University of PEARL https://pearl.plymouth.ac.uk

04 University of Plymouth Research Theses 01 Research Theses Main Collection

1983 HYDROGEOLOGICAL INVESTIGATIONS IN A GRANITE CATCHMENT, ,

Alexander, Jean http://hdl.handle.net/10026.1/1787

University of Plymouth

All content in PEARL is protected by copyright law. Author manuscripts are made available in accordance with publisher policies. Please cite only the published version using the details provided on the item record or document. In the absence of an open licence (e.g. Creative Commons), permissions for further reuse of content should be sought from the publisher or author. 4

•4

HYDROGEOLOGICAL INVESTIGATIONS in a GRANITE CATCHMENT, DARTOOOR, DEVON.

c Jean Alexander, B.Sc, M.Sc, Assoc. M.I.G.

A thesis submitted in partial fulfillment of the requirements for the de^ee of Doctor of Philosophy of the Council for National Academic Awards.

Dint: 1

Collaborating Est^b-lrshment South V/est Water -

September I983,

PLYMOUTH POLYTECHNIC

Department of Geographical Sciences

6 I •

r

\ PlYMOur/Y POLyTECllNIC

No ^'500143-1 Class. No. Memorandum The material presented In this thesis is the result of my own independent research work carried out in the Department of Geographical Sciences» Plymouth Polytechnic, and in the Narrator Brook Catchment, an experimental small drainage basin, on Dartmoor, Devon, during the period October 1978 - September 1980. The project was supervised by Dr. L. Ternan and Mr. P. Sims, Department of Geographical Sciences, Plymouth Polytechnic, with Mr. C. Tubb as the external advisor from the collaborating establishment. South West Water (S.W.W.). Any reference to previous work, published or unpublished is given full acknowledgement. To my parents "he dwelt from the beginning of Arda In the outer Ocean, and still he dwells there. Thence he governs the flowings of all waters, and the ebbing, the courses of all rivers and the replenishment of springs, the distilling of all dews and rain in every land beneath the sky."

From "The Silmarillon" J.R.R. Tolkein

"There is no life without water. It is a treasure indispensable to all human activity."

European Water Charter, Council of Europe, May 6th 1968. p Abstract: "HYDROGEOLOGICAL INVESTIGATIONS in a GRANITE If CATCHMENT, DAHTMOOR, DEVON". Jean Alexander I A network of 13 observation wells were utilized in the Narrator Brook i Catchment, Dartmoor, Devon, to monitor groundwater level fluctuations, in an attempt to assess the water resource potential of a weathered granite aquifer. Meteorological parameters, soil moisture data and reservoir ^A' levels together with waterlevel observations were used in a stepwise regres- sion model with a view to predict and account for the behaviour of ground• water levels in weathered granite areas.

Variations in groundwater and surface water chemistry were used as a basis for the hydrogeological grouping of the wells in the valley, whose individual charateristics were determined by site topography, lithology and slope position. Spring and river discharges were recorded and their recession characteristics determined. In a typical year approximately 7J^ of total discharge in the Narrator Brook channel is derived from groundwater, • ^ and 32% of the effective rainfall on the catchment finds its way into the J stream channel. The potential storage Rapacity of the weathered granite valley aquifer was estimated as 570x lOr m3 and an average value of (A 557 X id?. m3 per year was calculated for potential recharge.

Igneous bedrocks are not us\ially considered to make significant groxind- water contributions to riverflow, but Dartmoor in general, and the Narrator Brook Catchment in particular, have extensive deposits of weathered granite in situ overlain by periglacial deposits and river alluvium. These materials enhance the waterbearing properties of a granite based catchment, and combine to provide small isolated but significant valley aquifers.

i Acknowledgements: My thanks to geography, geology and ecology technicians for their assistance with fieldwork and laboratory work, particularly Mr, R, Cockerton and Mrs* J. Edwards for their help during the drilling and well completion procedures. To Mr. L. Cook, Mr. J. Harper and assistants in the Polytechnic's workshops, for their advice and assistance in the construction of field equipment used during this investigation. To Mr. J. Kltching of the Hydrographlc Department of the School of Maritime Studies, Polymouth Polytechnic, and the final year undergraduates (1979) for surveying in the well sites in the Narrator Brook Catchment. To Dr. C. Bowler, Geology Department, for the use of the Minute- man-drill, and to Mr. D. Clark for his services as a driller during inclement weather conditions in the catchment. To Mr. E. Durrance, Geology Department, University for the loan of the hammer seismograph. To Mr. C. Tubb and Mr. T, Jones for their advice and practical guidance during slug-testing procedures in the catchment, and to S.W.W. for the loan of the pressure transducer and recording equipment. My thanks also to Mr. G. Taylor (S.W.W.) manager at Burrator Reservoir, for the access to reservoir level data, co-operation during the drilling of the observation boreholes, and for his additional valuable local knowledge of the catchment area.

To the staff of the Computer Centre, Plymouth Polytechnic for their willing assistance, and to Dr. M. Kent for his guidance in the use of the Geog-Stats Package. Special thanks are extended to Mr. M. Coard for the use of, and subsequent adaptation of his stream discharge plotting computer program, and to Mr. A. Williams for many helpful discussions on the Narrator Brook Catchment. I am especially grateful for the skills and patience of Miss. S. Webber and Mrs. J. Jones in the preparation of some of the diagrams, and to Paula Barnes and Jane Moore who typed this thesis. My everlasting gratitude to Howard for his practical assistance and support at each stage of this investigation, and to his acceptance of the Narrator Brook Catchment as an exclusive topic of conversation. My thanks to colleagues in the Fluid Processes Unit of the Institute of Geological Sciences, for their continual encouragement and support. Lastly, I would like to thank two supervisors. Dr. L. Ternan and Mr. P. Sims, Department of Geographical Sciences, Plymouth Polytechnic, for their attention, discussions, recommendations and perserverance throughout the duration of this research project. /

Drilling in the Narrator Brook Catchment Chapter 1 INTRODUCTION TO GROUNDWATER STUDIES IN GRANITE AREAS

Introduction 1-5 1 Properties of Aquifers 5-8 2 A review of Aquifer Materials in the U.K. 8- 9 3 Unconsolidated Aquifers 9- 13 4 Characteristics of Alluvial Aquifers 13- 14 5 Development of Alluvial Aquifers 14- 17

2 Water Supply from Granite Areas 17 2:1 Sound Granite 17-20 2:2 Decomposed Granite 20-22 2:3 Decomposed Granite Areas in the U.K. 22- 23 2:4 Water in Granites of South West 23- 26 2:5 Water in Weathered and Solid Granites 26-28 2:6 Recent Granite Investigations 28-33

1:3 AtBg"aiia~ObTe'cCrve"s 33- 34 1:3:1 Broader terms of reference for hydrogeological Investigations in weathered granites 34- 35 1:3:2 Groundwater supply in the South West 35- 36 1:3:3 Surface water supply in the South West 36 1:3:4 Dam Construction 36- 37 1:3:5 Waste Disposal Interests 37- 38 1:3:6 Mining and Geothermal Interests 38- 39

1:4 Conclusions 39-42 Chapter 2 THE HYDROGEOLOGICAL BASIS

2J 1 Introduction 43-44 2: 1 1 Solid Geology 45-49 2 1 :2 Alteration of the Granite 49-50 2 1 3 Weathering Processes 50-52 2 1 :4 Nature of Decomposed Granite 52-54 2 1 :5 Classification of Weathered Granite 54-55 2 :1 :6 Other Superficial Deposits 55-56

2:2 Hydrogeological Properties of Valley Deposits on 56 Granites in 2:2:1 Burrator and 57-61 2:2:2 Hart Tor Valley 62 2:2:3 Cowsic Valley 62 2:2:4 Swlncombe 62 2:2:5 Huccaby 62 2:2:6 Taw Marsh 62 2:2:7 Fernworthy Dam 67 2:2:8 Sibleyback 67-68 2:2:9 Drift 69 2:2:10 Stithians 69 2:2:11 Summary of Site Conditions 69-70 2:2:12 Main Characteristics of Solid and Weathered Granites 70 (i) Fissures and joints 70-71 (11) Kaollnised sections 71 (ill) Veins, stringers and intrusions 71-72 (iv) The junction of weathered and solid sections 72

2:2:13 Water Movement in Weathered Granite 72-74

2:3 Factors Related to Hydrogeological Responses in 74-76 Weathered Granite Areas as Typified" by the NarratorBrook Catchment 2:3:1 Topography 76 2:3:2 Climate 76-82 2:3:3 Soils 82-86 2:3:4 Peat Deposits 86-88 2:3:5 Vegetation 88-89 2:3:6 Nature of the Valley Infill in the Narrator Brook 89 2:3:7 Borehole evidence 89-91 2:3:8 Particle Size Analysis 91-93 2:3:9 Mining Activities 93-95

2:4 Groundwater and Surface Water Relationships 95-96 2:4:1 Surface Water 96-98 2:4:2 Groundwater 98 2:4:3 Laboratory Derived Values of Porosity and 98-100 Hydraulic Conductivity of Granitic Material in the Narrator Brook Catchment 2:4:4 Gross Hydraulic Conductivity Determinations 100-101 2:4:5 Groundwater Transit Time 101-102 2:4:6 Watertable Fluctuations 102

2:5 Conclusions 104-105 Chapter 3

LABORATORY TECHNIQUES

3:1 Introduction 106 3:1:1 Water Level Data 106- 107 3:1:2 Well Design Considerations 107- 109

3:2 The Observation Network In the Narrator Catchment 109- 112 3:2:1 Installation of Observation Wells 112- 113 3:2:2 Well Dimensions 113- 114 3:2:3 Well Casing Materials 114- •116 3:2:4 Well Screens 116- •118 3:2:5 Slot Size Determination 118 3:2:6 Screen Length, Position and Effective Open Area 118- 121 3:2:7 Formation Stabiliser 121 3:2:8 Installation of Casing Components 121- 122 3:2:9 Well Development 122- 126

3:3 Wat'e"r~Xevie;l'8"and''Dlschargie''Mea8uremerit 127 3:3:1 Spring Ne twork 127- 129 3:3:2 Surface Water Measurements 130- 131 3:3:3 Supplementary Discharge Data 132

3:4 Ge'ochemical Sampling 132- 134 3:4:1 Field Techniques; The Collection of Water Samples 134 Groundwater Samples 134- 135 Surface Water Samples 136 Other Water Samples 136 3:4:2 Laboratory Techniques; Storage of Samples 136 3:4:3 Determination of the Properties of Water Samples 136 3:4:4 pH Measurements 136- 137 3:4:5 Specific Electrical Conductivity 137 3:4:6 Sodium and Potassium Determinations 138 3:4:7 Silica and Chloride Determinations 138- 139 3:5 Supplementary Measurements 139 3:5:1 Temperature 139 Water Temperature Measurements 140 3:5:2 Rainfall, Stemflow and Throughfall Measurements 141. Rainfall 141 Throughfall 141-142 Stemflow 142 3:5:3 Soil Moisture Measurements 142-144 3:5:4 Gravimetric Soil Moisture Measurements 144 3:5:5 Tensiometer Studies 144-145 3:5:6 Tensiometer Installation 145-148

3:6 Summary 148 Chapter 4 DISCHARGE CHARACTERISTICS IN THE NARRATOR BROOK CATCHMENT

4 1 Introduction 149 4 1 1 Sources of Streamflow 149 4 1 :2 Overland Flow 149-150 4 1 3 Interflow 150-151 4 1 :4 Groundwater Discharge 151 4 1 5 Stream Classification 151-152

4:2 Analysis of Data from Continuous Recording Stations 152-156 4:2:1 Hydrograph Analysis and Graphical Techniques 156-162 4:2:2 Deriviation of the Baseflow Component for the 162-163 Narrator Brook 4:2:3 Baseflow Recession in the Narrator Brook 163-169 Recession - Constants 169-173 4:2:5 Active Storage in the Narrator Brook Catchment 173-178

4:3 AflatygtB"gf"Dgtra "frgffl'Wgn-CgnTllltftfuB'-Sr^ 179-184 Gauging Sites

4:4 The Stability of Groundwater Discharge in "the 184 Narrator Brook Catchment 4:4:1 Flow Duration Analysis in the Narrator Brook Catchment 184-191 4:4:2 Flow Duration Curve Indices 191-193 4:4:3 Water Balance in the Narrator Brook Catchment 193-198

4:5 Conclusions 198-200 Chapter 5 GROUNDWATER FLUCTUATIONS IN THE NARRATOR BROOK CATCHMENT

5:1 Introduction 201 5:1:1 Controls on Watertable Fluctuations 201 5:1:2 Meteorological Factors 201-202 5:1:3 Topographic Factors 202-203 5:1:A Soil and Infiltration Factors 203-204 5:1:5 Hydrogeologlcal Factors 204-205 5:1:6 Miscellaneous Factors 205 Barometric Pressure 205-206 Loading Effects 206 5:1:7 Air Entrapment During Groundwater Recharge 206-207 5:1:8 Effect of the Unsaturated Zone on Watertable 207-208 Fluctuations

5:2 Contrlbutary Factors to the Groundwater Regime 208 In the Narrator Brook Catchment 5:2:1 Soil Moisture Conditions 208-214 5:2:2 Spring Discharge 214-218

5:3 Analysis of Groundwater Data - Graphical Techniques 218 5:3:1 Well Hydrographs 218-223 5:3:2 Groundwater Contour Maps 223-228

5:4 Analysis of Groundwater Data 228 Statistical Techniques 5:4:1 Standard Deviation of Water-level Fluctuations 228-231 5:4:2 Blvarlate Analysis 231-234 5:4:3 Analysis of Variance 234-238

5:5 Multivariate Analysis 238-240 5:5:1 Stepwise Regression 240-241 5:5:2 General Problems of the Use of a Stepwise 241-242 Regression Procedure 5:5:3 Variables Used in the Stepwise Regression Procedure 243 1. Rainfall 243-244 2* Potential Evapotranspiration 245 3,4 Soil Moisture Measurements 246 5, Sine Index 246-247 6,7 Antecedent Precipitation Indices 247-249 8. Soil Moisture Deficit 249 9. Reservoir Level 249-250 5:5:4 Data Transformation 250-251

5:6 Stepwise Regression Results for the 251 Narrator Brook Catchment 5:6:1 Correlation Matrix 251-260 5:6:2 Stepwise Regression Procedure Runs A and B 260-262 Well 1 262 Well 2 262- 263 Well 3 and Well 4 263 Well 6 263- 264 Well 7 264 Well 9 264 Well 10 264 Well 11 265 Well 12 265 Well 13 265 Well 14 and Well 15 265 5:6:3 General Patterns 266-267 5:6:4 Summary 268 5:6:5 Predictive Equations 269 5:6:6 Analysis of Residuals 269-276 5:6:7 Statistical Significant of 276-277 Multiple Regression Equations 5:6:8 Discussion of Stepwise Regression Procedure Results 277-278

5:7 Conclusions 278-280 Chapter 6 AQUIFER PROPERTIES IN THE NARRATOR BROOK CATCHMENT

6:1 Introduction 281 6:1:1 Definitions 281-282 6:1:2 Flow Conditions In an Aquifer Induced by Pumping 282

6:2 Groundwater Flow Equations 282-284 6:2:1 The Development of Methods for Determining Aquifer 284-288 Characteristics

6:3 Pump Tests 288 6:3:1 Pump Testing in the Narrator Catchment 289-290 6:3:2 Theis Method 290-297

6:4 Use of Slug Tests in Groundwater Investigations 297 6:4:1 The Development of Instantaneous Discharge or Recharge 297-300 Aquifer Tests 6:4:2 Field Equipment 300 6:4:3 Procedure to Perform a Slug Test 300-303 6:4:4 Observation Well Choice 303 6:4:5 Derivation of Type Curves 303-306 6:4:6 Aquifer Parameters Determined by Cooper, Bredehoeft 306-310 and Papadopulus Method 6:4:7 Bouwer and Rice Analysis 310-311 6:4:8 Determination of K (Bouwer and Rice) 311-314

6:5 Slug Test Parameters In the Narrator Valley 314-315 6:5:1 Results 315-316 6:5:2 Spatial Patterns 316-319 6:5:3 Slug Test Traces 319-320 6:5:4 Applicability of Results 320-321 6:5:5 Observation Well Effects 321-322 6:6 Comparison of Results from Differing Methods In the 322-324 Evaluation of Aquifer Constants In the Narrator Valley 6:6:1 Discussion 324-326 6:6:2 Comparisons with Slug Test Results In other Areas 326-328

6:7 Specific Yield Determinations In the Narrator Valley 328-332 6:7:1 Storage In the Narrator Brook Aquifer 332-333

6:8 Conclusions 333-335 Chapter 7 CHEMISTRY OF GROUNDWATERS IN THE NARRATOR BROOK CATCHMENT

7:1 Introduction 336 7:1:1 Ionic Content of Natural Waters 336-337 7:1:2 Origin and Source of Groundwater Constituents In 337 Granite Areas 7: 1 3 Silica 337-339 7 1 :4 Chloride 339-340 7: 1 5 Sodium 3AO-3A2 7 1 :6 Potassium 342-343 7 1 7 Atmospheric Contributions 343-344 7 1 :8 Chloride In Rainwater 344-345 7 1 9 Potassium and Sodium In Rainwater 343 7 :1 :10 pH of Rainwater 345-346 7 1 :11 Soli Zone Contributions 346-347 7 :1 :12 Geological Contributions 347-348 7 1 :13 Contributions to River Water 348 7 :1 :14 Summary of Source of Ions 348-349 7 :1 :15 Seasonal Variations in Groundwater Geochemistry 349

7:2 Presentation of Hydrochemlcal Data 3»*9-351

7:3 Chemistry of Groundwater and Surface Waters 351 in the Narrator Valley 7:3:1 Seasonal Trends in Water Chemistry: Temporal Patterns 351-362 7:3:2 Groundwater Temperatures 363-368 7:3:3 pH 368 7:3:4 Electrical Conductivity 368-370 7:3:5 Chloride 370 7:3:6 Silica 370-372 7:3:7 Sodium 372-373 7:3:8 Potassium 373 7:3:9 Summary of Temporal Patterns in Groundwaters from the 373-375 Narrator Brook Catchment 7:3:10 Hydrogeologlcal Groupings based on Water Chemistry 375-378 7:3:11 Comparison with Groundwaters of Other Granite Areas 378 7:4 Trends In Groundwater Chemistry - Spatial Patterns 379 7:4:1 The Use of Simple Statistical Techniques to Determine 379-382 Spatial Variations 7:4:2 The Degree of Intercorrelatlon of Solutes In 383 Groundwaters from the Narrator Valley 7:4:3 Temperature Variation with Depth 383-388 7:4:4 Summary of Spatial Patterns 388

7:5 Rate of Water Movement Towards" the"Wafertable-rn' the 389 Narrator Brook Catchment 7:5:1 The Use of Chloride as a Tracer 389-390 7:5:2 Variation In Groundwater Lag Times 390-395 7:5:3 Anomalous Groundwater Conditions 395-397

7:6 Geochemlcal Interpretation 397-399 7:6:1 Equilibrium Relations 399-402 7:6:2 Reaction Rates and Molecular Diffusion 402-403 7:6:3 Groundwater Quality 403-404

7:7 Conclusions 404-406 Chapter 8 THE POTENTIAL OF THE NARRATOR BROOK AQUIFER; SUMMARY AND CONCLUSIONS

8:1 Introduction 407-408 8:1:1 Characteristic Features of the Narrator Brook 408 Weathered Granite Aquifer 8:1:2 Lithologlcal Controls 408-411 8:1:3 Hydraulic Properties of Aquifer Materials 411-413 8:1:4 Water-level Fluctuations 413-415 8:1:5 Surface and Groundwater Discharges 415-417 8:1:6 Occurrence and Distribution of Springs 417-424 8:1:7 Geochemical Properties of Surface and Groundwaters 425-426 8:1:8 Temperature of Groundwaters 426-427

8:2 Hydrogeological Provinces in the Narrator Valley 428-434

8:3 Prediction of Groundwater Behaviour in the 434-435 Narrator Brook Catchment 8:3:1 Aquifer Response Times 435-436 8:3:2 The use of Slug Tests in the Assessment of 436-438 Groundwater Behaviour 8:3:3 The use of Regression Equations for Groundwater 438 Prediction Purposes

8:4 The Potential of the Narrator Brook"Aquifer 438-439 8:4:1 Groundwater Resources 439-441 8:4:2 Groundwater Quality 441-442 8:4:3 General Assessment of Weathered Granite 442-444 Resources

8:5 The General Form and Properties of'Weatliered 444 Granite Aquifers 8:5:1 The Extent of Weathered Granite Aquifers 444-446 8:5:2 Groundwater Movement 446-447 8:5:3 A General Model for Groundwater Conditions in 447-448 Weathered Granite Aquifers 8:6 Recommendations and Proposals for 449 Future Research 8:6:1 Aquifer Thickness 449 8:6:2 Extent of Joint and Fissure Systems 449 8:6:3 Spring Development 449-450 8:6:4 Delineation of Groundwater Flow Regimes 450 8:6:5 Mlcrostudles 450-451 8:6:6 Future Boreholes 452 Chapter 1 LIST OF FIGURES

Figure 1:1 South West England Location of the Narrator Brook Experimental Catchment Devon 1:2 Burrator Reservoir and the Narrator Brook 1:3 Generalised Distribution of Alluvial Deposits in England and Wales 1:4 South West England Geothermal Investigation Borehole Sites on Dartmoor Chapter 2 LIST OF FIGURES 2:1 Location of the Narrator Catchment

2:2 Distribution of Granite in the South West of England 2:3 Water Resource Investigations and Reservoirs on Dartmoor 2:4a Burrator Dam Geological Sketch 2:4b Sheeps Tor Embankment, Geological Sketch 2:5 Geological Cross-Section Hart Tor Valley 2:6 Geological Cross-Section Cowsic Valley Dartmoor 2:7 Geological Cross-Section of the Swincombe Valley 2:8 Geological Cross-Section Huccaby Area 2:9a Sketch-Map Location of Taw Marsh, N.E. Dartmoor 2:9b Diagrammatic Sketch of the Geophysical Survey, Taw Marsh 2:10 Geological Cross-Section, Fernworthy Dam, Dartmoor 2:11 Narrator Brook Experimental Catchment Illustrating the Groundwater Observation Study Area 2:12 The Narrator Brook Catchment Topography 2:13a Observation Borehole Elevations and Valley Profiles 2:13b Observation Borehole Elevations and Valley Profiles 2:13c Observation Borehole Elevations and Valley Profiles 2:14 Generalised Distribution of Vegetation and Soil Types in the Narrator Catchment 2:15 Diagrammatic Representation of the Main Soils of the Narrator Brook Valley 2:16a Particle Size Distribution Site 8 2:16b Particle Size Distribution Site 16 2:17 Distribution of Surface Water Around Burrator Reservoir 2:18 The Narrator Brook Catchment; Surface Water Distribution; Extent of Mining Areas 2:19 Graph to Illustrate the Relationship between Topography and Water Level in the Narrator Valley Chapter 3 LIST OF FIGURES

3:1 Observation Wells - Dimensions 3:2 Well Screen and Casing Arrangement 3:3 Surge Block-Hand-Operated Model 3:4 Groundwater Thermometer and Well Bailers 3:5 Water Level Recorder 3:6 Surface Water Distribution 3:7 Water Sampler 3:8 Tenslometer Installation and Plan Views of Tensiometer Sites 3:9 Total Instrumentation Narrator Brook Western Sector 1978-1980 Chapter 4^ LIST OF FIGURES

4:1 Regression Station Cutt vs Station 11 Discharges 4:2 Regression Head Weir (Burrator Reservoir Inflows) vs Narrator Brook Discharge (Station Cutt) 4:3 Hydrograph Separation Techniques and Components 4:4 Narrator Brook Station 11. Mean Dally Discharge 1978-1979 4:5 Groundwater Depletion Curves Station 11, Narrator Brook 1975- 1979 4:6a Master Recession Curve Narrator Brook Station Cutt 1975-1979 4:6b Master Recession Curve Station 11 4:7 Semi-Log Plot of Master Recession Curve Narrator Brook 1975-79 Station Cutt 4:8 Semi-Log Plot of Master Recession Curve Narrator Brook 1975-79 Station 11 4:9 Stream Gauging Study Section with Effluent and Influent Regions in the Narrator Brook Catchment 4:10 Contributions to Stream,Flow per unit drainage area, of discharge sites and stations, along a Study Section of the Narrator Brook Channel 4:11 Loss of Flow to aquifer between Site E and Station Cutt vs Discharge at Station Cutt 4:12a Flow Duration Curve Station Cutt 4:12b Flow Duration Curve Station 11 4:13 Log-Probability Plots Flow Duration Curves Station 11 4:14 Log-Probability Plots Flow Duration Curve Station Cutt 4:15 Log-Probability Plot 4-Year Flow Duration Curve Chapter 5 LIST OF FIGURES

5:1 Soil Moisture Variations with Depth Site 1 5:2 Soil Moisture Variations with Depth Site 9 5:3 Typical Soil Moisture Profiles in the Narrator Brook Catchment as Determined by the Gravimetric Method 5:4 Recession Constant a for Spring 12/13 5:5(a) Groundwater Level Fluctuations 1978-1980 Wells 3,4,6,11 5:5(b) Groundwater Level Fluctuations 1978-1980 Wells 1,2,9,15 5:5(c) Groundwater Level Fluctuations 1978-1980 Wells 7,10,12,13,14 5:6 Groundwater Contours Narrator Brook Catchment, July 1979 5:7 Groundwater Contours Narrator Brook Catchment, December 1979 5:8 Change in Groundwater Levels Narrator Brook Catchment 1979 5:9 Standard Deviation of Waterlevels in the Narrator Brook Catchment 5:10 Smoothed Well-Hydrographs and Rainfall Values in the Narrator Brook Catchment 1978-1980 5:11 Distribution of the Sine Index in the Narrator Brook Catchment 5:12 Well Groups Determined from the Stepwise Regression Procedure 5:13 Hydrogeological Provinces in the Narrator Valley 5:14(a))Dlstrlbution of Residuals Produced from the Regression 5:14(b))Equations for Individual Well Sites in the Narrator Brook 5:14(c))Catchment Chapter 6 LIST OF FIGURES

6:1 Theis Type Curve 6:2 Well 9 Recovery 6:3 Well 10 Recovery 6:4 Well 11 Recovery 6:5 Well 12 Recovery

6:6 Diagram to Illustrate the Slug Test Procedure at an Observation well 6:7 Diagram to Illustrate Idealised Slug Test Trace 6:8(a) Slug Test Traces Narrator Brook Wells 6,7 6:8(b) Slug Test Traces Narrator Brook Wells 1,7,14 6:9 Type Curves for Instantaneous Change in Wells of Finite Diameter 6:10 Slug Data Plots and Match Points for two Observation Wells, Narrator Brook 6:11 Curves Relating Coefficients A, B and C to L/rw 6:12 Partially Penetrating, Partially Perforated Well in an Unconfined Aquifer Chapter 7 LIST OF FIGURES

7:1a Mean Monthly Values: Temperature Narrator Brook 1979-1980 Wells 1,2,7,9,10,11,12,13,14,15, Sp.ll, Sp.12/13

7:1b Mean Monthly Values: Temperature Narrator Brook 1979-1980 Wells 3,4,6, Narrator Brook, Mean Air Temperature

7:2 Mean Monthly Values: pH Narrator Brook 1979-1980

7:3a Mean Monthly Values: Electrical Conductivity, Narrator Brook 1979-1980 Wells, 1,2,3,6,7,9,10,11,12,13,15, Sp. 12/13 7:3b Mean Monthly Values: Electrical Conductivity, Narrator Brook 1979-1980 Wells 4,14, Sp.ll, Narrator Brook, R.F. 7, R.F.9, R.F.ll 7:4 Mean Monthly Chloride Values: Narrator Brook, 1979-1980 7:5 Mean Monthly Silica Values: Narrator Brook, 1979-1980 7:6a Mean Monthly Sodium Values: Narrator Brook, 1979-1980 Wells i;2;3;4;7; 9;io,ii;i2;i3;i4;i5 7:6b Mean Monthly Sodium Values: Narrator Brook, 1979-1980 Wells 6, Sp.ll, Sp. 12/13 Narrator Brook^ R.F,7, R.F.9, R.F.ll 7:7 Mean Monthly Potassium Values: Narrator Brook, 1979-1980

7:8 Distribution of Standard Deviation of Temperatures **C for Wells and Two Springs in the Lower Narrator Valley 1978-1980.

7:9 Well Depth Versus Standard Deviation of Water Temperature, Narrator Brook 1979-1980

7:10 Distribution of Standard Deviation of pH, Narrator Valley 1978- 1980

7:11 Distribution of Standard Deviation of Electrical Conductivity, Narrator Valley 1979-1980

7:12 Distribution of Standard Deviation of Chloride, Narrator Valley 1979- 1980

7:13 Distribution of Standard Deviation of Silica, Narrator Valley 1979-1980 7:14 Distribution of Standard Deviation of Sodium, Narrator Valley 1979-1980

7:15 Distribution of Standard Deviation of Potassium, Narrator Valley 1979-1980

7:16a Mean Values of Temperature, pH, Electrical Conductivity, Chloride, Silica, Sodium and Potassium, Narrator Valley 1979-1980 Wells 1,2,3,4,6,7,9,10,11

7:16b Mean Values of Temperature, pH, Electrical Conductivity, Chloride, Silica, Sodium and Potassium, Narrator Valley 1979-1980 Wells 12,13,14,15, Sp.ll, Sp.12/13

7:17 Idealised Cross-Valley Section Illustrating Likely Disposition of Perched Groundwater Bodies Within the Weathered Granite Matrix In the Narrator Valley

7:18a Relationship between Depth and Temperature in the Well-Profile Wells 1,2 7:18b Relationship between Depth and Temperature in the Well-Profile Wells, 7,9 7:18c Relationship between Depth and Temperature in the Well-Profile Wells 10,12,14 7:19 Distribution of Lag Times in the Narrator Brook Catchment based on Chloride Values

7:20 Water Displacement Mechanisms In the Narrator Valley

7:21 Proposed Groundwater Pathways to Well 7 In the Narrator Brook Catchment

7:21 Stability Fields among some Silicate Minerals in the Gibbsite- Kaollnlte-Mlca-Feldspar System at 25**C and 1 Atmosphere Pressure, from Natural X^aters in the Narrator Brook Catchment Chapter 8 LIST OF FIGURES

8:1a Mechanisms for Spring Development 8:1b Mechanisms for Spring Development 8:1c Mechanisms for Spring Development

8:2a Origin of Springs in the Narrator Valley 8:2b Origin of Springs in the Narrator Valley

8:3 Hydrogeological Provinces In the Narrator Brook Catchment

8:4 Cross-Valley Profile of a Weathered Granite Aquifer Illustrating the Hydrogeological Complexity of Specific Sites Chapter 1 LIST OF TABLES

1:1 Aquifers, Flowtypes, Thicknesses and Yields in the U.K. 1:2 Summary of some of the Hydrogeologlcal Literature available on Major and Minor Aquifers in the U.K. 1:3 Geothermal Borehole Sites, Dartmoor 1979 1:4 Summary of some of the Hydrogeologlcal Literature available on Weathered and Solid Granite Areas Chapter 2 LIST OF TABLES

2:1 Average Chemical Composition of Dartmoor Granites (%) 2:2 Depths of Weathering In Granite

2:3 Depths of Incoherent Granitic Material In South West England 2:4 Meteorological Data for the Narrator Brook Catchment, Headwelr October 1978-February 1980 2:5 Hydraulic Conductivity of Various Peats after Boelter (1965) and Burke (1972) 2:6 Laboratory Determinations for Hydraulic Conductivity and Porosity in Altered Granites from the Narrator Brook Catchment 2:7 Dartmoor Granite Porosity Measurements, after Fookes et al. (1971) Chapter 3 LIST OF TABLES

3:1 Well Densities 3:2 Topographical Position of Wells 3:3 Observation Borehole Depths 3:4 Depths of Infilling of Observation Wells 3:5 'Levelled-ln* Altitudes of the Observation Boreholes 3:6 Summary of Data Collected in the Narrator Brook Catchment Chapter 4 LIST OF TABLES

4:1 Range of Discharge in the Narrator Brook 1976-1979 4:2 Comparison of Recession Times and Catchment Areas 4:3 Groundwater, Effective Infiltration and Storage Components, Station Cutt

4:4 Groundwater, Effective Infiltration and Storage Components, Station 11 4:3 Theoretical Maximum Groundwater Storage at Station Cutt and Station 11 4:6 Storage Factors in the Narrator Brook Catchment 4:7 Area of Contributory Surfaces to Discharge Sites and Stations, Narrator Brook 4:8 Streamflow per Unit Drainage Area, at Discharge Sites and Stations, along a Study Section of the Narrator Brook Channel, (m^ sec-^ km-^) 4:9 Flow Duration Analysis Narrator Brook 4:10 Variability Indices for the Narrator Brook, after Lane and Lei (1950) 4:11 Two-component Storage Factors 4:12 Components for the Calculation of Gross Underflow at Station Cutt 4:13 Water Balance in the Narrator Brook Catchment: Station Cutt 4:14 Water Balance in the Narrator Brook Catchment: Station 11 Chapter 5 LIST OF TABLES

5:1 Annual Groundwater Level Fluctuations In Four Aquifers for the Water Year 1971-72 5:2 Ratio of Water-level Rise: Rainfall Depth 5:3 Extremes in Water-level Data in the Narrator Brook Catchment 5:4 Correlation Coefficients for Preceding 7 Days Rainfall versus Water Level, and Soil Moisture Deficit with Water-level in the Narrator Brook Catchment 5:5 Variance and Standard Deviation of Water-levels for Wells in the Narrator Brook Catchment 5:6 Analysis of Variance (One-Way) 5:7 Analayis of Variance (2-Way) 5:8 Use of the Variables in the Stepwise Regression Procedure 5:9 Rainfall Periods 5:10 Mean Monthly Potential Evapotranspiration Values for North Hessary Tor and Head Weir, Narrator Brook Catchment 5:11 Transformations Chosen from the Variables in the Narrator Brook Catchment 5:12a,b)Matrix of Correlation Coefficients Produced by the Stepwise c,d,e, )Regresslon Procedure for Wells in the Narrator Brook Catchment f>8>h» )Summary of Significant Parameter Correlation Coefficients for i,j,k,m)the Narrator Brook Catchment from the Stepwise Regression l,m )Procedure 5:14 Summary of Correlation Coefficients of Variables Common to all Wells in the Narrator Brook Valley 5:15 Negative Correlations between Reservoir levels and Water-level Observations in the Narrator Brook Valley, (Run A) 5:16 Significant and Rejected Parameters in the Narrator Brook Valley (Run B) as selected by the Stepwise Regression Procedure 5:17 Summary of Hydrogeological Groupings of Wells in the Narrator Brook Catchment from Four Types of Analysis 5:18 Predictive Equations for the Narrator Brook Catchment as Selected by the Stepwise Regression Procedure 5:19 Significance of Stepwise Regression Coefficients and r^ (%) Values Chapter 6 LIST OF TABLES

6:1 The Utilisation of the non-equilibrium Theory in the Development of Procedures to Determine Aquifer Formation Constants 6:2 Transmissivity Values using the Theis Method of Analysis 6:3 Assumed Saturated Thicknesses of the Aquifer in the Narrator Brook Valley 6:4 Results of Slug Test Analysis, Narrator Brook Catchment, June 1979 6:5 Changes in Hydraulic Conductivity and Transmissivity with Saturated Thickness of the Aquifer 10%, 40%, and 60% Larger than the Estimated Value of Saturated Thickness at each Site 6:6 Hydrogeological Properties of Selected Sites in the Narrator Brook 6:7 Re-equilibrlum Times for Slug-Tested Wells in the Catchment 6:8 Well Response Times 6:9 Comparison of Permeability and Representative Aquifer Materials, U.S. Dept. of Interior (1977) 6:10 Transmissivity and Hydraulic Conductivity values from Water Resource Investigations in the U.K. in Unconflned Strata 6:11 Slug Test Rsults from two other Granite Areas in the U.K. 6:12 Storage Values for Selected Observation Boreholes in the Narrator Brook Chapter 7 LIST OF TABLES

7:1 Reactions for Incongruent Dissolution of some Alumlnosllicate Minerals, after Freeze and Cherry (1979)

7:2 Summary Statistics for Bulk Precipitation Chemistry 1977-1978, Narrator Brook Catchment

7:3 Annual Mean Values of Various Constituents (mg 1~^) in Rainfall over the UK for 1959-1964, from Stevenson (1968)

7:4 Maximum and Minimum Ranges of Temperature, pH, Electrical Conductivity, Chloride, Silica, Sodium and Potassium in Groundwaters, Surface Waters, Precipitation and Streamflow in the Narrator Brook Catchment

7:5 Mean Values of Temperature, pH, Electrical Conductivity, Chloride, Silica, Sodium and Potassium in the Narrator Valley

7:6 Standard Deviation of Water Temperatures and Soil Temperatures in the Narrator Valley with Well Depth and Site Elevation

7:7 Standard Deviations of the most Constant Groundwater Constituents in the Narrator Brook Catchment

7:8 Comparison of Groundwater Chemistry from Granite Areas in the UK

7:9 Correlation Matrix showing the Interrelationships between Site Geochemistry in the Narrator Valley Waters

7:10 Lag Times In the Narrator Valley

7:11 Drinking Water Quality Guide Chapter 8 LIST OF TABLES 8:1 Aquifer Hydraulic Properties

8:2 Hydrogeologlcal Parameters illustrating the Variations between Upstream and Downstream Sections of the Narrator Brook Valley 8:3 Standard Deviations of Well and Spring Concentrations measured in the Narrator Brook Valley to illustrate the Chemical Characteristics of Groundwaters from different localities within the catchment. 8:4 Summary of Hydrogeological Groupings of wells in the Narrator Brook Valley 8:5 Well Depths, Mean Depth to the Watertable and Well Elevations 8:6 Mean Aquifer Properties, Waterlevel Ranges and Saturated thicknesses in the Narrator Brook Valley 8:7 Groundwater Parameters indicative of Resource Potential in the Narrator Brook Valley APPENDICES Contents

Appendix 1 Drilling Logs, Narrator Catchment Dartmoor Devon

Drilling Logs, Taw Marsh N.E. Dartmoor

Appendix 2 Mean Daily Rainfall Narrator Brook 1976-79

Appendix 3 Streamflow at Discharge Sites and Stations Narrator Brook

Flow Duration Curve Tabulations

Mean Dally Discharge Narrator Brook (Station Cutt) 1976-1979

Discharge Data Spring 12/13

Appendix 4 Analysis of Variance Tables to test the significance of the Regressions chosen by the Step- Wise Regression Procedure Chapter 1 Introduction to Groundwater Studies in Granite Areas

1:1 Introduction This investigation is concerned with an evaluation of the hydrogeological characteristics of a small granite-based catchment area situated in South West England. Figure 1:1 illustrates the location of the Narrator Brook experimental catchment area on the southwest fringes of Dartmoor, Devon* As a result of the annual influx of tourists there are large seasonal and peak demands on the regional water supply. Over much of and Devon the means of meeting such extra demands have been by the construction of impounding reservoirs. As described in Chapter 2, many granite-based catchments contain considerable depths of unconsolidated materials, including granite in various states of decomposition, weathering products and sollfluctlon deposits. The potential for water resource development in these materials has not been fully investigated, since surface water supplies have received most of the attention in south west England, The physiography of south west England, (Cornwall, Devon and the Scilly Isles) and more specifically Devon, is such that there Is an uneven distribution of rainfall over the area. The increased cloudiness, humidity and rainfall lead to an average summer surplus of the water budget of 75 mm over Dartmoor, but an average deficit of 75 mm in the lowlands, (Brunsden and Gerrard 1977). As pointed out by these authors, when the need for irrigation and the heavy use of water by the coastal resorts is considered in the light of these figures, then the need to use the upland water catchments of Dartmoor for reservoirs becomes very apparent. Based on the above cited surplus and deficits and the use of upland catchments for water supply, then similarly the weathered granite catchments on Dartmoor may also be potentially useful in groundwater supply terms, and require assessment. The basic climatic and hydrological data support the case for judicious planning of the water resources of Dartmoor to include both reservoir sites and groundwater development* FIGURE 1:1 SOUTH WEST ENGLAND LOCATION OF THE NARRATOR BROOK EXPERIMENTAL CATCHMENT DEVON

i

Narrator Dartmoor Brook Catchment

Scale 1cm-20/cm Proposed reservoirs on Dartmoor have a history fraught with strong public opposition, despite the fact that new water supplies are essential to the population inhabiting the surrounding areas. In such environments where available land at hydrologically suitable sites are designated 'prime agricultural land*, 'areas of outstanding natural beauty' and 'National Park', the opportunities to provide the much needed increment to the water supply network, are not forthcoming. Against this unpromising background isolated investigations into potential groundwater resources have commenced which are dealt with more specifically in Sections 1:2:3 and 1:3:2. This research project has been prompted by a specific need in this region to find alternative public water supplies, without recourse to the construction of reservoirs. The Narrator Brook Valley Catchment, (within the National Park) is a major source of water supply to Burrator Reservoir, which supplies Plymouth and its environs. As such it seemed a suitable point from which to begin a groundwater investigation in a granite area. In addition this drainage basin has been instrumented since 1974, and a variety of records are available for Headwelr (the gauging station monitoring flow into Burrator Reservoir) and Burrator Reservoir, (Figure 1:2), providing valuable background information on streamflow. Since this Investigation is approached from a water resources viewpoint, brief mention is given to the present distribution of sources of public water supply in south west England. Further emphasis is given to the importance of the major sedimentary aquifers in this country for public water supplies, and the numerous investigations which have subsequently led to an increased knowledge of their hydrogeological characteristics. By comparison little work has been carried out on aquifers of weathered crystalline rocks, and hence the resultant paucity of hydrogeological knowledge of such aquifers* Recent developments on small localised sources in the form of "alluvial aquifers" and their hydrogeological properties are highlighted in this introductory Chapter, since they may well have common properties with weathered granite aquifers. Relevant literature on the occurrence of groundwater In crystalline rocks (the majority of which are granites unless otherwise ASharpitor

A Leather Tor

ADown Tor HEAD WE R ST CUTT

Combshead Tor ^

BURRATOR flESERVOIR A Sheeps Tor Bur rater Dam Sheops Tor Dam

Scale 1: 25.000

FIGURE 1:2 BURRATOR RESERVOIR AND THE NARRATOR BROOK specified) is reviewed, along with some Isolated cases of water resource development in granite regions. Other significant Investigations in granite terrains are discussed as they particularly highlight one of the alms of this project, that of assessing the potential use of weathered zones in granite as localised aquifers. By way of an introduction some standard hydrogeological definitions are outlined, and at a later stage re-assessed in the context of weathered granite aquifers. 1:1:1 Properties of Aquifers Most rocks and sediments contain numerous open spaces or interstices in which water may be stored and through which it can move. Water that exists In the interstices of rocks and soils is called subsurface water, Meinzer (1959), and that part of subsurface water in Interstices completely saturated with water is groundwater. According to Bouwer (1978) groundwater is that portion of the water beneath the surface of the earth that can be collected with wells, tunnels or drainage galleries, or that flows naturally to the earth's surface via seeps or springs. An aquifer is defined as a saturated bed, formation, or group of formations which yield water in sufficient quantity to be of consequence as a source of supply, (Walton 1970). The most Important requirement is that the stratum must have Interconnected openings through which water can move. Porosity of a material is the percentage of the total volume of the material that is occupied by pores or interstices. These pores may be filled with water if the material is saturated, or with air and water if it is unsaturated. Porosity is an index of how much water can be stored in saturated materials. Other aquifer parameters such as specific yield, storatlvity, transmissivity and permeability are defined in Chapter 6. An aquifer performs two Important functions; a storage function and a conduit function. The pores in a water-bearing formation serve both as storage spaces and as a network of conduits. By contrast an aquifuge is a formation which has no Interconnected openings and hence cannot absorb or transmit water, (Linsley e£ al. 1949). According to Todd (1959) solid granite belongs in the category of an aquifuge. An aqultard, as defined by Bouwer (1978) may contain large volumes of water but does not permit its movements at rates sufficiently large for economical development. The nature of each aquifer depends on the material of which it is composed, its origin, the relationship of the constituent particles and associated pores, and its relative position in the earth's surface, in conjunction with its exposure to a recharge source. By far the most common aquifer materials are the sedimentary rocks, as illustrated in Table 1.1 which provide the majority of groundwater supplies in this country. Sandstones, unconsolidated sands and gravels which occur in alluvial valleys, coastal plain deposits and glacial materials are common aquifer materials. Cavernous limestones with solution channels, caves, underground streams and other karst developments can also be high-yielding aquifers (Legrand and Strlngfield 1973). Other sedimentary rocks such as shales, and massive limestones do not generally make good aquifers, although small water yields may be possible where these rocks are highly fractured and weathered. The same is true for granites, gneisses and other crystalline or metamorphlc rocks, (Bouwer 1978). Most aquifers are of large areal extent and may be visualised as underground storage reservoirs (Todd 1959). Water enters a reservoir from natural or artificial recharge; it flows out under the action of gravity, or Is extracted by wells. Ordinarily, the annual volume of water removed or replaced represents only a small fraction of the total storage capacity. Aquifers may be classed as unconfined or confined depending upon the presence or absence of a water table. An unconflned aquifer is one in which groundwater possesses a free surface open to the atmosphere. Such an aquifer is often referred to as being under water-table conditions. Water infiltrating Into the ground surface infiltrates downwards through air-filled interstices of the material above the saturated zone and joins the groundwater body. The water table represents the surface along which the hydrostatic pressure Is equal to the atmospheric pressure. In the unconflned case there is no restricting material above the saturated zone so the groundwater level, the water table, is free to rise and fall. A confined aquifer is one in which groundwater is confined under pressure, by overlying relatively impermeable strata. When such

6 - aquifers are penetrated by wells, water will rise above the bottom of the overlying confining bed to an elevation at which it is in balance with the atmospheric pressure* This pressure condition in a confined aquifer is known as the potentiometric or piezometric surface* These potentiometric surfaces are produced on maps as continuous lines obtained by connecting equilibrium water levels in wells penetrating the confined aquifer. Groundwater contours, so produced are actually a map of the hydraulic head in the aquifer. Force potential causing groundwater flow is directly proportional to the elevation of the water levels in confined and unconflned aquifers* Potentiometric maps are best suited to describe water levels representative of a single flow system within an aquifer. Beds of clay, silt or other materials with relatively lower permeabilities than the surrounding aquifer material, may be present In some areas above the regional water table. Downward percolating water may be Intercepted and a saturated zone of limited areal extent is formed* This results in a perched aquifer with a localised 'perched' water table. Depending on climatic conditions, or overlying land use, a perched water table may be a permanent phenomenon, or may persist on a seasonal basis only* 1:1:2 A review of Aquifer Materials in the UK The importance of an aquifer depends on a large number of geological characteristics. The texture of a rock determines effective porosity and permeability, which In turn control the upper limit of groundwater storage and the ease with which the water can flow through the formation. Unconsolidated sediments of uniform size generally provide the highest yields per unit volume of rock, but in practice, other factors such as the thickness of the aquifer, structure, areal extent and annual recharge rate become dominant. This is true in the UK where the sedimentary rocks provide the main water-bearing strata for public supplies* Table 1:1 illustrates the major physical characteristics of the chief aquifers in this country* The most important aquifer is that formed by the large deposits of chalk stretching south of a line between Weymouth and the Wash, plus parts of east Yorkshire and Lincolnshire* The chalk supplies 40% of the groundwater abstracted for public supplies, with an outcrop of 12,950 km2 and another 18,000 km^ below the surface (Kirby 1979)* Table 1:1 Aquifers Flow Types, Thicknesses and Yields in the UK (after Jones, 1977)

Rank Aquifers Lithology Dominant Flow Max. Thickness Typical Yield Pattern (metres) (1/sec)

MAJOR Cretaceous: Chalk soft 1st. I/F 500 150 Triasslc : Bunter sst.mst. I/F 600 100 Sandstone Carboniferous: Coal Measures sst. F 2000 -

INTERMEDIATE Recent : Superficial Deposits sands, gravels I 10 — Cretaceous: Lower Greensand soft sst. I 250 50 Jurassic : Inferior Oolite 1st. sst. F 200 50 Triassic : Keuper Sandstone sst. I/F 300 100 Permian : Magnesium Limestone 1st. F 250 50 Carboniferous: Limestone hard 1st. F 1000 150

MINOR Pliocene/Pleistocene : Crag sands I 50 10 Tertiary : Lower London Clay sand, silts I 50 25 Cretaceous: Upper Greensand sst. I 50 25 Cretaceous: Wealden Sands sands I 50 25 Jurassic : Great Oolite 1st. F 50 50 Jurassic : Lias sst. I/F 50 25 Carboniferous : Culm Measures ) Devonian : Old Red Sandstone ) these strata alsc) provide small 'field s from borehc ties and springs Lower Palazoic ) 1 1 KEY: I « interstitial flow F = fissure/fracture flow 1st. = limestones sst. = sandstones mst. = mudstones Thicknesses vary from 90-500 m, and although It is a very fine-grained soft limestone of low permeability, as a bulk formation In the field It may have a permeability as much as a million times greater than locally measured permeability (Ineson 1962). This Is due to extensive fissure systems developed in the chalk mass* Chalk exhibits characteristic large-scale seasonal fluctuations in water levels (Headworth 1972). The Bunter and Keuper sandstones are the other major aquifer- system in the UK providing 25% of the abstracted groundwater in this country. Thicknesses vary from 12-350 m in the Nottinghamshire area, Alexander (1977), with a maximum thickness for the Bunter aquifer of 600 m, (Jones 1977). Total outcrop area is 4,532 km^ with another 3,237 km^ under more recent deposits, (Smith 1982). These coarse sandstones obtain their greatest development in the Midlands but extend from Cumberland through to the South West. They provide large yields, typically in the range of 100 1 minute"^, (Table 1:1), especially in the Bunter Pebble Beds, and are characterised by a wide range in grain size and in the shape of Individual grains, (Sherrell 1970). Their high permeability gives rise to only small fluctuations in rest water levels (Land 1966). The importance of these two aquifers in this country can be attributed to their large outcrop area rather than to their Intrinsic qualities of porosity and permeability (Smith 1972). Table 1:2 provides a summary of some of the literature available on the hyrogeological properties of the major and minor aquifers in the UK. The amount of literature available on groundwater investigations carried out on the Chalk and Bunter aquifers emphasises the importance of both these systems to water supplies In this country. More restricted aquifers, such as the Jurassic limestones, and sandstones In the Coal Measures may enjoy some local significance. 1:1:3 Unconsolidated Aquifers It is only in recent years, with increasing localised water shortages coupled with Increasing production and distribution costs, that water resources in unconsolidated materials and superficial deposits have been under investigation with a view to their development as potable water supplies. Such investigations, in addition to this current assessment of the potential of weathered granite aquifers, may help to redress the balance, and shift the emphasis away from the regional-scale aquifers to the more localised but still potentially useful aquifers.

Table 1:2 Sumary of Some of the Rjdrogeological Uterature Available on ^fejor and MLnor Aquifers in the UK

Type of Aquifer and Authors Investigation Chalk Bunter Others

General Ineson (1962) Land (1966) Ineson (1967) - h>drogeology fbster and >fLlton Bowet.al. (1970) Goal r-feasures and physical (1974) (1976) millams et al. Downiiig and properties (1971) MmianB (1969) Sage and Lloyd Jurassic l£ts. (1978) ThpplT (1971) - days CMrney (1972) tfegnesiun Lime• stones Trafford and Rycroft (1973) - clays ALdrick (1978) - (fagnesiun Lime• stones

Water level Halton (1954) fluctuations Headjorth (1972) (1978)

Recharge and Ineson and Sherrell (1970) infiltration Dawning (1965) Kitchlng (1974) capacity Ednunds et al. (1976)

Oonnorton and Vieed (1978)

Punp tests Ineson (1952) (1959)

Flow type Beeves et al. (1975) Crock and ttouell (1970)

Saline Foster et al. intrusion (1976)

10 - As a result of recent (1973) administrative changes In the water Industry, there has been a tendency for attention to be concentrated on the development of very large water resource projects. These serve regional schemes which cover extensive parts of a regional water authority area. Such major schemes are useful In that their implementation encourages the construction of extensive supply networks for the distribution of water from major sources, and results In an improved quality and reliability of supplies throughout the country.

There persists a few disadvantages in these major schemes which warrant attention here and exhibit the need for future development of small local water resources in valley gravels, weathered horizons and other similar superficial materials. A high element of capital cost is entailed in meeting peak demands in any region. Seasonal, diurnal and emergency peaks must have provision made for them in the capacity of all capital works built for a large water supply scheme. As a result of these peaks, works are used to full capacity only for very short periods, and the unit costs of water are correspondingly high. To deal effectively and economically with longer peak demands such as those caused by tourism and droughts, Jeffcoate (1977) suggested it may be possible to Incorporate a small local source, which, operated alone, would not be regarded as economic.

Where a local source can be developed and maintained without associated high costs of attendance out of proportion to the quantity of water obtained, they may prove invaluable in meeting peak demands and reducing a high distribution network cost. Such sources may be available in superficial deposits (detrital material transported and deposited by a river, commonly sands), gravels with some silts and clays; weathered material (transported by gravity such as talus, scree and sollfluction deposits); and horizons of weathered rock in situ. Sand and gravel deposits containing usable groundwater are found in many parts of England and Wales, as illustrated in figure 1:3 after Jeffcoate (1977). Most are of glacial origin, mainly outwash deposits from ice sheets but also flood-plain deposits and terraces formed by river action. It is not always easy to distinguish between the various types, but since most of the usable water is contained within

11 - FIGURE 1:3 GENERALISED DISTRIBUTION OF ALLUVIAL

DEPOSITS IN ENGLAND

AND WALES

V, '.ANewcastle

Liver poo

*^Bristol

Scale 3cm - 100km sand and gravel deposits directly associated with existing rivers, such horizons are termed "alluvial aquifers" by Monkhouse (1978). 1:1:4 Characteristics of 'Alluvial Aquifers* . The principal functions of an aquifer are to store and transmit groundwater, and its ability to do so can be quantitatively assessed by considering the properties of the material, particularly those relating to storage potential and transmissivity. The alluvial deposits include more recent alluvium composed of silts, clays and associated peats. Underneath and flanking these materials are frequently to be found layers of sands and gravels deposited in earlier periods when flows and stream velocities were greater and the river beds were at a higher level. These sands and gravels are variable in quality and characteristics, the variations being consistent in both depth and area. According to Price and Foster (1974) rapid lateral variations and complex boundaries are characteristic of most valley fill deposits. They are usually deposited in pronounced lenticular beds varying from fine sands to gravels, cobbles and boulders, only rarely are silts present. Such beds yield water in large enough quantities for public supplies. Grannemann and Sharp (1979) noted that large amounts of groundwater are stored in alluvial deposits whose mean grain size and hydraulic conductivity both increase exponentially with depth. The alluvial deposits, whatever their geological origin, have extremely variable hydrogeological characteristics. The clays and organic deposits are normally of extremely low hydraulic conductivity, whereas the coarser gravel deposits are amongst the most permeable of all geological formations. The hydraulic conductivity of alluvial deposits depends largely upon the homogeneity of the grain size. Todd (1959) considered that a clean, well sorted gravel might have a hydraulic conductivity of as much as 10^ m day"^. Measured values in England (Naylor 1974, Wilson 1975), in Ireland (Price and Foster 1974) , and in Wales (Welsh National Water Development Authority 1975) , are much less than this, being in the range of 5 to 1500 m day"^ with a mean of 350 m day"^. These relatively low values are probably due to the presence of silt, clay and fine sands within the gravel. With a maximum thickness of the order of 10 m,

13 this gives transmlsslvities of 50-15000 m^ day"^, with a mean of about 3500 m2 day-^. These differences occur in close proximity with the result that water resource evaluation of the formation, based on the assumed homogeneity of the aquifer is rarely reliable. An aquifer is normally thought of as a reservoir collecting water which infiltrates directly from rain falling on the exposed outcrop, though it may also receive contributions by seepage from other aquifers or surface flow. Alluvial deposits, while complying with the above description, may also receive replenishment by surface water from a river system or by interception of spring flows and seepage from adjacent formations. Because of numerous sources of replenishment the potential of alluvial aquifers cannot be assessed in the normal way by calculations of replenishment from the area of exposed formation and the estimated infiltration as is done for the deep aquifers. In general, alluvial aquifers provide relatively small supplies due to their limited thickness and extent. Wilson (1975) noted that as a proportion of total abstraction from principal aquifer groups only 2.2% was obtained from superficial deposits, including gravels. Glacial and flurio-glacial material are present in valleys in some parts of the UK. Such mixtures of valley infill material may be difficult to differentiate. Hydrogeologically they may be Inseparable from the alluvlals, but because of their fine silt content, the permeability of glacial deposits is much lower and they are rarely of value as an aquifer themselves. The presence of clay and silts reduces both the storage volume and the ability of a formation to transmit water. 1:1:5 Development of Alluvial Aquifers Where a major water supply scheme is proposed, it requires careful long-term planning and a suitable financial climate. When expenditure on public works is restricted, any major scheme may be deferred even though water shortages result. In addition to an unsuitable financial climate, public opinion may defer the implementation of a major water-supply scheme, which also results in local water shortages. This seems to be a trend, particularly in south west England, where potential reservoir sites in or near the National Park and other areas of natural beauty are vetoed by certain

14 sectors of the public. Such actions result in proposals for new reservoirs, such as Swincombe on Dartmoor in 1970, being rejected. In addition prolonged public debates, like the Roadford Enquiry, result in large bills for the rate-payers, with little other alternatives to supplement ever-increasing water demands in the meantime. Given such conditions, small source developments would be invaluable to ensure that growing water demands will be satisfied in any intervening periods. Short-term operational costs on such small sources are high but if severe water use restrictions are enforced as an inconvenient alternative, such expenditure may be Justified. Again aquifers of superficial materials may offer a suitable low-cost source which could be rapidly developed to meet immediate local needs. In the UK the development of small sources in the alluvial gravels for public supply and for Industrial and agricultural purposes have been common in the past, (Jeffcoate 1977). During the first half of this century the utilisation of all small sources for public supplies suffered a decline. There were exceptions to this, notably in the Nene and Ouse Valleys (Anglian Water Authority) where, according to Wilson (1975) in the absence of any obvious source of a suitable size to serve the scattered rural community, 94% of the water for the whole area (served by the Old Nene and Ouse Water Board) was taken from alluvial aquifers. With the advent of the Diddlngton Reservoir supply, which was designed primarily to meet the demands of more distant urban communities, this dependance on the alluvial aquifers declined. In a few other areas, notably in the Thames Valley, Naylor (X974), Trent Valley, Broadhead and MacKay (1972), and in Ireland, Price and Foster (1974), detailed investigations have been carried out. Elsewhere relatively little is known about the location of these aquifers or the groundwater resources contained in them. Exploitation of superficial deposits for useful sources of potable water have to some extent been precluded by the availability of large supplies of groundwater from the sedimentary aquifers in this country. This In turn has led to the lack of suitable hydrogeological information on shallow non-consolidated aquifers in general (Jones 1970). Jones (1970) noted that the term aquifer Is used as a restricted concept in that it has been thought of solely as a groundwater

15 - reservoir from which water could be extracted. The alluvial aquifer is in fact normally thought of as a vehicle for conveyance and natural filtration of surface water, rather than as an underground reservoir from which water may be abstracted by drawing on storage at times of reduced replenishment. Water obtained from shallow alluvial resources is more liable to contamination than water obtained from conventional deep-seated supplies. Wilson (1975) suggested that with increasing demand for water and the length of time and high costs needed to clean up rivers, possible sites from gravel sources where small supplies are required, should be Investigated. The object of Broadhead and MacKay's (1972) investigations into the use of the Trent alluvial gravels, was to utilise their natural filtration properties In order to obtain a supply of water superior in quality to river water. These authors concluded that limitations of suitable gravel deposits made their usefulness as regional schemes small, but emphasised that they could offer viable local sources of water.

Exploitation of the Thames alluvial aquifer as a source of groundwater has to some extent been precluded by the availability of large supplies of groundwater from the chalk and surface supplies from the Thames itself (Naylor 1974). A preliminary study was made of the alluvial gravels in the Middle Thames Valley to assess their usefulness as a major source of groundwater. Tentative average storage values were estimated as 168 x 10^ m^, and if it is assumed that only 10% can be developed, this storage would provide 168000 m^ day~^ for 100 days (Naylor 1974). This emphasises the potential value of the gravels as a storage area. It should be noted here that this is one of the few alluvial aquifers in the UK which could be developed along the lines of a conventional aquifer with storage facilities, due mainly to the thicknesses of sand and gravels, often up to 13 metres. Price and Foster (1974) found that rapid lateral variations and complex boundaries were characteristic of most of the alluvial aquifers Investigated in Ulster. Such variations resulted In corresponding changes in aquifer permeability, saturated thickness and transmlssivity. Groundwater bodies were often locally confined by overlying alluvial silts causing variations in the storage coefficient from less than 10"^ up to the specific yield values, which varied from 0.05 to 0.25 in the Ulster gravel formations. With regard to the

16 - overall depths of alluvial aquifers in the UK, Monkhouse (1978) noted that thicknesses of 4 m are common, up to 6 m not unusual, but thicknesses of more than 10 m are rare. In Devon and Cornwall alluvial deposits occur in most valleys, but with the exception of the Dart Valley (section 1:3:2) these have not been investigated or developed. Alternative supplies from the Bunter Sandstone in appear to be a more attractive proposition, but in an area with increasing water supply problems one is prompted to look closely at smaller potential water sources. Development of localised water supplies from alluvial materials and superficial deposits has become an important issue, but as yet little work has been carried out on the potential water-bearing horizons of weathered granite in this country. This present investigation hopes to remedy this situation. 1:2 Water Supply from Granite Areas This investigation is primarily concerned with the evaluation of weathered granite aquifers as potential water resources; however of necessity the water availability in the solid-granite basement area must also be assessed. The unconsolidated weathered granite zones and their resources may well influence the availability of deeper circulating groundwater. Conversely weathered granite zones may be recharged from deep-fracture systems, and spring waters may utilise vein pathways and fissures in the granite bedrock.

For the purposes of this discussion the granite-mass is divided into two zones: (i) Sound Granite (ii) Decomposed Granite The sound granite comprises the coherent mass of the intrusion crossed by faults, fissures and veining, while the decomposed area consists of chemically or physically weathered unconsolidated materials and hydrothermally altered zones. Further details concerning weathering of granites are presented in Chapter 2. By virtue of their physical state the water-bearing properties of the two groups are very different.

1:2:1 Sound Granite Most of the work carried out on the water-bearing properties of crystalline igneous rocks, and their weathered zones, has been

17 - conducted abroad. Ellis (1906) whilst drilling for water in Connecticut USA, made several observations concerning the water- yielding ability of crystalline rocks. He noticed that porosities were often less than 1% in granites and other crystalline rocks, and that the joint systems played an Important role in the circulation of water in granites. Ellis (1906) observed that the major groundwater source to wells in granites were supplied by horizontal fractures. This feature was particularly evident in the deeper wells (60-92 m in depth) in which the principal fracture source was usually close to the well base. Water yields from such wells were noted to have been constant and exhibited no appreciable annual fluctuations.

Melnzer (1923) suggested that water yield in crystalline rock was a function of joint occurrence, whilst Tolman (1937) indicated that water movement in superficial fracture systems does not bear the same relationship to the watertable as water movement in granular material. Le Grand (1949) mentioned the importance of joints and fissure systems on groundwater occurrence in granites and describes "sheeting", the development of planes in a massive igneous intrusion. These sheets were found to be approximately parallel to the surface, and water was envisaged as moving in a step-like motion along the sheeting planes down to the valley reservoirs. Le Grand (1949) pointed out that artesian conditions were extremely rare in igneous and metamorphic rocks, but may occur between different "sheets" in the granite. In contrast Ellis (1906) mentioned that wells in which water rises above the rock surface are common. In crystalline rocks the frequency of joints and fissures, especially their connections, dimensions and dips, determine the water-bearing capacity (Meier and Petersen 1951). From their work on the Swedish Archaen bedrocks these authors suggest granites are better aquifers than basic rocks such as diorltes and gabbros, and porphyries are better than dolerites or basalts. Yields from the Swedish granite areas vary from 0.2 to 0.4 m^ sec~^ (Meier and Petersen 1951). The idea has prevailed in some of the earlier literature (Ellis 1906, Gear 1951) that In cases where no water, or only very small quantities were struck at depths of less than 100 m, that there was little chance of improving quantities obtained with penetration to greater depths. Meier and Petersen (1951) however found that in the

- 18 Swedish granites considerable amounts of water are derived from depths of 100-200 ra. Unweathered and unfractured crystalline rocks generally have less than 1% porosity and, according to Davis and Turk (1964) have permeabilities so small as to be almost negligible. These authors attribute the water-bearing character of most crystalline rocks in California USA, to weathering and structural controls. Lewis and Burgy (1963) engaged upon hydrologlcal studies in the crystalline rocks in Sierra Nevada, USA, noted from well logs that there is a tendency for the number of open joints and fractures to decrease with depth, and hence reduce permeability. Larson (1968) makes similar observations on the volume and behaviour of joint and fault systems, outlining their importance in predetermining the availability of groundwater in the Pre-Cambrlan gneisses and granites of Sweden. This author concludes that an extensive study of the faults and joints of the rocks may favour a probable estimation of potential yield of groundwater in a granitic bedrock. The yield of any water-bearing horizon is predetermined by the physical properties of the materials such as porosity and permeability. Louis (1968) records that the hydraulic conductivity of the unweathered rock is small; ranging from 0.5 to 2.0 X 10""® m sec"^ , but that of weathered granite is greater. For the hydraulic conductivity of a granite mass Louis (1968) gives 0.7 m sec"^ in the direction of the fissures on the basis of one fissure 1 mm wide for every metre. Lugeon tests in Brazil, Franciss (1970) gave mass permeabilities of 10"^ to 10"^ m sec"^. Brown et al. (1975) noted that few measurements were available on the permeabilities of granites. Since 1975 the hydrogeology of granitic areas in countries like Canada, USA, Britain and Sweden has been under close scrutiny, with a view to assessing the feasibility of the geological disposal of high- level radioactive waste. As part of the UK investigations, the crystalline rock environment in the Strath Halladale complex In Caithness was subject to a variety of studies. Bulk hydraulic conductivity for the fractured crystalline rock was at an average of about 10-^ m sec"^, Mather et aj. (1982), while laboratory measurements of bedrock gives' porosities of only 0.01 (Glendinlng 1980). Brace (1980) gives a synopsis of the laboratory determined.

- 19 - in situ and inferred values of permeability of crystalline rocks, from the feasibility studies carried out In USA, UR and Sweden. For crystalline rocks in situ permeabilities ranged from 10"^^ to 10"^ m sec"^, but no systematic decrease with depth was evident. This brief review of the literature dealing with water resource investigations and recent studies on granite masses in association with other crystalline rocks, supports the view that groundwater occurrence in such materials is by no means insignificant. Enslin (1943) noted that most of the extractable groundwater In igneous formations is stored in the decomposed rocks. This author asserted that Igneous rocks contain no water, but that found is confined to the joint and fracture systems and in decomposed materials. 1:2:2 Decomposed Granite Enslin (1943) suggested that as decomposed crystalline rock has a high porosity and permeability, and if weathering extends to any depth, then such materials form ideal reservoirs for underground storage. This author was one of the first geologists to emphasise the Importance of chemical weathering in the formation of isolated basins of weathered granite. On the granite peneplain in the Kenhardt district. South Africa, basins of decomposed granites were found in excess of 60 m depth. In Igneous rocks a definite watertable only exists if the decomposed rock is saturated to a certain depth. Water which is stored in the granite mass in the joint and fracture systems are disconnected and too locally confined to form any uniform watertable (Bnslin 1943). This factor was commented upon by Dawklns (1901) from work carried out in Guernsey in the Channel Islands. Water-bearing beds formed by the decomposition of igneous rocks, rest in situ on solid granite. They occupy irregular hollows 9-12 m deep. As a result of these conditions water was stored in the porous decomposed rocks "in water-tight cells like those of a honeycomb though more irregular", (Dawklns 1901). Consequently, there is no free circulation and a well can only draw on its own reservoir, independent of adjacent wells. It seems unlikely that a large supply of water could be attained from any one centre of pumping in this situation, although there might be many useful wells sufficient for small local demands.

20 According to Le Grand (1949) rock type, structure, topography and the degree of weathering are significant factors which govern the occurrence of groundwater in granites. Weathered granites on the Piedmont, Georgia, USA, furnished appreciable storage space for groundwater resources. This residuum was found by Le Grand (1949) to be thickest in the lowlands where less surface runoff and more direct influent seepage occurred, feeding the groundwater reservoir.

Gear (1951), on a survey of water-wells in granite regions of Uganda, attempted to correlate borehole yield with rock-condition. Out of a total of 117 boreholes the majority of high supplies, 0.10 X 10-5 Q^^3 X 10-3 ^3 sec-^, were found in boreholes drilled in partially weathered rock. Boreholes below depths of 91 m in solid unweathered granite yielded few supplies, except where major fissures were intersected. The water yielding potential of gneisses, schists, amphibolites and other crystalline rocks in Georgia, USA, were investigated by Stewart (1962). These crystalline rocks were mantled by various thicknesses of weathered rock, referred to as "saprollte". The porosity of the saprolite was found to be 54% at depths of 9-12 m, but decreased with depth as the weathered mantle grades into unweathered rock, with porosity less than 5%. Weathering processes (see Chapter 2) cause differential expansion of various mineral grains due to partial hydration. The expansion and differential movement will produce intergranular pore space and circulating water may increase local porosities by the dissolution of unstable minerals. If a crystalline rock is originally coarse-grained with a moderate abundance of stable minerals such as quartz, a relatively high hydraulic conductivity may result from weathering. Porosities in such materials may be in the range of 30-50% (Davi s and Turk 1964).

The thickness of the zone of weathering depends on the geologic history of an area. According to Davis and Turk (1964) maximum depths of 92 m may be encountered in areas of little erosion, but depths of 3-30 m are more typical in California, USA. Later work of Turk (1973) in the Llano Area, Texas, emphasised the importance of the relationship between well-yield and depth of weathered material.

21 Largest yields were found from wells in thick layers of weathered granite found primarily in valley bottom locations. Similar conclusions were drawn by Uhl and Sharma (1978) while investigating water supplies from weathered crystalline aquifers in the Satpura Hills, India. Thicknesses of saprolite vary from 1 to 35 m in depth, and transmisslvity values ranged from 2 to 382 m^ day~^ in the weathered zones. Significant differences in the hydraulic conductivity with depth were observed, and groundwater circulation in jointed and fractured bedrock was restricted, by the closing up of fractures, to depths of 45-60 m in most areas (Uhl ^ al. 1979). The crystalline-rock country is gently undulating in the Satpura area, and groundwater basics were found to be conterminous with surface drainage sub-basins which are a few square kilometers in area (Uhl et al. 1979). The groundwater flow systems were of a local type, where each local system has its recharge area at a topographic high and discharge area at a topographic low. These authors came to the conclusion that intermediate and regional groundwater flow systems do not exist in these Indian crystalline areas because of the negligible hydraulic conductivity with depth.

1:2:3 Decomposed Granite Areas in the UK Generally water-bearing beds formed from decomposed granite resting in situ on solid material below, occur over a limited area in the British Isles. Accumulations of weathered rock In situ are to be found mainly in Devon, Cornwall and the Channel Islands. It is generally accepted that major glacial ice sheets did not extend over the land south of the 'Bristol-London line'. As a consequence the granites of the Lake District, Scotland, Isle of Man and Ireland are not expected to have such depths of weathered materials as are found in south west England.

There are however some exceptions to the above generalisation. Fitzpatrlck (1963) working on aspects of soil formation in Scotland discovered Igneous, metamorphic and consolidated sedimentary rocks weathered to depths greater than 12 m. These were overlain by glacial till in some areas, and by drawing comparisons with similar deposits in Asia and North America, Fltzpatrlck (1963) attributed their formation to weathering in a pre-glaclal period.

22 Moore and Gribble (1980), while investigating the suitability of aggregates from weathered granites at Peterhead, Scotland, discovered that altered granite penetrated by boreholes extended to depths of 60 m. These authors note that large areas of Scotland have been affected to some extent by weathering processes, leaving a weathered mantle, and the coastal granites in the Buchan area of north-east Scotland seem particularly prone. The great thicknesses of alteration in parts of north-east Scotland and Its apparent uniformity lead Moore and Gribble (1980) to suggest that such features may be taken as evidence of hydrothermal alteration. During a recent drilling programme in the Scottish granites of Caithness, Glendining (1980) reported that boreholes penetrated areas of greater than 40 m of weathered granite. As Is typical of granite terrain in the UK depths of weathered rock varied from site to site. From 24 holes drilled the extent of weathering varied from 2.4 m to greater than 40 m below the surface, (S. Glendining pers. comm. 1980). 1:2:4 Water in Granites of South West England From the foregoing review it is evident that the most prolific parts of a granite area, from the point of view of water supply, are those zones of decomposed and fractured materials. The potential water-bearing characteristics of the granite and its weathered zones in south west England have been noted in the past, Sandeman (1901) and discovered during the course of trial excavations for the construction of dams in the area in connection with the development of impounding reservoirs. Sandeman (1901), during the excavation of a trench for the Burrator Dam, observed that the quantity of water draining into the trench from crevices in the rock was small. When the Sheepstor trench was excavated water was hit less than one metre below the surface in this weathered zone, and increased in quantity with depth until the pumps were lifting 2,045 m^ day~^ to heights greater than 30 m. It was observed that as greater depths were attained the decomposed granite became harder until it gradually merged into hard (unaltered) rock. As the sinking of the Sheepstor trench proceeded, water which had formerly bubbled up like a strong spring in the bottom of the trench

- 23 was observed to be coming more from the sides of the trench. Two veins which conducted water to the surface of the granite via the "hard quartz rock" were piped and the water carried up above the water level of the reservoir. This description of the occurrence of water in the weathered granite at this location. Is of particular Importance here since the valley upstream of Burrator Reservoir Is the subject of this present investigation. Other evidence relating to groundwater occurrence in granite and weathered zones in the vicinity of Dartmoor are obtained in sinking of wells and shafts for water supplies. At Sampford Splney N.G.R. SX 534724, west of the Narrator Brook Catchment, two-well bores were put down on a farm revealing the following: (1) Level of ground surface 231.64 O.D. Depth of bore 18.89 m Water table depth 4.26 m The well log revealed 0.30 m of subsoil and 18.59 m of soft granite. A pump-test yielded 4.54 m^ sec~^ after one hour's pumping. (2) Level of ground surface 259 m Depth of bore 18.28 m Water table depth 0.91 m The well log revealed 1.52 m of peat and drift material and 16.76 m of 'extremely soft decomposed granite'. A two hour pumping test yielded 1.51 m^ sec~^ with a 7,62 m water level depression below the well top. Rest level below the well top was 0.91 m with a recovery period of half an hour. As part of a research project into the distribution of underground temperatures in the granites of south west England, with a view to assessing possible future geothermal resources, (Whelldon et^^. 1977), some exploratory boreholes were drilled into the Dartmoor granite. Six sites (Figure 1:4) were investigated and depths to solid granite and the watertable are tabulated in Table 1:3. Solid granite and dry holes were a pre-requlslte for suitable temperature measurements, Francis (pers. comm. 1980) and several obstacles presented themselves in the form of varying and Inconsistent thicknesses of weathered granite in association with varying amounts of water experienced in the Dartmoor sites. These borehole logs nevertheless provide very useful background information for this

- 24 A

" .2 N .1 X

•3

r

PLYMOUTH

Scale 3 cm =10 /cm

FIGURE 1:4 SOUTH WEST ENGLAND Geothermal Investigation Borehole Sites on Dartmoor •1 Blackingstone •2 Rock Artillery Range •3Soussons Down •4Foggintor Quarry •BLaughter Tor •6 Quickbeam Hill Investigation. More than one attempt at drilling was carried out at each site, giving the range of depths to solid granite encountered, and the variation In depths to the watertable. At Laughter Tor the first watertable was hit at 3.96 m In weathered granite. From 6-8.5 m below this level the borehole through the granite was dry until the second watertable was hit at 8.53 m. These Investigations highlight the variation In the watertable or potentlometrlc surfaces In weathered granite zones.

Table 1:3 Geothermal Borehole Sites Dartmoor 1979

Site National Grid Depth of Depth to Depth to Reference borehole Solid watertable Granite

Blackingstone SX 7850 8593 100 m 3-4 m 9 m

Winter Tor 6117 9156 100 m 3.96->16 m 5.79, 14.02

Soussons Wood 6733 7971 100 m 6 m 11.43

Foggln Tor 5663 7334 100 m 2.5-7 m 9.14

Laughter Tor 6562 7549 100 m 3.96-4.5 m 3.96, Second WT at 8.53 m

Quick Beam 651 647 100 m 7-12.5 m Dry Hill

1:2:5 Water in Weathered and Solid Granites Specific investigations on potential igneous aquifers for water supply purposes In south west England and the UK, appears only to be instigated where other water supplies are inadequate. This is the case in the Isles of Scllly, where hydrogeologlcally only the granites and recent alluvium are of any significance, (Burgess et al. 1976). The granite has secondary permeability and the system is unconflned although head variations were recorded during drilling. These were attributed to differential pressures in different fissures and the local absence of hydraulic continuity. The granites of the Scllly Isles are well jointed with a system of inclined conjugate joints and sub-horizontal pressure relief

26 - Joints. The horizontal joints appear to be dominant In controlling groundwater flow (Burgess and Clowes 1975). The exposed granite in the Scilly Isles is everywhere weathered in situ to produce angular blocks in a matrix of quartz and finer clay material. This is known locally as 'Ram' and strongly resembles Pleistocene head deposits. Thicknesses of these two materials in association with recent alluvium varies from 1-15 m. Where the granite is covered by Head, a low permeability material, the aquifer may be partially confined (Burgess and Clowes 1975). With respect to aquifer characteristics, standard pumping test analysis was discounted in the Scilly Isles on the basis of the secondary nature of the permeability of the granite and the absence of observation wells. The specific capacity data for wells in the granite indicated a wide range, 0.35 to 5.4 m^ day~^, which makes the regional allocation of permeability difficult (Burgess and Clowes 1975). As the granite aquifer possesses non-homogeneous permeability the qualitative analysis of groundwater in granites and weathered granites involving normal Darcy principles is not feasible. Binnte and Partners (1971) estimated transmisslvities of 74 to 112 m^ day"^ and a storage coefficient of 0.01 in one of the three groundwater basins on St. Marys, Isles of Scilly. As is expected in a granite aquifer its ability to transmit water is limited and hence individual boreholes were found to give low outputs. Binnie and Partners (1971) report specific capacities of wells varying from 0.32 to 120.1 m^ d"^» and estimated the potential yield of all existing sources of water supply on the island (rain tanks, reservoirs and groundwater resources) was sufficient to meet a peak week demand of 2.10 m^ day-^. On Tresco, Scilly Isles, two depresslonal areas in the granite exist around Grimsby and the Great Pool. Head and gravels have accumulated in these overlying weathered granites, thus creating two areas with small unconfined aquifers (Sherrell 1964). Yields in wells are in the order of 0.052 m^ day"*^ with a water table 3 m below the surface in the Grimsby area, to a yield of 0.073 m^ day~^ with a watertable 0.9 m below the surface in the Great Pool basin. Both these weathered granite aquifers are tapped for water supply by large well-chambers 1.50 m in diameter, constructed at depths of 4.6-6 m.

27 with lateral feeding channels at depths of 3.65 m to permit extraction of water over a maximum area (Sherrell pers* comm. 1980). The results of these two studies In the Scllly Isles give an Indication of the difficulty of assessing groundwater potential in both weathered and unweathered granite areas. They also serve to Illustrate the usefulness of Isolated weathered basins in augmenting public water supplies, and emphasise the fact that water supply from granites may be derived from both decomposed and weathered materials. In terms of water supply purposes it Is not a valid approach to assess weathered and non-weathered granite water supplies independently as the two systems are strongly Interrelated. This feature is shown in Table 1:A which summarises some of the hydrogeologlcal literature available dealing with weathered and solid granite areas. Most of this literature refers to the combined hydrogeologlcal conditions in both weathered and unaltered granite masses. 1:2:6 Recent Granite Investigations More specialised Investigations into the hydrogeological characteristics of a granite mass have been conducted recently. Llndblom, Lundstrom and Stllle (1979) point out that the assessment of the permeability of crystalline rocks has become an Important issue in nuclear fuel safety analysis in countries like Canada, U.S.A., Britain and Sweden, where repositories in hard rock are considered. The rock mass permeability together with the regional hydrology and the induced temperature field are factors that will determine the rate of groundwater flow. The only possible way for deposited radioactive material to reach the biosphere is through future transport in mobile groundwater. Although such investigations appear in the first Instance to be unrelated to groundwater resources in weathered granites, these unconsolidated zones and their water resources may well Influence the availability of deeper circulating groundwater. Conversely weathered granite zones may be recharged from deep-fracture systems and spring water and seeps may utilise vein pathways in the granite. The two veins transmitting significant amounts of water from an unknown depth which had to be covered and piped up above the water level of the

28 Table 1:4 Summary of Some of the Hydrogeologlcal Literature Available on Weathered and Solid Granite Areas

Type of Investigation Granite Areas and Authors

Decomposed Solid Decomposed Location Solid Location

General General Sandeman (1901) UK Ellis (1906) USA description description Dawklns (1901) Channel Melnzer (1923) USA Basins of Water yielding Islands Gear (1951) Uganda decomposition properties of Enslln (1943) South and thickness crystalline Africa of weathering rock types Le Grand (1949) USA Fitzpatrlck (1963) Scotland Davis & Turk (1964) USA Moore & Gribble Scotland (1980) Glendlning (1980) Scotland

Well yields Well yields Gear (1951) Uganda Tolman (1937) USA \ and depths jointing, Turk (1973) USA Le Grand (1949) USA faulting, Glendinlng (1980) Scotland Meier & Petersen Sweden and water (1951) movement in Larson (1968) Sweden crystalline Burgess et al. UK rock (1982) Physical Hydraulic Sherrell (1964) Scilly Lewis & Burgy USA properties properties Isles (1963) of decomposed of massive Blnnie & Partners Scilly Franciss (1970) Brazil crystalline crystalline (1971) Isles Brace (1980) USA materials rock Uhl & Sharma (1978) India Mather et al. UK Stewart (1982) USA (1982) reservoir during the excavation of the Sheepstor embankment, Sandeman (1901), Illustrate this point. Investigations attempting to elucidate hydraulic properties of a rock mass with a view to assessing their use as radioactive waste disposal sites are not new. Work on radioactive waste disposal in excavated vaults in crystalline rock was carried out in the early 1960's in South Carolina. Proctor and Marine (1965) emphasised that the most significant driving force for the migration of radionuclides from the storage site was derived from the natural water movement, coupled with the effects due to dispersion and ion exchange. Characteristics of the waste, heat generation and radlolysls were found to have only small effects on the migration of radionuclides In the gneisses, schists and granites of South Carolina. The detailed hydrogeological investigation of Proctor and Marine (1965) revealed that the very low permeability of the rock in which the storage vault Is located, the virtually impermeable clay layer separating the rock and overlying sediments was capable of confining the radionuclides used with the plant boundaries for a much greater time than the 600 year period required to render wastes Innocuous.

Large-scale permeability tests carried out in the granite of the Stripa mine, Sweden, Linblom et al. (1979) were conducted 360 m below the surface. The rock mass permeability was estimated as being between 1.6 to 5.1 x 10"^^ m sec"^, but the authors emphasised that new test methods must be developed for crystalline rocks with permeabilities below 10"^ m sec"^. Black (1979) attempted an analysis of the hydrogeological properties of the Carnmenellis granite in Cornwall, by using single borehole tests and multiple borehole tests. He concluded that multiple borehole test techniques have considerable potential but require adaptation to low permeability, low porosity crystalline rock environments. The permeability of the Carmenellis granite was found to be in the order of 0.1 x 10'^ m sec"^. The Institute of Geological Sciences (I.G.S.) have been concerned in recent years with research into the relevant geological criteria for safe long-term radioactive waste disposal in this country. A drilling programme in the granites of Caithness has been carried out (Glend;:ning 1980).

- 30 Preliminary investigations Into the Scottish granite at Altna Breac with a view to assessing such rock types used as nuclear waste repositories have produced valuable hydrogeological data. Three deep boreholes (300 m) and 24 shallow boreholes (40 m) were drilled penetrating weathered horizons in the granite often greater than 40 m in thickness. These granites at 180 m depth exhibited solution cavities with remnants of calclte crystals in them, while at depths of 280 m they showed staining along fractures, indicating paths of circulating water. Laboratory measurements on unweathered core material give porosities of 1%, possibly related to micro-cracks of the constituent grains (Alexander et al. 1981). Borehole hydrograph analysis suggested a gross interconnected porosity of 12-15% for the system as a whole. Much of this porosity is probably related to the drift and upper weathered and highly fractured zones in the top 40- 50 m, (Glendining, pers. comm. 1980). Throughput estimates, from recharge to discharge points are in the range of 5-100 years, being very fast in the top 40 m of the water profiles as Indicated by britlum values.

A drilling programme, as can be seen from the previous discussion, gives more detailed Information on the geological factors relating to water in, and the mechanical properties of, the subsurface crystalline rock. If crystalline rocks are to be used safely for high level radioactive waste disposal purposes, then data needs to be available from a variety of similar, if less suitable sites, so the spectrum of properties of crystalline rocks relevant to waste containment can be fully established. Similar reasoning necessitates more detailed study of fluid migration in weathered crystalline horizons, and the determination of the hydraulic connection between these two circulation systems is crucial. Such investigations will furnish important hydrogeological data which will be relevant to studies of the potential of groundwater resources from both weathered and solid crystalline bodies. Further evidence for the volumes of groundwater available from granite areas in south west England, is provided by hydrogeochemical studies undertaken to investigate the nature of groundwater circulation in the Carmenellis granite (Burgess et_ al^. 1982, in preparation). The existence of thermal saline springs in working tin

31 mines of Cornwall has been evident for a long time, (Henwood 1843), and suggests groundwater circulation at depth. At South Crofty, Pendarves, Wheal Jane and Mount Wellington, the four deep active tin mines sampled by Burgess et_al. (1982), the total amount of groundwater pumped from workings amounts to 76,000 m^ day"^. In kaolinlzed zones of Pendarves mine, and Mount Wellington, at depths of 250 m, percolation of groundwater into the workings is common in these 'wet* mines. Discrete inflows are also provided by fissure and fracture systems* Wheal Jane and South Crofty are much drier at their working levels of 300 m and 690 m respectively, with water more commonly encountered as discrete Inflows relating to joints, cross- cutting faults or lod€s. On the Carmenellls granite groundwater is commonly encountered in boreholes which Intersect fracture planes in granites. Such boreholes are less than 50 m in depth and Intersect watertables several metres below ground level. Piezometric heads vary in such boreholes, and in the case of boreholes at Rosemanowis Quarry, those within 20 m of each other vary, giving a maximum head difference of some 22 m, (Pearson pers. comm. April 1980). Fissure-fed springs on the Carmenellis granite, with discharges from 1-3 litres sec~^ are common occurrences, (Burgess et al. 1982). Recent (post 1953) meteoric water penetrates down to depths of at least 700 m in the Carmenellis granite, mixing as it does with much older saline waters in the fracture-systems. Such saline waters owe their hydrogeochemistry to reaction with the granite, (Burgess et al. 1982). The circulation system operating at present has been induced or greatly accelerated by mining activity in the area. The geochemlcal evidence implies circulation to a depth of 1.1 km and therefore the existence of hydraullcally effective fractures In the Carmenellis at these depths are inferred. At present there is a limited understanding of the behaviour at depth of arrays of rock fractures, of the paths and rates of fluid migration, of the permeability of crystalline rocks. The effects of joint-systems in relatively homogeneous granite at depth, and their present contribution to groundwater migration is little known. Any interconnections between the shallow water circulation system in weathered granites and that of the deeper circulation system of the

32 - granite mass, are of great Importance and, as yet, are not documented. Any knowledge concerning the water movement and its relative rate through zones of weathered granite, may aid in determining the effectiveness of a crystalline mass as a whole, to store, transmit and discharge water. Such hydrogeological information will be useful not only for the radioactive waste repository investigations and civil engineering concerns, but also as a basis for assessing the groundwater resource of a crystalline (granite) region. As a summary of the previously discussed factors, it is apparent from the literature survey that there is a lack of documented investigations on the water resources of granites and associated weathered zones. This deficit appears to be due mainly to the fact that large supplies of groundwater can be obtained from the sedimentary aquifers of the Permo-Trlas and Cretaceous formations, along with subsidiary minor sedimentary horizons as indicated In Table 1:1. Consequently, It is on these aquifers that the major hydrogeological investigations for groundwater supplies have been concentrated over the last decade* More recent developments of water supplies from alluvial aquifers have become an important Issue, but as yet little work has been conducted on granitic regions of this country for potential water supply resources. This present investigation hopes to remedy this situation. 1:3 Aims and Objectives The primary aim of this investigation is to determine the hydrogeological characteristics of decomposed materials in granite valleys using a small drainage basin on south west Dartmoor. The Narrator Brook Is typical of many valleys on the Dartmoor granite and other granites of south west England. One of the intentions of this present investigation in the Narrator Valley is to accrue data on the transmissivity and hydraulic conductivity of the weathered granite aquifer. Such information may act as a guideline in future hydrogeological assessments of weathered crystalline aquifers. Aquifer characteristics derived from the Narrator Brook may be of relevance to other areas of crystalline rock in the British Isles, in addition to those in other temperate environments.

33 Some of the more specific aims of this Investigation are summarised as follows: (1) a study is made of groundwater level fluctuations in order to assess groundwater behaviour in a weathered granite aquifer. Such results may be useful in an attempt to predict further groundwater conditions, and may also provide some basic generalisations on groundwater conditions in weathered sequences, which may be extrapolated to other regions; (ii) to assess the time taken to transmit recharge to the watertable through the unsaturated zone; (ill) to determine hydraulic conductivity, porosity, transmissivity and storage values, representative of a weathered crystalline aquifer; (iv) to determine the nature and thickness of the saturated horizons, and depth to solid bedrock; (v) to delineate recharge and discharge areas; (vi) to determine the direction of groundwater movement and the extent of hydraulic continuity between surface and subsurface systems; (vii) to assess the degree of success with which watertable observations in a portion of the aquifer, (under more detailed investigation) may be used to detect changes on storage in the Narrator Brook aquifer as a whole. The potential of an aquifer in weathered granite as a localised water pocket to augment public supplies in times of severe drought, is also an element under consideration. As this is the first attempt to examine groundwater resources in the Narrator Brook Catchment, it is felt that this project will fulfill its main objectives if some initial parameters of the groundwater regime can be Identified. 1:3:1 Broader Terms of Reference for Hydrogeological Investigations in Weathered Granites It can be seen that hydrogeological knowledge, concerning the occurrence and distribution of groundwater in igneous rocks, is desirable in an attempt to assess the usefulness of a potential aquifer in a weathered granite area* Such Information also has important Implications for the design, and construction of civil

- 34 engineering structures, such as reservoirs, which again have a special significance in south west England. The importance of hydrogeologlcal Investigations in small weathered granite catchments in south west England, can best be seen in relationship to water supply, waste disposal, mining and geothermal fields of interest. The hydrogeological properties of an area form the basis on which designs for dam construction, waste disposal and mining projects rest. 1:3:2 Groundwater Supply In the South west Much of south west England is underlain by Devonian and Carboniferous slates and shales of low porosity and permeability. Their generally small yields to wells and boreholes depend largely on the presence of fissures which according to Edmonds et al. (1975) are best developed where these rocks have been 'baked' by the intrusive granites, or are affected by faulting. Such fissure systems are rarely well interconnected so supplies commonly diminish or fail in dry summers. Edmcnds et al• (1975) refer briefly to water occurring in the fissures of the granite, and notes the low permeabilities of the fresh rock. Mention is made that weathering of granite may extend to a depth of 15 m or more, and where it does so good supplies of soft acid water can be obtained from these horizons. These authors allude to even more favourable conditions for water supplies from granite where the weathered rock is overlain by drift. Apart from this brief mention of water supplies from granitic areas in the South west there is a paucity of literature on this issue a point discussed more fully in sections 1:1:3, 1:1:5 and 1:2:5. Small quantities of hard water are obtained from the Devonian limestones near , and Plymouth. 30 Ml day~^ are obtained from the narrow alluvial aquifer in the Dart Valley near , but as yet this constitutes the only productive aluvial site in South west England. The breccias and sandstones of Permian and Triasslc age constitute the most important aquifer in the South west at the present time. The water-bearing Otter sandstones are being developed for municipal supplies in the Otter Valley east of Exeter. Large areas of east Devon and west Somerset are underlain by the thick impermeable Mercia Mudstone Group (Keuper Marl) and Lower Llassic clays, from which no commercial supplies are obtainable.

35 iouth west Water in Cornwall, north and south , and west Somerset, draw most of their supplies from surface water. These are mainly from reservoirs and river intakes. 1:3:3 Surface Water Supplies in the South West The presence of a number of storage reservoirs reflects the demand made on surface water for public water supply in south west England. There are 27 such reservoirs in Devon and Cornwall, covering a total area of greater than 8 km^. The largest of these is Wimbleball (1.51 km^) capacity 19,500 Ml on the Devon/Somerset border; Stlthians (I.10 km^) capacity 5202 Ml north of in Cornwall is the second largest; and Burrator Reservoir, Devon (0.60 km^) capacity 4210 Ml subject of this present investigation, being the third largest. All large centres of population in south west England, except Exeter, depend almost wholly on river intakes and reservoir supplies. 1:3:2 Dam Construction The main problems which arise due to the creation of a reservoir are related to (1) the construction of the dam; (11) the assessment of the water tightness of the reservoir basin; and (ill) the recognition of leaky zones and the extent of any potential leakage in the basin. In south west England there are special questions which arise in relation to the principle of constructing reservoirs and factors Involved in their location. As there are few aquifers of significance, it seems inevitable that the summer demand can, according to Knill (1972)4, only be supplied by storage in reservoirs. The physiography of south west England Is such that water may be best supplied from a series of small reservoirs rather than a single major source. Since the areas of high rainfall rates are attributed to the granite massifs, these are deemed suitable locations for storage reservoirs. As all natural materials have a finite permeability, the question as to whether seepage or leakage will take place around a dam is determined by the groundwater conditions and specifically the hydraulic gradient in the region. In consequence, the pre-existing groundwater conditions are of considerable importance in the assessment of reservoir watertightness. In south west England the reservoir basins are likely to be founded on granite, and very little

- 36 published material is available concerning specific hydrogeological details of either fresh or weathered materials. The assessment of seepage from reservoirs can be best derived in the first instance from a knowledge of the groundwater flow pattern and distribution of groundwater pressures. On the flanks of a reservoir, four general groundwater conditions may be recognised, as suggested by Knill (1972)band outlined briefly below to emphasise the importance of hydrogeological investigations into the groundwater regime of granite catchments: 1. Watertable and piezometric conditions in the groundwater system are In excess of the top water level (of the proposed reservoir). 2. Watertable, but not the piezometric conditions at depth are in excess of the top water level. 3. Both the watertable and the piezometric conditions are less than the top water level, but higher than the reservoir floor. 4. The watertable is depressed below the base of the reservoir. It can be seen that the detailed knowledge of groundwater conditions in a potential reservoir basin is of considerable importance not only because of the reservoir, but also because of implications on the design of groundwater control measures and cut• offs at the dam site. 1:3:5 Waste Disposal Interests A regional example of the necessity of determining groundwater conditions for dam construction on granite areas is afforded by the Kernlck Dam. This is located North of St. Austell in Cornwall. The dam is designed to retain mtcaeous residues from the China Clay mining operations, only allowing clear water to flow into the local river systems (Anon. 1974). Although the Kernick Dam is required to drain off water as quickly as possible, whereas a water-retaining dam is required to impound it, knowledge of the site hydrogeological conditions are just as Important. Existing records of boreholes sunk for mineral prospecting showed the bedrock to consist of metamorphic and sedimentary rocks of lower Devonian age in the south west part of the Kernick site, (Illsley

37 - ^ al. 1976). A wrench fault crossed the line of the dam but the fault zone was found to be reasonably tight. The superficial deposits consisted of sandy gravels containing a variable proportion of clay which were partially alluvial and partly sollfluction in origin. During flood tests on the first stages of the Rernick Dam (now built), 47.32 m^ sec"^ were pumped in and a surprisingly high loss was encountered. Water escape, through the jointed granite valley base beneath the mica residues was at an estimated 0.19 m^ sec"^. This discharge found its way into the next valley giving artesian back pressures in the valley gravels, (Brlstow, pers. comm. 1980). A lag time of three days was noted for the artesian effect to become apparent, and 20 days from tritium measurements in the test waters. From this isolated example it can be seen that adequate knowledge concerning the hydrogeologlcal characteristics of decomposed granites is essential. Inferred properties from one area may well be useful in the assessment of dam sites and waste disposal Interests in other decomposed granite zones in the South West region. 1:3:6 Mining and Geothermal Interests Hydrogeologlcal knowledge on granite and weathered granite areas have other uses besides those directly related to groundwater resources and reservoir construction conditions. Such information, if available on specific areas, may be helpful in mining operations. Hemerdon Ball lies on the south west edge of Dartmoor, 1.5 km outside the National Park boundary. It has been the site of mining activity since at least the 19th century when underground operations were designed to exploit tin, copper and arsenic lodes. Local mining activity is currently restricted to the exploitation of the kaolinised granite to the north and east of Hemerdon. At present Amax Exploration U.K. are on site evaluating the country's largest known tungsten deposit. Since 1977 500 boreholes totalling 25,000 m have been drilled into the granite and surrounding Rillas. Information concerning the permeability and porosity of the granite and associated weathered zones, and the distribution and likely amounts of circulating water are required to assess dewatering operations for deep mining schemes in the area, (Jones pers. comm. 1980).

- 38 With the advent of potential geothermal energy sources from the Cornish granites, (Bachelor and Pearson 1979), there is an increased demand for hydrogeological studies on the granites in south west England. The heat Is contained in a mass of granite, like the Carnmenellis granite and can only be liberated by cooling the rock. The natural hot springs (43*C) of South Crofty mine in Cornwall are formed by fresh water that circulates through fissures and extracts the heat over many square metres (Bachelor and Pearson 1979). The most satisfactory method for heat extraction is the creation of a zone of broken rock so that water can permeate it with ease. Knowledge of the presence and mobility of water in the granite masses under observation for these types of investigations are desirable. From current research on the Carnmenellis granite the deep circulation of water seems to be limited to depths of 2000 m with residence times of 20-30 years. This flow is in the joints and the geochemistry appears to be related to a water/granite system. On this basis it is unlikely that the circulation of water in a Hot Dry Rock system will produce major toxicity hazards (Bachelor, pers. comm. 1980). 1;A Conclusions

The best chance of finding water in a granite region is where the granite itself is covered by a thick layer of weathered material, specially if this overlies a depression in the weathered granite, which is Itself intersected by a well developed joint system. Where there is sufficient permeability in the weathered and disintegrated granite, groundwater Is likely to be contained in a manner comparable with that In rocks having Interstitial porosity. Such potential aquifers have never been held In a favourable light since little is known about the hydraulic and water-yielding properties of weathered granite areas, or of massive granite zones. It seems likely that such supplies would only be small-time concerns for intermittent use in times of severe drought, and the expense of developing such isolated sources may be prohibitive. Perhaps they could be Incorporated into the water distribution net work in a similar way to alluvial aquifer resources as envisaged by Monkhouse (1978) and Jeffcoate (1977). Certainly these two types of unconsolidated materials have similarities; both weathered granite and

•4

39 alluvial deposits being of limited lateral extent; sometimes lenticular In disposition; of restricted depth; and with a great variation in hydraulic properties. It is felt, however, that in regions such as south west England, which has severe water-supply problems, particularly in the summer months with its huge seasonal influx of tourists, that any potentially available source should be Investigated. Indeed, small sources could prove Invaluable in this region, especially if drought conditions such as experienced in 1976 occur again. That this groundwater investigation has been instigated fulfills one of the objectives of this research project, and to clarify further objectives the major aims are developed in the ensuing chapters as summarised below: Chapter 2 "The Hydrogeological Basis"

This deals with the bedrock geology illustrating the differences between solid unaltered granite and the decomposed granites consisting of chemically and physically weathered materials in association with hydrothermally altered granites. Physical properties of the Narrator Brook Catchment are outlined in conjunction with hydrologlcal features, and the resulting groundwater and surface water characteristics are discussed. Drilling in the catchment area, in as much as It relates to the hydrogeological properties of the decomposed granite, is also discussed in this chapter. Chapter 3 "The Data Collection Network in the Narrator Brook - Field and Laboratory Techniques"

This chapter deals exclusively with the experimental design and the resultant implementation of the data collection network. The emphasis on the type of data collected from this catchment, is biased towards the acquisition of hydrogeological properties which may be representative of weathered granite aquifers elsewhere.

Chapter 4 "Discharge Characteristics in the Narrator Brook Catchment"

40 since surface water and groundwater regimes are Interrelated, the analysis of stream flow data will provide valuable Information regarding the groundwater component In a weathered granite catchment. This chapter deals with stream flow sources» discharge measurementy hydrograph analysis, groundwater recession, active storage, the extent of the hydraulic continuity between the stream and the aquifer, and presents a tentative water balance for the Narrator Brook Catchment. Chapter 5 "Groundwater Fluctuations in the Narrator Brook Catchment"

It was considered that some of the hydrogeologlcal properties of the weathered granite aquifer In the Narrator Brook Valley could be assessed through the monitoring of groundwater level fluctuations. This chapter presents an analysis of the water level data collected during the research period In the form of groundwater contour maps; statistical analysis of water levels, soil moisture and hyrometeorologlcal parameters; and a discussion of such results in conjunction with other factors particular to the Narrator Brook which may Influence the groundwater regime. Chapter 6 "Aquifer Properties in the Narrator Brook Catchment" The techniques utilised and the data collected in an attempt to derive aquifer properties for weathered granite areas, are outlined in this chapter. The use of pump tests and slug tests in deriving hydraulic conductivity and transmisslvity of weathered granite materials are assessed in the light of conditions in the Narrator Brook. Chapter 7 "Chemistry of Groundwater in the Narrator Brook Catchment" The groundwater chemistry of a weathered granite aquifer Is presented and discussed in the light of determining the direction and transit time of groundwater recharge. Recharge and discharge areas as

- 41 delineated In Chapter 5 are substantiated by the geochemistry. Chapter 8 "The Potential of the Narrator Brook Aquifer, Summary and Conclusions" The potential of weathered granite aquifers as illustrated by studies in the Narrator Brook are summarised and discussed. Recommendations for future work relevant to weathered granite aquifers are also presented.

- 42 - Chapter 2 The Hydrogeological Basis

2:1 Introduction Although this Investigation is based primarily on observations in the Narrator Brook catchment, south west Dartmoor, it is considered important to review factors relating to the hydrogeological conditions prevailing in all granite areas of south west England. This serves the dual purpose of providing supplementary data on probable conditions within the Narrator catchment in addition to demonstrating the degree to which the Narrator catchment may be taken as representative of other valleys on the granites of south west England. It is in this light that the results from the Narrator catchment may provide a basic model of groundwater flow conditions and hydraulic properties in weathered granite aquifers. According to Dixey (1950), the geological character of an area largely determines the pattern of water supply. Since the present emphasis is on potential groundwater resources, the discussion here will focus on those factors related to aquifers, although some parameters may be relevant to both surface water and groundwater resources. Parameters such as porosity, permeability, hydraulic conductivity and saturated thickness of an aquifer are relative terms, dependant on local conditions. This necessitates a detailed assessment of the geological properties of the material underlying a catchment, and subsequent considerations of their hydrogeological implications in a groundwater investigation. This is achieved through an Initial examination of the geology of an area as a whole, to bring any problems into perspective. As illustrated In Figure 2:1, the Dartmoor National Park includes the whole of the granite outcrop and its associated metsimorphic aureole with its rich mineral zone, set within a varied assemblage of sedimentary rocks. For the purpose of this study 'Dartmoor' will refer to the granitic intrusion alone, and 'the catchment' will refer to the Narrator Brook experimental catchment as used by the Department of Geographical Sciences, Plymouth Polytechnic,. Devon.

43 FIGURE 2:1 LOCATION OF THE NARRATOR CATCHMENT

/ ^ '-^ + + + ^ ^-.v'- .-•+ + + + + + n ^ N >'*-r + + + + + + + +^^ ••+ + + + + + + + + + + ^ + + + + + + + + + + + + :)- + + + + + + + + + + + + + -> :/+ + + + + + + + + + + + + + -? ;, + + + + + + + + + + + + + + + + + + + + + + + + + + + + +\ *+ + + + + + + + + + + + + + + +, + + + + + + + + + + + + + + - , + + + + + + + + + + + + + + + +/ + _+ + + + + + + + + + + + +^+ +>• »+ + + + + + + + + + + +/ yi- + + + + + + + + + + +, •+ + + + + + + + + + + • If + + + + + + + + + + + + ' %+ + + + + + + + + + + + + + + +^/ '+ + + + + + + + + + + +^ '- + + + + + + + \+ + +-.+ + + + + V + + t^-t>-\- + + + + + ^ + + + + + + / ^4- -iS^^^^f^ + + + + -V * + + + +/ ^••^'^•^^^^^i^ + + + + , Biucrator Reservoir ^^;%%%%/ ++++++++'-+ + + + + + ^^ +,+ + + + + + +» ~"^ + + + + + + -*'' J+ + >\ + + +

' + + + ^ /+ + + I + + +1 PLYMOUTH - + + •

Scale 1-5cm =5km

KEY + + + + + + DARTMOOR DARTMOOR + + + GRANITE NATIONAL PARK BOUNDARY .pliiinni BURRATOR • + * i.*NARRATO R RESERVOIR BROOK CATCHMENT. GRANITE BASED CATCHMENT 2:1:1 Solid Geology The granites of south west England occur as six large outcrops lying on a continuous sub-surface 'ridge' trending west-south-west from Dartmoor, to the Isles of Scllly, Figure 2:2. Dartmoor itself is a cupola of adamellitic granite composed of several distinctive rock types. Structurally it may be an asymmetric laccolith, in which the magma appears to have risen steeply from the south, from a depth of 16—19 km, and spread northwards as a sheet whose bottom surface Is about 9.5 km below ground level (Edmonds et al., 1975).

Gravity measurements by Bott et al. (1958) suggested that the southern slopes of the cupola are steeper than their northern ones, and later work by Dearman (1964) verifies this. The granites of Devon and Cornwall have age determinations, from Isotopic methods, of 295 million years old, (Fitch and Miller, 1964; Miller and Mohr, 1964). Dangerfield and Hawkes (1969) suggested that these granite intrusions occurred during the Upper Carboniferous thus agreeing with Fitch and Miller (1964) and Miller and Mohr (1964). The Dartmoor pluton Is composed essentially of two granite types termed by Hawkes (1968) the 'Big Feldspar Granite', which is equivalent to the 'Giant Granite' of Brammall and Harwood (1923), and the 'Poorly Megacrystic Granite' which is equivalent to the 'Blue Granite' of Brammall and Harwood (1923). The 'Giant Granite' overlies the 'Blue Granite' and has been interpreted as a marginal fades by Hawkes and Chaperlin (1966). Bramall and Harwood (1932) suggested that the Dartmoor magma was Intruded as two sheets, an upper one of 'Giant Granite' which forms most of the tors and is characterised by large rectangular feldspar crystals; and a lower one of 'Blue Granite' which intruded the upper sheet and contains fewer and smaller raegacrysts.

These two types of granite are Important in the light of this Investigation since they exhibit differing petrographlc and structural features which may have some bearing on their water transmitting characteristics. The giant granite is more massive and its major divisional planes are well developed with widely spaced vertical joints, often 3.6 m apart, with almost equally well developed pseudo- bedding planes 1.5 m apart. The closer set pseudo-bedding planes are

- 45 - FIGURB 2:2 DISTRIBUTION OF GRANITE IN THE SOUTH WEST A

Dartmoor Granite Scilly Isles Granite Moor Granite

•o

^^ :p/^VGranite

Lands End Granite Carnmenellis Granite

Scale 1cm • 10/cm minor features which although controlling surface etching, according to Edmonds et al. (1975), do not penetrate far into the rock. The blue granite, which Is finer-grained nearly non-porphyritic, with little biotite, is divided by several sets of closely spaced vertical joints. These define prisms or columns, the faces of which have been little affected by etching. The average chemical composition of the granites are outlined in Table 2:1 which summarises the geochemical analyses of Brammall and Harwood (1923), and Al-Saleh et al, (1977).

Table 2:1 Average Chemical Composition of Dartmoor Granites (%)

Ions Big Giant Poorly Blue Feldspar Granite Megacrystic Granite * ** *

SiO 72.3 71.56 74.6 74.77 Al 0 13.6 14.08 12,7 13.33 K 0 5,49 4.87 4.79 4.75 Na 0 2.91 3.01 3.24 2.87 FeO 1,74 2.21 1.11 1.38 H 0 1,23 0.88 0.95 0.90 Fe 0 1,17 0.55 0.67 0.44 CaO 0.92 1.55 0.50 0.63 MgO 0.50 0.60 0.22 0.47 TIO 0.40 0,42 0.17 0.20 P 0 0.19 0.21 0.20 0.18 MnO 0.06 0.06 0.05 0,08 Li 0 0.06 — 0.10 -

* Al-Saleh, Fuge and Rea (1977) ** Brammall and Harwood (1923)

The Dartmoor granite consists of quartz, perthitic orthoclase feldspar, plagioclase feldspar, biotite mica with secondary muscovlte mica being locally present. The commonest accessory constituents are tourmaline, zircon and apatite. Cordierite and andalusite occur at the edges of the granites and are likely to be the result of

47 contamination of the magma by argillaceous country rocks now forming the metamorphic aureole. More specifically, work carried out on the Swlncombe and Huccaby granites (4.5 km N.E. of Burrator) by Morton (pers. comm. June 1978) showed that these granites were of the coarse-grained 'Giant Granite'. At these two locations the granites were composed of large (up to 10 cm) phenocrysts of feldspar, with quartz and biolite mica. Muscovite was present with tourmaline as an accessory, and on the whole the granite was found to be more lime-rich than the other southwest granites, with less muscovlte and more plagloclase feldspar. According to Morton (pers. comm. 1978) the Dartmoor granite is less altered by hydrothermal action than the other granites in south west England. This is partly due to its composition, blotlte and plagloclase being more easily altered than muscovlte and orthoclase. There are, therefore, relatively few elvans, grelsen veins and kaollnlsed seams In the granite. The sediments which the granite intrudes are only slightly metamorphosed Upper Devonian deposits, but at the end of the period of Intrusion a burst of low-temperature activity produced rich mineralisation. This resulted In large deposits of tin, arsenic, copper, iron, wolfram, quartz and fluorspar. Tin was mined in the upper reaches of the Narrator Brook valley, and such activities and their hydrogeological implications in the catchment are discussed in section 2:3:9.

The probable underground shape of Dartmoor suggests that the present high tors lie near the top of the laccolith. The tabular joints, typical of most tors may parallel the roof of the intrusion or stress surfaces in the consolidating magma. Although the joints commonly appear to follow the topography it is unlikely according to Edmonds' ^(1969) that they are a product of weathering, A number of authors (Brunsden 1964; Oilier 1969) however, do attribute the horizontal joints of the tors to pressure-release or dilatation processes brought about by the removal of overburden by erosion. So far, the term decomposed granite has been used to distinguish incoherent crystalline materials from solid granite. This term

48 - includes chemically and physically weathered granites, and the alteration and subsequent breakdown of the crystalline mass due to hydrothermal processes. The specific types of alteration of the crystalline granite mass are individually assessed since each process results in increased permeability and water transmitting properties of the granite.

2:1:2 Alteration of the Granite During the final stages of cooling at depth, the granite gave off active solutions and volcanic gases above their critical temperatures, including volumes of fluorine and boron, along with sulphur dioxide, superheated steam and carbon dioxide. Changes brought about by the action of hot gaseous substances, other than water, associated with igneous activity are known as pneumatolysis. The distinction between this process and hydrothermal processes is largely arbitrary, but the distinction is useful in terms of the specialised minerals and rock types produced. The commonest volatiles involved in pneumatolysis are fluorine, hydrofluoric acid, and boron fluorides. Other substances may locally be present giving rise to unusual mineralogies. Pneumatolysis is a process associated with a late stage in the cooling of an igneous mass and may therefore affect both the country rock and the main mass of igneous material. The development of the borosillcates such as tourmaline may occur in both granites and thermal metamorphic aureoles. Three main pneumatolytic processes may be recognised: Tourmalinisation, Greisening and ore mineralisation, (i) In tourmalinisation, boric vapours altered the granite and adjacent calcareous muds and limestones, and partly converted them to axinite and garnet rock. The feldspars and micas are replaced by tourmaline, (ii) In greisening the feldspars are replaced by pseudo-morphous aggregates of quartz and muscovite, (iii) Tin and wolfram are ore minerals found in the south west granites attributable to pneumatolytic origin, but may also commonly be found in environments thought to be produced by hydrothermal processes* Hydrothermal processes are associated with igneous activity which involve heated or superheated water. Kaolinisation is the most

49 important hydrothermal process, resulting in the production of the clay mineral kaolin from the feldspars in granite. All granite masses In the south west are kaollnlsed in parts which Exley (1958) attributed to hydrothermal alterations effected by movement of acid solutions along joints. In kaolinisatlon the principal changes in the granite take place at the expense of the feldspars. Charoy (1975) suggested that kaolinisatlon can be visualised simply as a strong removal of alkalis and the production of silica accompanied by a strong hydration. The reaction can be expressed as 2 orthoclase + 2H+ + HjO -»• 1 kaollnite + 4S102 + 2Kt (Na+) The destruction of feldspar produces Important quantities of silica. Observations in Brittany (Charoy 1975) suggested that such alteration took place without change in volume, and the structure of the granite was perfectly conserved in the kaollnised rock. In the south west region In particular, tourmallnisation, greisening and kaollnlsatlon processes acting collectively on the granites break down the rock matrix. Such a physical change in the rock cohesion may locally Increase its porosity and permeability and hence give rise to better water transmitting facilities throughout the granite matrix. 2:1:3 Weathering Processes Workers on Dartmoor from the earliest days, (e.g. De La Beche, 1853) have observed and commented upon the Incoherent nature of the granite near the ground surface. Reid (1912) and Ussher (1912) distinguished two types of decay: one showing signs of chemical decomposition; and one exhibiting a predominantly mechanical disintegration. These two states are found in differing degrees at sites all over Dartmoor, and their characteristics vary. The agents of chemical weathering, carbon-dioxide and oxygen in association with water, (often with organic and other acids), together act effectively both at surface and subsurface levels. The

50 decomposition of a granite is produced by an alteration of the feldspars and micas. The first stage of decomposition is oxidation resulting in the ferrous oxide in biotite mica being converted to ferric oxides on the rock surface and Joint planes. Staining of the rock surface and Joint planes to a depth of a few centimetres follows, caused by the slight alteration of the feldspars and mica by hydration and hydrolysis. Hydrolysis Is a chemical reaction between the minerals and water, between the H or OH ions of water and the ions of the mineral. Hydration is the addition of water to a mineral. This reaction is exothermal and involves a considerable volume change (Oilier 1969). Eventually the increase in volume due to hydration produces sufficient stress for the rock surface to spall off, thus offering fresh surfaces to the attack of water. The actual chemical reactions involved in the process of hydrolysis are exceedingly complex, (Lumb 1962). They can be illustrated however, for present purposes by the following simplified equation for the change of potassium feldspars (orthoclase) to kaolinite: KjO AI2O3 . 6Si02 + 3H2O orthoclase feldspar water

AI2O3 . 2SIO2 . 2H2O + 4Si02 + 2 KOH kaollntte silica potash The silica and potash released by this reaction are in colloidal form and are eventually removed by leaching, resulting in a net loss of volume (Lumb 1962). With further weathering the mineral grains are loosened, allowing more rapid alteration of the feldspars and mica, while the quartz remains largely unaltered. The breakdown of the feldspars causes most of the calcium and sodium in the rock to be released and these elements are eventually leached out. At this stage in the weathering of granite the porosity increases rapidly due to the leaching out of the colloids. The quartz grains, being unaltered, remain in the same state and position as In the fresh rock, so the resulting material in situ still preserves the texture of

- 51 the original granite, although it can be excavated easily with a shovel (Moye 1955). Elluvlatlon of the clay minerals and of iron and calcium, may result in zones within the weathered granite matrix possessing high permeability. This process was considered to be an important one by Brunsden (1964) but not considered to be significant by Eden and Green (1971). The downward leaching of clay materials and fine silts and their deposition resulting in llluviated horizons effect the porosity of the regollth In situ, and result in textures with a different form to the original crystalline granite matrix. The decomposed granite on Dartmoor is referred to as 'growan' (Ussher 1912). Several authors, Linton (1955), Waters (1957) and Brunsden (1964) attributed growan largely to chemical weathering processes probably occurring during the Tertiary era or during Pleistocene Interglacials. Te Punga (1957) however attributed 6 metres of 'growan* on Bodmin Moor to Pleistocene frost shattering, and Palmer and Neilson (1962) suggested that some of the growan found in valleys Is largely sollfluction debris, Brunsden (1964) suggested that growan owes its origin to chemical and mechanical weathering along with hydrothermal alteration, since all forms of granite break• down and alteration are found on Dartmoor, often in the same location. Brunsden (1964) also suggested that some kaollnisation of the Dartmoor granite occurs under the action of chemical weathering. Brlstow (1977) mentioned that some features in china clay deposits of Cornwall and Devon, seem to argue against a weathering origin. The lack of evidence of any zonatlon in the profile, (a kaolinised granite is the same at the surface as it Is 200 m down), is unusual for a weathered sequence, especially as there Is no evidence of de-sill- cation of quartz grains near the surface. Bristow (1977) concluded that the only acceptable explanation for the origin of weathered granite is a combination of hydrothermal alteration and late supergene origin, in association with weathering processes. 2:1:4 Nature of Decomposed Granite While discussing the weathering products of granite the emphasis will be more on the growan which, because of its porous nature, is potentially more useful in terms of its water transmitting and storage

52 capabilities. While kaollnisation results in a physical breakdown of the massive granite, its by-products, namely kaolin clay, is not a suitable water-bearing material, and such deposits may represent local aquicludes, since their permeability is negligible and specific yield is low. Decomposed granite consists either of moderately decomposed growan, or of greatly altered china clay (Green and Eden 1971). Kaolinlsatlon results in a material containing relatively high silt plus clay content. Samples of china clay from Lee Moor contained 1.5 to 30.5% clay and 30.0 to 72.0% silt plus clay (Eden and Green 1971). Quartz forms the bulk of the sand fraction in kaollnlsed areas, with some feldspar, mica and tourmaline fragments often present. China clay is normally white to pale grey in colour except in superficial layers where oxidation has occurred. Growan has a lower silt plus clay content (13.5 to 28.0%) and a clay content of 2 to 10.5%. These values were obtained from in situ growan samples, where a high residue of feldspar crystals were found in the remaining sand fraction (Eden and Green 1971). Under field examination quartz and mica show little modification and some tourmaline is present. Growan exhibits commonly brown to reddish brown colours, but whitish and greyish materials have also been described by Brunsden (1964). In situ growan normally retains a visible crystalline rock texture which can be broken in the hand. Clay movement along joints and mineral grains are evident in some samples (Brunsden and Gerrard 1970). The final product is a sandy or sllty layer that is rich in iron minerals, and oxidation of Iron from micas and other ferromaghesi'onminerals produces the distinct red/brown colour. Growan is only moderately decomposed, although Brunsden (1964) refers to it as 'well-rotted and incoherent granite'. Implying that in places substantial decomposition of the rock has occurred and has been followed by the removal of weathering products from the growan. However, taking into consideration the high feldspar content and persisting rock texture of the growan, one may infer that there has been only limited weathering and removal of such products. According to Eden and Green (1971), the low clay content can be attributed more to limited decomposition than leaching processes.

53 Growan generally occurs as a laterally extensive layer, frequently overlain by head deposits, and having a lower boundary with more solid rock. Transported growan, termed 'bedded growan' by Waters (1964) is also observable at different locations on Dartmoor, and Palmer and Nellson (1962) suggested that much of this, and the growan observed from valley floors, is the result of solifluction processes.

The superficial deposits of the Narrator Brook valley may consist of a mixture of 'bedded growan' and in situ growan. No attempt was made during this investigation to differentiate between the types of incoherent valley fill deposits but an appreciation of their possible origin, in the Narrator Brook valley is recognised. Soil material over sedentary growan is usually 40-60 cm in depth and is relatively compacted, normally lacking the original rock texture observable in the growan, A higher degree of mineral breakdown is encountered giving 57-62.5% sand, 26,5-29,5% silt, and 11-13.5% clay (Eden and Green 1971). More specific details concerning soil properties on Dartmoor and the Narrator Valley are in section 2:3. 2:1:5 Classification of Weathered Granite At this stage in the discussion it may be useful to define the plethora of terms used to describe the decomposed granite on Dartmoor. The following categories are based on Moye's (1955) classification which he drew up in order to closely define the terms used to describe the various degrees of weathering in granitic rocks. Hamrol (1961) and Lumb's (1962) observations are also incorporated as both authors agreed that an objective method of determination of the degree of weathering was required. Granite soil: Derived from granitic rocks, often does not possess any recognisable granite fabric. Exhibits a higher degree of mineral breakdown, consisting for the most part of sandy material with small quantities of silt and clay. Completely decomposed granite: A matrix of sllty coarse sand comprising quartz grains and alteration products (clay minerals) along with 'core-stones' and boulders. The original granitic texture is still apparent, although no feldspar remains. This material can be broken up with the hands.

54 Slightly decomposed granite: Silica and potassium are freed in the colloidal form and removed by leaching. The mineral grains are subsequently loosened allowing more rapid alteration of feldspars and mica. Relative to the unaltered granite bedrock, the porosity increases. The resultant material is mostly quartz and feldspar with a little clay. Some staining is evident on joint surfaces. The rock can be broken by kicking the surface. Altered granite: Shows cracks, and grains of feldspar and mica are slightly altered on their surfaces by hydration or hydrolysis. The increase in volume by hydration produces stresses so the rock begins to break up. The rock can be broken with a hammer. Fresh granite: This lies immediately beneath the various types of weathered granite and frequently shows limonite stains along Joints, indicating paths of water movement. The minerals are fresh and the mass is solid and generally impermeable except where intersected by fissure and joint systems. A hammer blow tends to bounce off the rock.

Terms such as rotted or incoherent granite are generalised terms which refer to the altered to completely decomposed granite as defined in the above classification. Growan would be placed in the slightly decomposed to the completely decomposed material In this system, 2:1:6 Other Superficial Deposits The direct effects of glacial ice in south west England are restricted in extent since there is no evidence of an overriding ice sheet, the nearest ice front being in . The results of Pleistocene climatic changes, particularly cold conditions, are much more evident and widespread, with the result that Dartmoor is often described as a relict periglaclal landscape (Gregory and Ravenhill 1969).

Periglacial conditions in the south west manifest themselves in deposits of 'head*. These deposits in general consist of angular fragments embedded in a finer matrix, (Waters 1971). In most cases they are typified by an anomalous particle size distribution in that there is a large amount of coarse angular material and much fine-grade material with a conspicuous lack of material of intermediate size.

55 It is inferred that the fine matrix which encloses the larger angular fragments may have been derived from weathering products under earlier climatic conditions. Two head deposits have been distinguished and described at various inland and coastal sites in south west England. These upper and lower head deposits are fully described elsewhere (Stephens 1966; Waters 1971), but their main features will be outlined briefly here as they may be important in terms of water-bearing and transmitting capacities in this area. The lower head is usually the thicker of the two deposits and consists of a dull yellow to strong brown sandy or gritty loam containing coherent but easily broken granite stones up to 15 cm in length (Waters 1971). The upper head is mostly thinner and fresher in appearance than the lower or mainhead deposits. On the granite outcrops around the tors the head is represented by the 'Glitter* of granite blocks strewn over the surface with very little fine matter in its matrix, (Stephens 1966). Such deposits are characteristic of the slopes of Sharpitor and Leather Tor in the immediate vicinity of Burrator Reservoir, and of Combshead Tor in the Narrator Brook Catchment. These may form valuable infiltration sites for groundwater replenishment of the valley aquifer in the Narrator Brook region. 2:2 Hydrogeological Properties of Valley Deposits and Granites in South West England The disposition and depths of superficial materials are important factors in determining the potential for localised aquifers on the granites of south west England, and the Narrator valley in particular. Granite may be weathered to great depths as illustrated in Table 2:2. Evidence concerning the depth of decomposition of the granite on Dartmoor is somewhat scattered and often incomplete. Information compiled from experimental trenching and trial boreholes on site investigations for proposed reservoirs, and other water supply schemes have provided the most useful data. Fortuitously a large number of these investigations have been carried out in close proximity to the Narrator Brook valley. These investigation sites are illustrated in Figure 2:3.

- 56 Table 2:2 Depths of Weathering in Granite

Depth of weathered Location Author granite (metres)

12 Scotland Fitzpatrick (1963) 40 Altna Breac Glendlnlng (1980) Caithness, Scotland 60 Hong Kong Ruxton and Berry (1957) 60 Peterhead, Scotland Moore and Gribble (1980) 91 Uganda Oilier (1960) 100 Czechoslovakia Demek (1964) 274 Snowy Mountains, Moye (1955) Australia

2:2:1 Burrator and Sheepstor Burrator Reservoir is situated 1 km south west of the Narrator valley. The and the Narrator Brook discharges contribute to Its impounded volume. The reservoir (N.G.R. S.X. 552680) is formed by two embankments, one of masonry across the narrow gorge of the Meavy, the other of earthwork lying between the rocky outcrops of Sheepstor and Burrator. Sandeman (1901), during the construction of Burrator Dam (in 1893-1895) discovered solid granite at depths varying between 0.60- 12,19 m in the Meavy valley. The average depth to solid granite was 12.19 m, with 16.15 m below the old river bed and 21.33 m on the west side of the valley. The disposition of solid and weathered granite is illustrated in Figure 2:4a. Two trial pits were sunk in the region of Sheepstor Dam (N.G.R. S.X. 558673), but these gave a deceptive picture of subsurface geology since they were situated on rock pinnacles from which the bedrock shelved away rapidly. These are Illustrated in Figure 2:4b* Depths of weathered granite as discovered in an excavated trench varied from it>26 to 30.48 m.

- 57 FIGURE 2:3 WATER RESOURCE INVESTIGATIONS AND RESERVOIRS ON DARTMOOR

Dartmoo Granite

PLYMOUTH

Scale 1-5cm = 5km

KEY Reservoirs Investigation Sites Groundwater Development

A Avon C Cowsic TM Taw Marsh B Burrator H Huccaby F FernworthyHt Hart Tor K Kennick NB Narrator Brook M Meldon S Swincombe To Tottiford Tr Trenchford V Venford FIGURE 2:4a

SHEEPS TOR EMBANKMENT, GEOLOGICAL SKETCH

Elvan Rock Decomposed Veins and Stringers OGranite Q

+ + +C;ranitordiiiit;e ^ + + + + + + + + + + + + + + + + + + Blocks

FIGURE 2:4b BURRATOR DAM, GEOLOGICAL SKETCH

W Vertical and Level of Dam Horizontal Scale + + 1mm ° 1m + + + H + + + + + + + + + + + + H + + + + + + + + + + + + + + + + H + + + + + + 'djrock + + + + + + + + + ++•+++ + + + + + + + + + + + + +.+ + + + + + + +'+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + c^itj r^^^-^i++ + + A + + + + + + + + + + + + + + + + + + + + 4: + + + + Solid Granite+ + + + + + + FIGURE 2:5 GEOLOGICAL CROSS SECTION HART TOR VALLEY after E. Mortons borehole logs (1944) showing disposition of decomposed and solid materials

Position of NW boreholes SE (A 0) 1 ye//ow sands and 360 qranite boulders . River Meavy nan Tor Brook "with .coarsergranitic E sand at dept 350 partly decomposed granite + + + + + + + +^.^ \ •T + + + + Q + + + + + + >s 4 + + + + + + O + + + + + + +'^. ^ _ + + + + + + + + +. + + + + + ^ + + + + T- ^ + + + + + + + + + + + + + + + + +^—^. + + + + + + + -^^-F + + + +.+ + + + + + + + + + 4- + + + +^+ + + + + +' + + + + + + + + + + + + + + + + + + + + I + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + CO + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +-* + + + + + + + + + + + + + + + + HCnIiH nrpnito \A/ith+ + + + + + + + + + + + + + + + O '5 + + + + + + + + + + + + + + + + ^0110 uranite wiin + + + + + + + + + + + + I 300 + ^ + ^ + ^+ + + + + + + + + ^horizontal and^oblique jqints*^. '^^a.^a.^a.^ . ^ . ^ . ^ . ^ . ^ ./"^^o.^ 100 200 300 400 500 530 2:2:2 Hart Tor Valley The Hart Tor site, (N.G.R, S.X. 577717), Is 2.5 km due north of the Narrator Brook Catchment astride the river Heavy and Hart Tor Brook. Ten boreholes were sunk In a northwest-southeast direction. These revealed solid granite at depths of 4.87 to 20.11 m. An appreciation of the disposition of the rotted granitic material in the Hart Tor valley can be obtained from the cross-section in Figure 2:5. The section runs northwest-southeast and it is noticeable that a deeper layer of loosely consolidated material is evident on the south east side of the valley. 2:2:3 Cowsic Valley The Cowsic site (N.G.R. S.X. 596765) is approximately 7.5 km north east of the Narrator Brook Catchment, astride the Cowsic river valley and part of the . Eight trial boreholes were sunk in 1944 at this location and revealed granite described as being of low bearing potential. This weathered granite was proved to depths of up to 36.57 m as illustrated in Figure 2:6. 2:2:4 Swincombe The Swincombe site (N.G.R. S.X. 625712) lies 4.5 km north east of the Narrator Brook Catchment astride the Swincombe river valley. Preliminary work was carried out on this site in 1968-1969, with the intention of constructing an impounding reservoir. Morton (1969) published a report outlining the geological findings and recommendations. Access to the borehole records have allowed the reconstruction of the geological cross-sections as illustrated in Figure 2:7, Twelve rotary cored holes were drilled, and the unweathered granite was encountered at 1.37-15.24 m under the two hillsides. 2:2:5 Huccaby The Huccaby site (N.G.R. S.X. 653735) lies 8 km north east of the Narrator Brook valley, just below the Dart river and its confluence with the Swincombe Valley. Ten rotary cored boreholes were drilled, from which the geological cross-sections (Figure 2:8) were constructed, which revealed bedrock from 8.14 to 34.23 m below the surface. All boreholes in this instance terminated in slightly weathered granite. 2:2:6 Taw Marsh Taw Marsh (N.G.R. S.X. 619915) situated 5 km south east of Okehampton on northern Dartmoor, was first examined seriously as a source of - 62 - FIGURE 2:6

GEOLOGICAL CROSS-SECTION OFTHECOWSIC VALLEY, DARTMOOR

(after Borehole logs, E.Morton,(1944))

Position of X boreholes (D +\ (5* + \ 2: 0) + + \soft decomposei + '*' + '*" + "* :^ aranite with coarse'"'"""^^ 2 Devon Port f + + + + ^^ranitic sands at deptT Leat CD + + with boulders + + Cowsic O + + + + + + ^ • + + + + + +> River D + + + + + + + > + + + + + + + ^ ^ 4 3 + + + + + + + + +^^ + + + + + -*• + + + +'^ 400 + + + + + + + + 4- + + -,^ 3 + -h + + + + + + + -f +V + + + + • • , -t- , ' • + + ^ ^ iS + + + Hard Granite with + H ^ (0 oblique and vertical joints + + + + + + + ^ and crystalline veins + + + + + + + -h + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 4- + + + + + + + + + -(- + •*• + + + + + + + + + + + + + + + + + + + + + + 4- •!- + + + + + •h + + + + + + + + + + + + + + + + + + + + + + + + + + + -»- + + + + + + + + + + + -J- + + + -I- + + + + + + + H- H- + + -t- + + -H + -I- + -t- + 60 + + + + + + + + 4- + + + + + + + -h + + + + + + + + + + + + + + + + + 4- + t- f- + + + + + + + -t- -H + + + + + + + + 4- + + + + + + + + + + + 100 200 300 400 Scales Vertical 1cm • 10m Horizontal 1cm » 20m FIGURE 2:7

GEOLOGICAL CROSS SECTION OF THE SWINCOMBE VALLEY SHOWING PROBABLE DISPOSITION OF ALTERED AND SOLID GRANITES N SE 380

o Position of T3 O boreholes CO ^ 57^ •o Peat and thin topsoil + + + 3- + + + > overlying patches of 350 o' + + + +7\ highly-slightly ^ SL + + + '+ \ altered granites ^7'^'*' I + + + + + + + + + +x CD + + + + + fA Granitic Soil , ^+ + + (5' + + + + + 4- -iN + + + + + + iS + Solid + + + ^ Possible depth^ > + Granite^ + + + ?^ 7+^^ of solid granite b r + H- • + + + + -t- + + + +N -Highly weather boundary , + / + + + 1 . . . + D + + + + + + + + + + tv graniie /+ + + + + + + + + 3 /+ + + + + + + + + CD + + + + + + + + + + -t^^ •>f + + + + + + + + + 300 r& /+' + + + + + + + + + -1 + + + + + + + ,+ + + + + + + + + + + CD H- + + + + + + + + + + +\ .+ + + + + + + + + + + V) + + + + + + + + + + .+v V + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +•\ < + + + + + + + + + + + + + + + + + + 4-4. + + + +^^,-.4: + + + + + + + + + + + + 200 400 600 800

Scales Vertical 1cm » 10m Horizontal 1cm • 50m FIGURE 2:8 GEOLOGICAL CROSS SECTION OF THE HUCCABY AREA SHOWING PROBABLE DISPOSITION OF ALTERED AND SOLID GRANITES

Scales Vertical 1cm » 10m Horizontal 1cm = 50m

SW Position of 3301 HB boreholes H4 I Thin topsoil H6

I West Dart 300 + + + + + /River + +?+ + + + + + H5 + + + + + + + • H2 + + 4- + + + + + + + + + + + + + + + + • ++ '•** ^ "T^ranitic sands and f raamer^s----rrr^'^ + + + ^ + -t- + 4- + + + + + + + + + -i- i- "CS.^ X '+ + + H + + + + + + + + + + + + + ^ + + + + + 1 ?:*: Hiqhly weathered + -•+ + + + + + Solid Granite + + + + + + + •^'"i-. granite + + + + + + . T X -r + + + + + + + + + + + + + + + + + + + + + + + •t- + 4- + + -I- + Z^ + '^H Possible depths 1 •4- + + + + + + + + + + + + + + + + + + + + + + -I- H- + > -I- 4 V \ *i*tiU^.-'—' X+ + + of solid granite + + + + + + + + + + + -f + + + XSIiqhtly weathered.^ + + + + + + + + + + + + + + + + ^• V materiaL>f^ + + ^/*^ + ^ boundary ^ + + + + + + + + + + + + f f + + + + + + + -t--rTr-H + 250! + + + + -H + + + + + + + + + + + -f- + + -t-+f+ i-H- + -I- ^- + f- + + + H- + + + h + + + + + + + + + + + + + •*• + + + + + + + + -t + + + + 1- + + + + + + + 4- + + .I- + + + + + + .f- + .f. ^ -I -I- + + + + + + + •(• 100 500 1000 1300 Irishmans / Wall Scale 5cm - 3km

2 + KEY

•ia,i,2,3,4 Borehole position Taw-e.^:^^:^ (logs in appendices) Marsh ~

£^=ifMarshy region t

I FIGURE 2:9a SKETCH MAP LOCATION OF TAW MARSH, N.E .DARTMOOR

FIGURE 2:9b DIAGRAMATIC SKETCH OF THE GEOPHYSICAL SURVEY (EAST OF )(1956) SHOWING DEPTH TO SOLID GRANITE

2-43m 6-40m 46-asm 25-90m 3 •04 m NNE 12-19m SSW 4-57m 975m 38 •40m I + + I- + + + + + + + + + + + + + + + + + + + + + + + + + + f + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + f + + + + + + + + + + + + + + + + + + + + + + + + + + water in 1956, (Figure 2:9a). A geophysical survey across the river Taw valley in a north northeast-south southwest direction, found bedrock in the valley centre at a depth of 36.57 m. Later borehole evidence supported this finding, recording depths to bedrock of between 2.43 m and 46.93 m, as illustrated in Figure 2:9b. This cross-section of the Taw valley, as elucidated from geophysical evidence, is of a diagrammatic nature only, suggesting the mode of disposition and the relative depths of incoherent material overlying the granite at this location. The overburden in the Taw valley consists of sands, gravels and clays whose bedding was described as being extremely lenticular. Strata in adjacent boreholes therefore showed little correspondence in lithology. 2:2:7 Fernworthy Dam This dam (N.G.R. S.X. 655842) which impounds the South Teign river, is situated in the north east edge of Dartmoor 17.5 km north east of the Narrator catchment. The depth of superficial material removed for the excavation of the dam extended to an average depth of 4.57 m. This included peat, surface boulders, sands, gravels and loosely decomposed rock. Below the superficial layer were found granite and elvan, partially decomposed, broken and fractured in the centre of the valley. This necessitated extending the excavation in many places to reach sound granite to depths of 15.24 m (Figure 2:10). Further evidence on the depth, disposition and nature of weathered potential water-bearing materials overlying the granite can be illustrated from data collected at reservoir sites on the granite and granite contact zones elsewhere in the south west. Although these are not related specifically to Dartmoor and the Narrator Brook they serve the purpose of highlighting potential localised water sources in decomposed granites, and so are included here. 2:2:8 Sibleyback Reservoir Sibleyback Reservoir, completed in 1969, is constructed on a tributary of the river within the area underlain by the Bodmin granite. The sites investigated were on the higher gathering grounds of the , and decomposed granites were found to extend to a depth of 21.33 m (Farrar 1969).

- 67 FIGURE 2:10 GEOLOGICAL CROSS - SECTION, FERNWORTHY DAM,DARTMOOR

Peaty subsoil Original with boulders Surface of Sedimentary materials Grouni of sands, gravels and River boulders South Teign

r + + + mmmrnm^ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Fractured + + + + + + % Rock mm + + + + + . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +^ + + + + + + + 4--^',., + + + + + + + + + + + + + + + + Black schor + + + + + + + + -1 bond + + + + + + + + + + + + + + + + + + + + + + Granite >- + + + + + + + + %%%%%%%^withvertical< + + + + + + + + + joints + Imnlndp + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 1 ^iruniuut? ^ 4. + ^ + + + + + + + + + + + + + + + + + +

Scale 70mm =25metres 2:2:9 Drift During the construction of the Drift Dam> on the Lands End granite (N.G.R. 435290), a steep-sided belt of altered granite, 20 m in width was encountered in the valley floor. This material had not been penetrated by vertical boreholes drilled during the site investigation* 2:2:10 Stithians The Stithians Dam is located 2.5 km north west of Falmouth within the outcrop of the Carnmenellis granite. Near the surface, rocks are severely weathered to a depth of 1.5 m, being altered to a gravelly granitic sand with occasional core-stones distributed along the valley sides* Excavations and boreholes demonstrated that fresh rock occurs at 15-20 m depth. The rock here is well-jointed, and has been kaolinised to varying degrees. In the centre of the valley the kaolinisation extends to more than 24 m in depth. 2:2:11 Summary of Site Conditions Table 2:3 summarises the previous information concerning depths to solid rock in the granites of south west England* Th.'sedata substantiate Brunsden and Gerrard's (1977) generalisation that many valleys in granite masses are underlain by broad zones of altered granite. In these Dartmoor valleys the thicknesses of unconsolidated materials lenses out towards the valley sides, with interfluKe areas being composed largely of solid granite. This can be seen from the geological cross-sections (Figures 2:4-2:10) particularly in the Hartor and Cowsic valleys. Hydrogeologically this feature Is significant. If groundwater is to be held or transmitted through weathered granite alone, and assuming no drainage fissure-systems out of the solid granite base of the aquifer, then such catchments may be considered to be watertight. Decomposed granite extends to deeper levels beneath the valley centres because differential weathering has worked preferentially along weaker features* Major decomposition is associated with mineralisation of the granite, as in the case of Fernworthy and Burrator. The mineralisation is probably late Carboniferous or Permian in age. Dines (1950), although the valley formation however is probably not earlier than the Tertiary (Palmer and Neilson 1962). Despite the absence of dominant regional trends in jointing, (Worth

69 - 1930), there is considerable local accordance between valley and joint directions, as in parts of the Cowsic, Meavy, Webburn, Taw, Dart and Plym valleys (Brammall 1926).

Table 2:3 Depths of Incoherent Granitic Material in South West England

Site Location Depth to solid granite

Burrator W. Dartmoor 0.60-12.19 m Sheepstor W. Dartmoor 4.26-30.48 m Hartor 3 km N. of Burrator 4.87-20.11 m Cowsic 7.5 km N.E. of Burrator 12.19-36.57 m Taw Marsh 5 km S.E. of Okehampton 2.43-46.93 Drift Dam Lands End granite 20 m Stithians Dam N.W. of Falmouth Cornwall 1.5-24 m Swincombe 4.5 km N.E. of Burrator 1.37-15.24 m Huccaby 8 km N.E. of Burrator 8.14-34.23 Sibbleyback Bodmin Moor, North of 21.33 m

2:2:12 Main characteristics of Solid and Weathered Granites Upon examination of the available geological data for these sites it is evident that there are four characteristics common to all locations of weathered and solid granite, which may be of considerable hydrogeological importance. These features are as follows; (i) the presence of fissures and joints in the solid granite bedrock (ii) kaolinised sections in decomposed granite areas (Hi) veins, stringers and intrusions (iv) the nature of the solid rock/decomposed rock interface. (1) Fissures and joints. During the excavation of the trench at Burrator several fissures in the solid granite, full of decomposed material were unearthed. These fissures were wedge-shaped and ranged from 0.6-3.0 m. In width, but decreased with depth. When solid granite was exposed it was found to have many joints and cracks, some.

- 70 according to Sandeman (1901), particularly on the sides of the valley being quite open* Other observations on jointing in granite were noted from the Swincombe and Huccaby sites where the granite mass is bisected by two sets of joints roughly at right angles to each other, and a third set, referred to as pseudo-bedding or dilation features, are apparent. Open jointed granite was apparent down to 25 m in the excavation trench at the Fernworthy site* (ii) Kaolinised sections The excavation trench at Sheepstor was cut to a depth of 32 m through an extensive layer of decomposed granite which, according to Sandeman (1901) exhibited a remarkable variety of colours* A vein of white china clay crossed this trench at its deepest part cutting across veins of solid red elvan in places. Evidence from the boreholes in the Taw valley also provide an indication as to the nature of the weathered deposits and kaolinised areas* In general the Taw marsh valley boreholes exhibited finer material at the top, with coarser sands and gravels at their bases* Kaolinised biotite-mlca granite was found at depths of 41.14 m which were red in colour and showed evidence of tourmallnisatlon. Clays found were predominantly grey in colour, and typical of kaolin deposits, having high silt contents. (ill) Veins, stringers and intrusions A large mineral vein was encountered during excavation of some of the more solid bedrock at Fernworthy as illustrated in Figure 2:10. This vein varied from 1.82-3.0 m in width inclined at ly from the vertical* Open jointed granite was apparent around this vein so excavation was taken down to below 25 m, after which a wedge was driven into the centre to plug the zone and prevent water movement downwards. Veins of red elvan were recorded at the Sheepstor trench as being cut by a larger china clay vein. Veins traversing the area interected the trench at angles of 45-70*** These veins were found to be practically parallel in nature, trending northwest and southeast, and ranging from 2 cm to 91 cm in thickness, providing adequate pathways to facilitate water movement in this locality*

71 When some veins, 15-30 cm wide, at Burrator, were followed down to their junction with solid rock, it was found that a black line (caused by peat staining), approximately 2*5 cm thick could be observed which continued downwards through the rock, beneath the centre of the vein. These were interpreted as possible pathways for further breakdown of the rock matrix. Sandeman (1901) noted that the decomposed granite along the whole of the trench would have been a serious leakage problem and so was removed. (iv) The junction of weathered and solid rock sections In the cases of Jointed granite seen at the Burrator site, as the excavation depth increased the joints closed up entirely or became filled with material of a 'water-tight nature' (Sandeman 1901). The decomposed granite at Burrator was observed to become harder with depth. In the Fernworthy valley the south face of the granite was found to be generally solid and unbroken, as opposed to the north face which consisted entirely of open jointed and partly decomposed granite as illustrated in Figure 2:10. With depth these joints closed up in a nature similar to that described at the Burrator site. Other indications concerning the weathered and solid granite interface on Dartmoor are less specific. At the Swincombe and Huccaby sites, no solid or fresh granites were recorded in the original drilling logs, giving a slightly less well defined zone between weathered and essentially sound materials in the geological cross- sections (Figures 2:7, 2:8). 2:2:13 Water Movement in Weathered Granite The existence of groundwater in weathered granite areas Is illustrated by surface expressions such as springs and seeps. There is little published evidence concerning water movement through solid or weathered granite material on Dartmoor. With reference to near surface conditions/horizons on valley sides, Williams et al. (1981) distinguished four interflow pathways through the soil and weathered granite regolith. In interflow monitoring pits in the Narrator Brook Catchment, total volumes of flow from above the (i) iron pan, (11) above a fraglpan, (ill) beneath an impeding layer, (iv) flow caused by saturation upwards, (v) and overland flow were recorded. Of the subsurface flows recorded those in order of importance were (ii) above

72 a fraglpan, (1) above the iron pan, (iv) flow caused by saturation upwards, and (Hi) beneath an impeding layer. Evidence for the movement of considerable volumes of water at deeper levels are derived from the accounts of excavations and trial- borehole sinkings at various locations on Dartmoor. The trench excavated at Sheepstor hit water Just below surface level, and had to be pumped continuously to be kept clear. It was observed, by Sandeman (1901), that as the trench was deepened, the water which had previously bubbled up spring-like from the trench bottom, came from the trench sides. Veins traversing the Sheepstor locality and intersecting the trench were inferred to have acted like 'land drains*, since material In their upper portions were deeply stained with peat, carried down by infiltrating surface waters. This staining was observed to have gradually disappeared with depth. The quantity of water flowing into the trench increased with depth until pumps were lifting 2045.70 m^ day^^ to a height of 30.48 m (Sandeman 1901). During excavation at Fernworthy, water was found to be entering the foundations from the surrounding granite at an estimated 133 sec*"^. This flow took place mainly through the fractured granite as there was only a shallow (4.57 m) thickness of weathered granite at this site. It seems likely that such discharges are fed from a larger fissure-system source at this locality. The large central vein found in the Fernworthy valley was regarded as a potential leakage pathway at this site, and was subsequently wedged off to maintain the water- tightness of the valley. Groundwater from weathered granite material is drawn from the Taw Marsh area for public supply purposes. Nine underground pumping stations draw water from wells 15.24-18.28 m deep in the residual sands and gravels in this valley. The scheme provides a useful localised source of some 3785 m^ day-^. In general the major factors governing the water movement in weathered granite aquifers appear to be the thickness and disposition of weathered materials and the nature of the unconsolidated-solid Interface, along with the occurrence of joints, fissures and vein systems. The ability of the material to store and transmit water is a function of the incoherent material's permeability and the occurrence

73 - of clay materials in the succession. The amounts of water available from granite systems and their relative depths, appear to bear no relationship to one another despite locations being nearby as in the case of Burrator and Sheepstor. This emphasises the point that it is unlikely that such aquifers on Dartmoor can be characterised by investigations at one location. Aquifer parameters, because of the heterogeneous nature of weathered materials in situ, are likely to be site specific with respect to location, despite the fact that the valley aquifers have a granite basement in all cases. 2:3 Factors Related to Hydrogeologlcal Responses In Weathered Granite Areas as Typified by the Narrator Brook Catchment As the hydrogeology of an area relates to the total environment, this necessitates a review of hydrometeorologlcal influences and physical features of such an area. The Narrator Brook catchment is very representative of Dartmoor as a whole not only In terms of geology, but also in terms of other factors likely to affect groundwater conditions, namely climate, soil type and vegetation. The Narrator Brook catchment is one of three river basins on south west Dartmoor which supply Burrator Reservoir. The catchment lies 20 km to the north east of Plymouth, Devon, within the Dartmoor National Park (Figure 2:1). The Narrator valley is aligned northwest- southeast with a catchment area of 4.75 km^. The Narrator Brook is nearly 4.5 km in length and joins Burrator Reservoir on its eastern margin. The drainage basin constitutes approximately 30% of the total surface water catchment area feeding the reservoir. The Narrator catchment is therefore a major source of water supply to Burrator, and its importance in the maintenance of an adequate water supply to Plymouth is evident. The catchment area of the Narrator Brook is illustrated in Figure 2:11. The boundary represents the surface water divides between the drainage basins to the north and south of the Narrator Valley. These are easily identified by the topographic highs of Sheepstor, Roughtor, Eylesbarrow, Combshead Tor and Down Tor. The experimental catchment area of the Narrator Brook, as utilised by the Department of Geographical Sciences, Plymouth Polytechnic, terminates at the lowest outflow point, a stream gauging station known as Station Cutt, just before the reservoir. For the

74 FIGURE 2:11^NARRAT0R BROOK EXPERIMENTAL CATCHMENT ILLUSTRATING THE GROUNDWATER OBSERVATION STUDY AREA

DOWNTOlK

COMSSHEAD \ TOR ROUG

ISHEEPSTOR---^

KEY A Tors Catchment Divide /I = = - Tracks == Road Scale 7cm = 1km r_"__ Study Area purposes of this present Investigation the catchment area has been extended to Include the region around Headwelr (Figure 2:11) and the eastern part of the reservoir. This additional area Includes the entire groundwater observation well network. The last 250 m flowlength of the Narrator Brook Is by means of an artificial channel. The purpose of this Is to divert the Narrator Brook through the gauging site at Headwelr, run by South West Water, to monitor Inflows to Burrator Reservoir, 2:3:1 Topography Elevation of the Narrator Brook valley ranges from 220 m at Burrator Dam to over 400 m In the east near Eylesbarrow as illustrated In Figure 2:12. To the south of the groundwater observation-well network, Sheepstor (368.80 m) and Roughtor (304.80 m) provide the highest points In the western part of the valley. At the eastern end of the catchment the valley Is markedly asymmetrical In cross-profile being steeper on the southern side. Generally slopes are linear or convex. The steeper slopes facing N.N.E. are between 10' and 20" with maximum slope angles of 25*'-30** where the granite outcrops at the surface, tors (Sheepstor, Downtor, and Combshead Tor) and their associated blockflelds are evident, and provide local steep slopes. Previous tin mining activities in the valley bottom have resulted in abrupt breaks in the slope profiles, particularly at the eastern end of the catchment. The cross-valley profiles. Figures 2:13 a-c, show the location of the observation wells on the valley slopes of Sheepstor and Roughtor, indicating the steeper valley slopes on the south side of the Narrator Brook valley. 2:3:2 Climate The south west Is characterised by a mild wet climate typical of much of Atlantic Britain. Maritime air masses, especially from Polar source regions are experienced at all seasons and appear over south west England for more than half the year. Distance from the sea and rising altitude give increasingly cool and wet conditions on Dartmoor. Generally much of the rainfall is concentrated In the autumn and winter months, with June being the driest month. During the period of this investigation, November 1978- February 1980, precipitation as recorded at Head Weir was 2747.20 mm. The total precipitation for 1979 was 1896.73 mm. This is high when

76 Downtor

ombshe

Eyiesbarrow I /449'58m 1 \ Sheepstor

Key Catchment ^ Boundary

•^Contour (20m interval) Scale 6cm " 1km

FIGURE 2:12 THE IMARRATOR BROOK CATCHMENT TOPOGRAPHY sw 274-3

2438

228-6

Horizontal scale 1cm=200m

274-3 Sheeps Tor

I 259-0 Sheepstor 243-8-t 10/Beck

228-64-

213-4

Horizontal scale 1cm = 200m SW 2743 259-0

243-8 7 Road

213.4

Horizontal scale 1cm = 230m

FIGURE 2:13a Observation Borehole Elevations and Valley Profiles

7S FIGURE 2:13b

OBSERVATION BOREHOLE ELEVATIONS AND VALLEY

PROFILES

S4 2895

274 3 Small tributary of 259 0 Narrator Narrator Brook , Brook 2438

2286

213 4

Scales Vertical 1cm " 15m Horizontal 1 cm « 200m FIGURE 2:13c Observation Borehole Elevations and Valley Profiles

Scales Vertical 1cm = 15m Horizontal 1cm =200m

S O 2590m R Meavy

213 4 t compared to the long term average rainfall (using 1941-1970 period) calculated for Sheepstor at Redstone Gauge, (0.75 km due west of Sheepstor N.G.R. S.X. 558682) of 1588 mm. The wettest periods were recorded from November 1978 to March 1979, and October 1979 to February 1980, with most of February's average monthly rainfall occurring in the first half of the month. The maximum total daily rainfall of 180.60 mm was recorded at Head Weir on 27th December 1979, during one of the wettest periods experienced during this investigation. Maximum one hour rainfall intensities were recorded in the same period on 24th December 1979, being some 6 mm/hour. The driest period occurred in June, July and September 1979, as can be seen from Table 2:4. August 1979 was an exceptionally wet month. Throughout the investigation period no months were without rain. Snowfall during this investigation period fell during late December 1978 through to late February 1979, with intermittent hail and snowfalls in mid-March and early April 1979. Snowfall in May 1979 accounts for some of the precipitation record for this month. Snowfall and periods of exceptionally heavy frost were experienced in late December 1979 to early January 1980, interspersed with torrential rains and widespread flooding in the south west as well as In the Narrator valley itself. Graphs of the mean daily rainfall in the Narrator Brook Catchment for 1976-1979 are presented in Appendix Lowest mean dally temperatures occurred in November 1979 to January 1980 with -8.8*C on 14th January 1980.being recorded at Head Weir. Maximum daily temperatures occurred in July and August 1979 with 28.20**C on 27th July 1979, recorded. The annual temperature range for this investigation period, obtained from available data from the automatic weather-recording station at Head Weir was 15.0**C. Winds on Dartmoor are generally strong with the prevailing direction west to south westerly. Some severe storms were experienced during the Investigation period, which resulted in large areas of the coniferous stands around the reservoir being uprooted. Wind speeds in December 1979 were recorded in excess of 193 km/hr. Potential evapotranspiration (P.E.) amounts were evaluated by the Thornthwalte equation for the Narrator Brook catchment. Total P.E. for the eighteen months under review in this investigation was 868.6 mm, with 685.8 mm occurring during the twelve months of 1979.

81 Table 2:4 Meteorological Data for the Narrator Brook Catchment, Head Weir, October 1978-February 1980

Rainfall Temperature **C P.E.

Month mm/month Max Min Range mm/month

1978 October 17.72 23.4 -0.2 23.6 36.2 mm November 206.6 15.6 -5.6 21.2 21.0 December 231.35 16.8 -7.2 24.0 16.2 1979 January 139.22 8.0 -7.2 15.2 11.5 February 231.45 10.4 -7.8 18.2 19.5 March 245.55 11.2 -3.2 14.4 34.0 April 98.15 10.4 -4.4 14.8 37.5 May 146.55 25.0 -4.4 29.4 71.2 June 93.2 25.2 3.6 21.6 97.5 July 68.25 28.2 3.2 25.0 135.5 August 189.95 23.4 3.0 20.4 101.5 September 84.36 21.6 1.6 20.0 72.7 October 109.05 18.2 1.2 17.0 60.2 November 167.9 14.2 -4.6 18.8 28.5 December 323.1 13.0 -8.0 21.0 16.2 1980 January 199.6 9.8 -8.8 18.6 11.5 February 149.6 10.4 -1.6 12.0 25.2

2:3:3 Soils The soils of the Narrator Brook catchment are typical of Dartmoor soils in general and consist essentially of: (1) Peaty gley soils/iron pan stagnopodsols (11) Brown earths/brown podsols. The two soil types are Illustrated on the generalised map in Figure 2:14 along with their associated vegetation types. (1) Peaty Gley/Iron Stagnopodsols The soils on the gentler moorland slopes between 300-500 m form part of the Hexworthy series (Harrod e£ £l. 1976). These are coarse loamy iron stagnopodsols. A typical profile, diagrammatic in form. Is Illustrated in Figure 2:15, and discussed below in terms of its horizons, since their nature may Influence the rate of infiltration to the groundwater system.

82 FIGURE 7: \4 GENERALISED DISTRIBUTION OF VEGETATION AND SOIL TYPES IN THE NARRATOR BROOK CATCHMENT Downtor

rr^iililili '^'iiilili r

5 /Eyiesb arrow

Sheepstor

Key Moorland/Stagnopodsols

Coniferous forest/Brownpodsols "=——^ iiili Deciduous forest/Podsol Scale 6cm -Ikm BROWN EARTHS / PODSOLS PEATY GLEY/ IRONPAN STAGNOPODSOLS

Moretonhampstead Series Hexworthy Series cm OT

10-

20-

30- AO- -^'(^Jl B.S. BO •

60 •

70-- B/C B/C SO-

SO--

FIGURE 2:15 DIAGRAMMATIC REPRESENTATION OF THE MAIN SOILS

OF THE NARRATOR BROOK VALLEY Horizon 0-5 cm This consists of a grass root mat and a litter layer.

Oh Horizon 5-15 cm Black (N 2/0) amorphous peat, often containing bleached sand grains with a coarse angular blocky-structure. This horizon is greasy when wet.

A h/E Horizon 15-30 cm Black (10 YR 2/1) humose sandy sllty loam which is slightly stoney with gravel and bleached granite fragments. It possesses a moderate subangular blocky-structure and has a friable consistence.

Eg Horizon 30-35 cm Greyish brown to light brownish grey, or brown pinkish grey (10-7.5 YR 5-6/2), gravelly sandy silt with a few mottles of strong brown (7.5 YR 5/6-8) and commonly heavily humus stained. It is slightly stoney with gravel, and exhibits a subangular blocky-structure with a friable consistency. Dead roots tend to concentrate at the base of this horizon on top of the Iron pan.

Bf Horizon at 35 cm Thin hard continuous iron pan which restricts vertical water movement.

BS Horizon 35-45 cm Strong brown (7.5 YR 5/6) gravelly sandy silt loam with some granitic fragments present. It possesses a very fine subangular blocky structure and a friable consistency.

BS Horizon 45 cm plus Brown (7.5 YR 4-5/4) gravelly sand-silt loam, usually stoney with gravel to large angular granite fragments. This horizon has a moderate-fine subangular blocky structure.

B/C Horizon 60 cms plus Brown to light brown (7.5 YR - 5-6/4) gravelly sandy silt loam which is very stoney with gravel and large subangular granite fragments. It exhibits moderate or strong fine subangular blocky structure with moderate platy structures often apparent. Usually friable but sometimes has a firm consistency.

The Hexworthy Series are usually waterlogged for long periods in the upper part of their profiles due to the impedence of drainage water by the iron pan. The soil is characterised by a peaty surface horizon, and developed in moorland areas experiencing high annual rainfall totals. If the Bf horizon can be breached drainage is free in the lower horizons due to the very permeable nature of the subsoil.

- 85 (11) Brown Earths/Podsols These soils in the Narrator Brook valley are typified by the series found at elevations of 229-340 m. These are usually well drained coarse loamy brown podsollc soils. The profile is illustrated in Figure 2:15 and described below:

Ah Horizon 0-25 cm Dark brown (10 HY 3/3) or very dark greyish brown (10 YR 3/2), gravelly sandy silt loam, usually of a slightly stoney to gravelly nature with moderate fine to medium subangular blocky structure. The horizon has a friable consistency.

Bs Horizon 25-60 cm Brown (7.5 YR 4-5/4) or strong brown (7.5 YR 5/6) gravelly sandy silt loam, often being stoney with coarse gravel and the occasional boulders. It possesses a moderate fine to subangular blocky crumb structure of a friable consistency.

B/C Horizon 60-90 cm Brown to strong brown to yellowish brown in colour (7.5 YR 5/4-6, - 10 YR 5/4), gravelly sandy silt loam. This horizon is stoney with gravel and large subangular granite fragments. Generally it exhibits a subangular blocky structure with a friable consistency. Sometimes thin platy structures occur giving the characteristics of a Cx horizon (fragipan).

The fragipan is an indurated horizon with a platy structure. It is of localised occurrence on Dartmoor and associated with areas of former perlglacial activity. Fitzpatrick (1956) emphasised the importance of fragipan horizons to water movement in soils. When the fragipan is highly Indurated it can completely prevent the vertical percolation of water, so infiltrating waters move laterally. Brown podsolics are described by Harrod et al. (1976) as generally being well drained due to its permeable nature and the characteristics of the substratum of Incoherent granite. 2:3:4 Peat Deposits Upland peat deposits on Dartmoor are found where annual rainfall is in excess if 2,000 mm and altitude ranges from 366 to 590 m (Clayden and Manley 1964). Such deposits may be in excess of 1.2 m thick in some areas but are normally less than 40 cm elsewhere. A discussion of the physical properties of peat is appropriate since they form part of the soil profile in the Narrator Valley. Peats absorb and retain large quantities of water and hence may be important In terms of their effects on infiltration rates, runoff and ground• water recharge in the Narrator Brook Valley.

86 Boelter (1963) suggested that the hydraulic conductivity of peat soil horizons Is one of the more Important physical properties that affect the hydrologic features of a catchment. The structure and density of different kinds of peat vary considerably, Boelter (1964), as do their water storage characteristics. Consequently, water storage characteristics and movement rates In peat will determine the amount and timing of outflow from watersheds made up entirely or partially by organic soils. Conway and Miller's (1960) work on catchments covered with 3.65- 4.26 m of sphagnum peat in northern England, determined water retention properties of peat. Sphagnum peat has a water storage capacity of several centimetres in its surface layers, and this store Is replenished by the retention of precipitation which follows dry periods. Water flow from peat margins was found to continue steadily throughout dry periods. The burning of sphagnum alters the loose- textured surface peat into a substance which is nearly amorphous and highly resistant to water movement through It. Rough grazing lands In the Narrator Brook are subjected to intermittent burning, a practice which may Improve the quality of the pasture but appears to be detrimental with regards to potential infiltration and groundwater recharge facilities. Peak flows from peat covered catchments are delayed and smaller than those catchments with no peat development (Conway and Miller 1960). This suggests that surface runoff over peat areas is greatly reduced. The hydraulic conductivity of peat-types cover a wide range of values which are related to specific yield and pore size distribution (Boelter 1965). Table 2:5 illustrates this. Peat deposits in the Narrator Valley consist mainly of sphagnum species, (Rent and Wathern 1980), and although no work has been done specifically on their physical properties, generalisation of their hydraulic conductivities can be made from reference to work by Burke (1972) and Boelter (1964, 1965).

- 87 Table 2:5 Hydraulic Conductivity of Various Peats after Boelter (1965) and Burke (1972)

Peat types Hydraulic conductivity cm d~^

undecomposed 3292 peat

woody peats and deep 48 undecomposed peats

dense decomposed 0.65 and herbaceous peats

blanket peat 1.8-13.0

2:3:5 Vegetation The Narrator Brook catchment Includes within its boundaries all major vegetation types which characterise Dartmoor as a whole. Broadly speaking the catchment can be divided into a number of vegetatlonal units as Illustrated in Figure 2:14. The upper part of the basin Is open moorland with peaty areas consisting predominantly of sphagnum species, (Kent and Wathern 1980). The middle section is one of rough pasture, grassland heath, bracken and heather mixtures. The lowest part of the area was planted with sitka spruce in the early 1920's to utilise the land rendered 'unproductive* by the establishment of the reservoir and to invest in the timber as a long- term commercial proposition* Prior to this present study, coniferous forest occupied more than 10% of the catchment area, as illustrated in Figure 2:14. This proportion of forested area has since been altered by two main factors: (1) Severe storms in January 1979 and December 1979 which uprooted more than 300 trees. (2) Sllvlcultural activities (felling, logging etc) from March 1979 up until February 1980.

- 88 Both the above factors have altered the vegetation pattern and disrupted soil horizons which, in turn, may have important consequences on groundwater recharge rates. These factors may be difficult to assess in terms of this short period of Investigation, but are points worthy of consideration, 2:3:6 The Nature of the Valley Infill in the Narrator Brook Catchment Apart from investigations carried out on the nearby Burrator and Sheepstor dams (Sandeman 1901), very little information Is available on the nature and depths of incoherent materials overlying solid granite In the Narrator valley* Past mining activities, discussed in section 2:3:9, have yielded no information on the depth and subaerial extent of weathered granite In this locality. The first exploratory boreholes in the Narrator Brook valley were put down by the Geology Department, Plymouth Polytechnic, September 1976, (Gibson 1977), with a Craellus 'minute-man* rig utilising the rotary drilling method. Logs from these sites show that at Site 3, 12.A3 m of fine saturated sands were penetrated and at Site I, 11.12 m of fine sandy material with broken fragments of granite were encountered. Neither of the two boreholes reached bedrock. The watertable was recorded at 0.5 m below the surface. Both these sites were further east than the present investigation area. 2:3:7 Borehole Evidence In November 1978 fifteen boreholes were sunk in the western end of the Narrator Brook catchment, using an air rotary system for the down-the-hole-hammer technique. Thirteen of these were utilised for the data collection network for this investigation. The logs are presented in Appendix I. These boreholes revealed llthologies containing large quantities of clay, sandy clay, sands and grits, with gravels and cobbles. All deposits were of varying thicknesses as can be seen from the logs, and revealed a more complex sub-surface lithology than previously assumed. The terms clay, sandy clay and other terms used are based on visual Assessments of the materials drilled and recorded in the field logs, consequently they are of a qualitative nature only. The shallow sites, 3, 4, 5, 6, as can be seen in the appropriate logs, contained the coarsest materials in the form of gravels and

89 cobbles often greater than 6 cm In diameter* Interstltlces between these larger deposits were filled with clays, silts and water which caused excessive caving, resulting in Site 3 being abandoned. Sites 10, 11, 12, 13, 14, revealed the largest thicknesses of clays and sandy clay material, varying from 8.5 m of yellow sandy clay in Site 11 to 13 m in Site 13. The great variation in the depths and nature of the deposits result in poor correlation between neighbouring boreholes, which can be seen by comparison of the borehole logs in Appendix 1. The great lateral diversity of the deposits is particularly well Illustrated in the case of Site lA. Here two boreholes were attempted, the first drilled to a 12 m depth was later abandoned. This, as can be seen from the borehole log for Site 14, consisted primarily of yellow sandy material down to 6 m, where the clay content increases markedly towards the base. The second borehole, eventually completed for the observation well at this site, was drilled approximately 2 m from the first borehole and penetrated a different sequence of deposits* The variation in thickness and colour of the materials penetrated by the two boreholes show a very poor correlation between llthologles. Sites 9 and 15 contained from 9-12 m of dry sandy material whose clay content appeared to Increase towards the base of the borehole. At Site 15 clay nodules brought to the surface from approximately 7- 8 m depth contained large feldspar crystals in their centres, which showed signs of alteration along their cleavage lines. These nodules were probably formed via the transit of the feldspar crystals up the annulus of the borehole by compressed air. Sites 1 and 2 on the south facing slopes of the catchment exhibited larger thicknesses (14-18 m) of sands and gravels, and probably penetrated odd granite blocks at some depths. Site 2 exhibited approximately a 7 m thickness of coarse sands and gravels at its base. Site 7 also penetrated 6.5 m of yellow/grey sands with a slight orange tinge. Granitic material interpreted as boulders was penetrated at approximately 8 m, directly beneath which was found 2 m of pale grey clay, very typical of a kaolin-type clay. This borehole was terminated in granitic material at 13,5 m which may have been another Isolated granite block.

- 90 - Time and financial constraints, along with problems of drilling through unconsolidated material without casing, prevented further exploratory drilling. The exact thickness and lateral extent of the valley fill in the Narrator Brook is not known definitively at present. The boreholes drilled for this investigation extended at their maximum 20 m below the surface, but no information regarding depth to solid granite was obtained. The nature and lithologlcal variations found in the Narrator Brook catchment correspond markedly to those found in the Taw Marsh area as described in section 2:2. Further investigations into the nature of the shallow or near surface valley deposits in the Narrator Brooks (Gomez and Sims 1981) indicate a crevasse-splay type of deposit near the river banks, characteristic of rivers which overflood their banks periodically. Such an environment may well account for periodic lenticular deposits of clays and sands found in the immediate vicinity of the Narrator Brook, but the occurrence of lithologlcally similar but thicker deposits at the base of Sheepstor may necessitate some other mechanism whose origin is outside the scope of this Investigation. 2:3:8 Particle Size Analysis Extra drilling was attempted at two more sites in the Narrator Brook using a Minute-Man drill. The two sites are referred to as Site 8 and Site 16, whose location is illustrated in Figure 2:17. At Site 8 a total depth of 9.44 m was drilled into a mixture of sands and clays. No core samples were taken due to the caving nature of the hole. This hole collapsed once the drill stem was removed. At Site 16 a total depth of 9 m was drilled with the watertable being hit at approximately 2.7 m. Materials penetrated were described as yellow sandy clays. An attempt to bore such material proved fruitless and only larger particles of weathered granite were found remaining in the core when it was brought to the surface. Drilling was terminated at 9 m. From both sites samples were taken for particle size analysis from the auger flights at depths from 1-3 m below the surface. Particle size analysis curves were produced and are illustrated in Figures 2:16a and 2:16b. It is evident that the valley fill penetrated at these two points contains a larger proportion of sands

91 - 100* / / 90 SiT£ 8 S/75 16 / 80

70- / /

1 60 / / / 50. / / I 40

30

20

iO

-r- * *2 +1 -2 *2 -2 -3 006, 0-2. 0-6,

FIGURE 2 : 16a Particle Size Distribution FIGURE 2:166 (0*06-2 mm diameter) than Is apparent upon visual examination of material In the field giving over 75% at Site 8 and 56% at Site 16. Less than 2% slit and clay fractions was recorded at both sites. At Site 16» 43% gravels (2-60 mm diameter) were apparent In the profile, while at Site 8 only 24% of gravel material was present. The higher topographic position of Site 16 may account for the higher percentages of coarser materials within the first 3 m of the profile, smaller particles having been transported further downslope by perlglacial processes. Particle size analysis therefore indicates areas of permeable materials, possibly with well sorted materials in some sections, with a smaller amount of clay and fines throughout than is apparent from visual examination alone. 2:3:9 Mining Activities No discussion of superficial deposits in valleys on Dartmoor granite is complete without reference to past mining activities in the area. This is particularly true of the Narrator Brook catchment area, and although such evidence refers mainly to the western end of the valley, as Illustrated in Figure 2:18, the area outside the groundwater observation network, some discussion is required for former mining activities have markedly changed the landscape and structure of the ground materials. Commercial china clay, a product mainly of hydrothermal kaolinlsatlon, is found in southern Dartmoor around Lee Moor, but over most of "the noor, kaolinlsatlon was not Intensive enough to produce economic deposits (Reld et al. 1912). Decomposed granite Is common in the areas around tin mines, while the commercial kaolin deposits of Lee Moor, are elongated along the regional trend of the tin mines In this region. Based on the evidence of the mineral vein exposed at Burrator during excavations for the dam, Brunsden (1964) suggested some hydrothermal alteration of the granite has occurred In this valley, in addition to chemical weathering by percolating waters choosing selective pathways and contiributing further to the in situ breakdown of the granites. Tin was first worked from Sheepstor In 1168, Harris (1972), but the main mining concerns In the Narrator Brook took place in the 1800s. Gomez and Sims (pers. comm. 1980) have a Cm date for an upper peat horizon (at 0.55 m depth) in the Narrator valley of 1350 AD.

- 93 - FIGURE 2:17 DISTRIBUTION OF SURFACE WATER AROUND BURRATOR RESERVOIR

Scale 4cm = 1-5 km Sharpitor A Leather Tor ^

Downtor

Combshead -a

1 aerator Sheepsto

Sheepstol>ia*A

R.Meav KEY Intermittent springs * Perennial springs o Wells Dry leat stretches rfc^ Extra drilling sites They believe that the deposition of this organic horizon within the floodplain may be linked to water management activities of the tin workers in the early mining period in the valley. Mining activities also took place to the north of the Narrator valley in the Newlycombe lake valley, around Clasey Wheal (Crazy well pool) and Plymouth Consols (N.G.R. S.X. 386699). The majority of Dartmoor Lodes lie in an E.N.E.-W.S.W. direction, and are tin or copper bearing (Dines 1956). In west Dartmoor the lodes are mainly disposed in an east to west direction with a higher percentage of copper lodes. The spoil-heaps from former tin-mining and streaming activities are still evident near the head of the Narrator Brook valley, as illustrated in Figure 2:l2, and appear to form a major part of the valley infill at various localities. Tin was separated in the valley bottom from the rubble waste which was then thrown up in heaps along the river banks. These grew into sizeable mounds, now in places overgrown but still a persistent feature of the landscape. Since they are coarse, granular poorly-sorted heaps they form valuable infiltration sites, facilitating recharge to the underlying valley alluvium and rotted granites. Little information is obtainable relating to the size and lateral extent and production of the subsurface mines in the Narrator valley. Adits and shafts are still evident (N.G.R. S.X. 578683) but their underground extent is unknown. A few of the adits now act as springs feeding the surface water network of the catchment. The general disposition of such spollheaps and shafts are indicated in Figure 2:18. From the evidence presented in the previous two sections, it can be seen that there is a wide variation in the type of unconsolidated materials in situ in the weathered granite of the Narrator Brook valley. Aquifer properties of such deposits are fully discussed in Chapter 6, but porosity and hydraulic conductivity values derived from laboratory measurements of materials from the Narrator Brook are discussed here in relationship to groundwater and surface water regimes. 2:4 Groundwater and Surfacewater Relationships In the literature, surface and subsurface water regimes are commonly treated as being separate and distinct from one another.

- 95 These are usually hydrologlcally interconnected and they function together to form an integrated dynamic system in which water commonly moves, alternating from one environment to another. The two regimes are dissimilar in many respects, but because of their hydrologic Interdependence it is technically incorrect to treat them separately, except for purposes of classification, when evaluating water resources of an area. This latter approach will be adopted here in the description of groundwater and surfacewater relationships In weathered granite areas, as typified by the conditions in the Narrator Brook catchment. 2:4:1 Surface Water The uniformity of underlying geological structure on Dartmoor suggests that topographic and groundwater divides are near coincident. Dartmoor Is characterised by a complex system of purpose built leats which transfer water across catchment divides as illustrated in Figure 2:17. The Narrator Brook catchment is the only one of the three major sub-catchments of Burrator Reservoir which does not have water transported into or out of the catchment by leats. The distribution of surface water in the Narrator Brook valley is illustrated in Figure 2:18. Numerous artificial channels drain the south side of the Narrator Brook in the vicinity of the groundwater observation network. These artificial channels of approximately 1 km total length, (Murgatroyd 1980) are related to forestry activities in the area and may contribute significantly to main channel flow in the Narrator Brook during intensive rainfall periods. To the south west of Combshead Tor, down towards the marshy valley bottom, (N.G.R. S.X. 585684), water can be traced flowing below the surface, towards the stream channel. Several distinct drainage lines can be observed, flowing under boulders and just below the surface utilising crevices and channels within the matrix of the valley floor. This may be a form of piping, which Jones (1971) suggested Is a feature of moorland soils with steep hydraulic gradients. Such subsurface plpeflow Increases the transmisslvlty of a soil mass and speeds throughflow, resulting in a quicker hydrograph response than might otherwise be expected.

- 96 Scale 6cm = 1km

® Boreholes Downtor 1976 350-52m \37277m * Intermittent springs * Perennial Combshead .:r^ springs Tor 370-94m =:^arsh a eas • Deancombe ^ 1 Tinners -0 - = qPlruins)** heaps F ' Footpaths * ^ Combshead \ I (ruins) Roads old minei^ Catchment workings boundary Yellovjimead 304-8nrK shaft A Tors own Sheeps Tor Spot Heights 3688m

(uoCl&SStf ltd) FIGURE 2: 18 •368-8m THE NARRATOR BROOK CATCHMENT Surface water distribution Extent of mining areas

OS This is a particularly noticeable feature in wet periods when rapid downslope movement of water takes place towards the Narrator Brook from the contributing spoil heaps and surrounding clltter slopes, resulting in a temporary extension of the stream channel and surrounding marshy area, 2:4:2 Groundwater The headwaters of the Narrator Brook rise from perennial springs emerging from the valley sides (N.G»R. S.X. 592692) and from amongst the tinners' spoil heaps (N.G.R. S.X. 589682). The source of the Narrator Brook varies seasonally, and according to Murgatroyd (1980), 0.19 km of the main stream channel is ephemeral in nature. Many tributaries in the catchment are fed by perennial springs and intermittent springs (those which cease to flow during dry periods). Figure 2:18 Illustrates these hydrogeologically significant sources in the Narrator catchment. Sheepstor Beck, with its headwaters coming from the lower reaches of Sheepstor, is fed by three main springs. The upper reaches dry up during the summer period and begin to flow only after the onset of winter rainfall, a characteristic feature of an intermittent stream. Other smaller tributaries feeding the Narrator Brook have their source in marshy areas. The granite based aquifer of the Narrator catchment appears in the first instance to be unconflned. However, alternating sands, gravels and clay lenses shown by drilling, and described in section 2:3:7, may result in a series of stacked, perched water tables, operational at different times of the year and rendering parts of the aquifer to be both confined and unconfined in places. Variations in thicknesses and lateral extent of differing textural horizons gives rise to wide ranges in hydraulic properties of the aquifer. 2:4:3 Laboratory Derived Values of Porosity and Hydraulic Conductivity of Granitic Material in the Narrator Brook Catchment Area Laboratory measurements of slightly altered granite from the Narrator Brook valley were carried out at the Water Research Centre (W.R.C) laboratories, to determine hydraulic conductivity (K). The results are presented in Table 2:6.

- 98 Table 2:6 Laboratory detertalnations for K and 0 In altered granites from the Narrator Brook

Hydraulic Sample 1 Sample 2 conductivity

vertical (K) 1.5 X 10-3 ^ ^1-1 8.1 X 10-^ m d-^ horizontal (K) 3.2 X 10-3 ^ 5.0 X 10-2 ^ jj-l porosity (0) 0.068 0.101

These values were measured using a gas permeameter. The hydraulic conductivities are much higher than expected for slightly altered granite samples, Rowland (pers. comm. Jan. 1980), and can be attributed to: (1) leakage around the edge of the samples during measurement (11) In the weathered portions the clay particles may contract upon drying out giving higher porosities than expected. The Y. values of these granite samples are still lower than those of the sands, gravel and clay materials of the water-bearing horizons In the Narrator valley, as derived from the Slug-Test data (Chapter 6). A series of porosity measurements were carried out on a variety of differing grades of weathered granite by Fookes et al. (1971), particularly in the vicinity of Burrator and at other locations on Dartmoor. These determinations are present in Table 2:7. These Illustrate the range of likely porosities to be expected In weathered granite areas on Dartmoor.

99 Table 2:7 Dartmoor Granite Porosity Measurements after Fookes et al. (1971)

Granite Location Porosity %

Merrlvale Quarry Not determined (n.d.) Burrator Quarry n.d. n Roadside n.d. ti Quarry 1.65 Roadside 5.5 •• Quarry 8.85 Roughtor Quarry 3.9 •* Quarry 8.7 n Quarry 1.56 Burrator Roadside n.d.

2:4:4 Gross Hydraulic Conductivity Determinations The variety of materials in the Narrator Brook aquifer with different hydraulic conductivities indicates the hydrogeologlcal complexity of the region, and makes resource assessment difficult. An attempt was made to derive an overall average gross catchment hydraulic conductivity value as follows: Using the data available for the water year 1978/79 an estimate for (K) was determined using Darcy's Law given by Q = KIA catchment area A = 4.75 x 10^ m^ effective infiltration per year = 1.093 m year"^ average hydraulic gradient =1 =0.03

volume of infiltration per year = 1.093 x 4.75 x 10^ m^year"^

K = 1.15 X 10-^ m sec"^

For a refinement of the gross catchment K-value above:

- 100 - Gross catchment hydraulic conductivity K*^

Number of wet days in the year K X Days in the year

The number of wet days in the year refers to those when the soil moisture deficit (S.M.D.) is zero, and therefore it is assumed that any water available will infiltrate to the groundwater. In the water year 1978-1979 117 days were recorded when S.M.D. = 0, so the K* value derived for the Narrator Valley is 3.6 x 10"^ m sec"^. It is emphasised that this calculation is merely an attempt to quantify the gross hydraulic characteristics of the Narrator Brook catchment as though it were transmitting groundwater as diffuse flow. The real physical situation Is most unlikely to be one of homogeneous diffuse flow because of the joint and fissure systems in the more solid bedrock in the higher reaches of the catchment. 2:4:5 Groundwater Transit Time An estimate for the actual transit time for groundwater movement can be derived using the relationships

V, V = -Ji after Todd (1959) a

where V ° actual velocity of flow a V = darcy velocity of flow d n = effective porosity and = Ki d Using the total annual daily flow of 78.62 m^sec"^ for 1978-1979, and Q = KIA, then Ki =» 1.7 x 10"^ msec"^. Porosity measurements of weathered but coherent granite sections In the Narrator Brook catchment by W.R.C, (section 2:4:3) give values of 6.8% and 10,1% using n = 0.068 then the actual velocity of flow is estimated at 2,5 X lO^'* m sec"^, and choosing a 2 km flowpath, le, from Combshead Tor to Station Cutt, transit time for water from one end of the catchment is estimated at 0*254 years (3 months). This is only a tentative suggestion since several physical features In the catchment may profoundly alter the estimate. The porosity values, as determined by the W.R.C,, are likely to be higher than those expected for partially weathered granite, due mainly to

- 101 contraction of weathered clay minerals, and leakage around the sample. On the other hand areas of gravelly deposits will have much higher porosities which will reduce flow velocities, while the granite bedrock will have negligible values. Water infiltrating into the surface layers of the catchment may not necessarily flow directly to the water table, and contribute to groundwater flow. The impedence of the infiltration of precipitation waters in the catchment by the iron- pan is a factor in point. Certain areas in the groundwater observation network appear to be underlain by clay layers that support perched water bodies. The direction and speed of flow, if any, within these perched levels is difficult to Judge, and may not be related to the normal down valley flow within the main groundwater zone. The fact that clay layers are overlain by coarser sands and gravels, may result in quite high localised porosities that will reduce flow velocities. 2:4:6 Watertable Fluctuations The watertable fluctuations have been monitored for the groundwater sub-catchment from November 1978 to February 1980. The watertable is near the surface by the Narrator Brook and its lower tributaries but varies in depth below the surface elsewhere. The topographic location of the observation wells in the Narrator Valley are illustrated in the cross-valley profiles in Figures 2:13a- c. Gustafsson (1966) suggested that if groundwater surface levels are plotted against topographic locations of individual wells on a graph, and the cluster of points falls on or closely below a line drawn through the origin, then groundwater levels follow the ground surface very well. Figure 2:19 illustrates this procedure for wells and water levels in the Narrator Valley. It can be seen from this portrayal that there is a close association with the position of the watertable and the topographic height of the wells in this region. Generally the watertable is high in the valley bottom and lower or indeed periodically absent as one progresses up the valley sides. This feature is related to the relative porosity and permeabilities of the different zones in the weathered profile, and to the nature of the materials penetrated by the observation wells.

102 FIGURE 2:19 GRAPH TO ILLUSTRATE THE RELATIONSHIP BETWEEN TOPOGRAPHY AND

WATER LEVEL IN THE 250 n NARRATOR VALLEY O O (A 240

E c 230H o

(0 o a 220H a> n (0 •J ^0) 210H

200 200 210 220 230 240 250 260 Ground surface AOD metres 2:5 Conclusions It is evident from material presented in this chapter, collated from site investigations and dam excavations, that the occurrence, distribution and depth of weathered granite and its potential for water storage is not insignificant, on granite areas la south west England. Evidence for considerable volumes of water from excavations and trial borehole sinking accounts, in the weathered and fractured granites in the south west testify to this. In cases like the Fernworthy site, where the depth of weathered granite is shallow, but large quantities of groundwater were apparent, movement of groundwater Is facilitated via fissure and joint systems in the granite. Like many drainage basins on Dartmoor, the granite of the Narrator Brook is overlain by a variable thickness of weathered and Incoherent granitic materials. During excavation of the trench of Burrator Dam, solid granite was reached at depths of between half a metre and 12 m (Table 2:3) with wedge-shaped fissures extending down to 33 m into the granite (Sandeman 1901). The decomposed granite was recorded as becoming more resistant with depth and merging into hard rock, in some cases abruptly. Within the Narrator Brook valley the unconsolidated material is known to be greater than 30 m at one point in the valley floor,, but no information is available concerning a more detailed picture of weathering depths in the catchment. Depths of weathered granite in the south west have been collated. Table 2:3, and exhibit varying thicknesses of less than half a metre to greater than 34 m. This weathered zone (growan) frequently contains large bounders of undecomposed granite. Within the top few metres of the growan in the south west, layering is present. Such bedded deposits have been attributed to the action of downslope transportation processes, possibly accentuated under perlglaclal conditions (Waters 1964). Drilling logs from site Investigations on Dartmoor show great similarity in that their lithologies show such complex variations that neighbouring boreholes cannot be correlated. Such was found to be the case in the Narrator Brook catchment. Here thicknesses and the disposition of materials of differing permeabilities within the weathered granite matrix vary widely. A mixed assemblage of clays, sands, grits, cobbles and the occasional granite boulder were found to

- 104 - be the main constituents of the weathered granite In the Narrator valley. Particle size analysis suggests that the clay content of the valley Infill may not form the larger proportions implied from Initial visual assessment of borehole drilling returns. Gravels form 24-43% of the weathered material, sands 56-76% with silt and clays being less than 2%. It must be emphasised however that these particle size analyses were only carried out to depths of 3 m below the surface, and such quantitative studies at depths greater than this may prove contrary conditions. A range of porosity values measured at Burrator, in the vicinity of the Narrator Brook valley, and at other sites on Dartmoor have been presented in this chapter. The porosity and hydraulic conductivity ranges reviewed highlight the hydrogeological complexity of weathered granite regions, and the uncertainty Involved in groundwater resource assessment in such areas. Porosity values range from 1.65-11.56% for weathered granites with a gross hydraulic conductivity value of 3.6 X 10"^ m sec"^ being presented for the Narrator Brook aquifer. Past mining activities may enhance the subsurface flow pathways in weathered and fractured granites by utilising the old adits and drainage channels. There is little detailed documented evidence of the extent of these activities of parts of Dartmoor, and In particular the Narrator Brook Catchment. As a consequence of this. In association with the variety of materials of differing hydraulic properties, estimates of velocities and throughput or transit time of 3 months for the Narrator Brook Catchment must be viewed with caution. The disposition and thickness of weathered materials on the granites in south west England are likely to be unique for each location. As such, sites will exhibit hydrogeological complexity which is specific to the location, making potential resource assessment difficult.

105 - Chapter 3 The Data Collection Network In the Narrator Brook; Field and Laboratory Techniques

3:1 Introduction As outlined In Chapter 2 studies were undertaken In the Narrator Brook Catchment, with a view to elucidating the hydrogeological characteristics of weathered granite areas. The experimental network used in the Narrator Catchment was designed to provide data for the following purposes: (a) to determine annual changes in the groundwater regime in weathered granite regions (b) to determine the effect and magnitude of the interrelationship of precipitation, evapotranspiration, spring discharges, river stage and reservoir levels with water level fluctuations (c) to indicate the directions of groundwater flow (d) to determine aquifer hydraulic characteristics (e) to delineate areas of recharge and discharge in the catchment. Information on groundwater conditions were obtained from measurements from three main sources: (1) water level fluctuations In wells in the observation network (11) spring discharges (ill) surface water discharges. 3:1:1 Water-Level Data Water-level data provide records of short-term changes and long- term trends of fluctuation of storage within a specific aquifer. The amount of information derived is dependent upon the length of the observation period, and the standard of data collected. A water-year, running from October of one year to September of the following year is the standard observation period utilised by the water industry in the UK. Such a time-period covers marked seasonal recharge and discharge of water to aquifers, and Is useful for water resource evaluation in this country. The 16 month period of observation utilised for this

106 - investigation 'bridges* the standard water years October 1978 to September 1979, and October 1979 to September 1980. Water level fluctuation data was collected on a routine basis from a number of wells penetrating the aquifer. Wells generally serve as devices for extracting groundwater from aquifers, (Walton 1970). A well Is a hydraulic structure which, when properly designed and constructed for specific investigations, permits the economic withdrawal and measurement of water from the water-bearing formation. How adequately it will accomplish this purpose depends on the skill in drilling, and well construction that Insures taking best advantages of the geologic conditions, and the careful selection of casing materials, (Johnson 1975). In theory, a well intersects the uppermost surface of the zone of saturation, either the watertable in the unconfined case, or the piezometric surface in the confined condition. It can be fully or partially penetrating in nature, and may be dug, angered or drilled; cased or uncased; and may vary in diameter and depth (Cruse 1979). 3:1:2 Well Design Considerations The type of well adopted must depend on the purpose of the Investigation, the depth to groundwater, geologic conditions, and economic considerations (Todd 1959). Wells installed for the purposes of studying groundwater conditions in an area are termed observation wells or piezometers (Ward 1976). An observation borehole, as defined by the Research Panel, Institution of Water Engineers (1969), is a boring, generally of small diameter which is drilled into underground strata to penetrate the aquifer either partially or fully. It thereby provides a means of measuring the physical and chemical properties of groundwater and the physical characteristics of the aquifer (Walton 1970). The type of observation well to be used for water-level fluctuation data, must be considered In relation to the validity of results to be obtained from them. Two problems arise: firstly, to what degree is the water-level in the wells a true reflection of the surrounding watertable; and secondly, to what extent Is the validity of the water-levels affected by the diameter of the well and other characteristics of the casing arrangements?

- 107 As a result of comparisons of direct (open wells, piezometers, tensiometers) and indirect (neutron scattering, gamma densitometry, seismic and resistivity) methods of measuring water-table elevations, Myers and Van Bavel (1963) suggested that open wells give a true measurement of the water-table elevation. This holds providing there is no appreciable vertical flow component and equilibrium obtains in the soil water system. Lag due to storage volume of a large open well may cause appreciable errors in soils of low permeability. During an investigation of shallow groundwater in sandy soils of the Coachella valley, California, Richards et al. (1973) noted that the water-level in a well is the same as the water-level in the soil. If the water in the soil is static, or if the depth of water in the well is shallow. In this case the water-table position will be reasonably correctly determined in a well. Richards ^ ^* (1973) mention however that if the water in the soil is moving vertically, if the soil is stratified, or if the water in the well is deep, there may be a considerable difference between the true watertable and the water level in the well. Since a well is permeable it may act as a short-circuit between layers of different material. Water storage in the well may also cause water-level changes in the well to lag behind water-table changes in the soil or aquifer material. In the evaluation of shallow groundwater movement in a Boulder Clay catchment, Bonell (1978) utilised a nework of perforated polythene tubes of 5.08 cm and 6.35 cm diameter, to monitor groundwater level fluctuations in Holderness, Yorkshire. These provided 'satisfactory data' of water-level movements for his investigation. Ward (1962) came to the conclusion that the level of the watertable could be accurately determined in an observation well, but the accuracy of results will increase as the diameter of the well decreases. The underlying principle is that the wider the well is, the more it departs in size from the interstices in which groundwater is naturally held, and the more likely it is to improve artificial flow conditions which result in a lowering of the watertable at the observation point. In the analysis of a variety of pump tests, Kruseman and De Ridder (1970) found that rapid and accurate measurements of water-

108 levels can be obtained by small diameter piezometers. If the diameter was large the volume of water contained in the piezometer may cause a time-lag in changes of drawdown. Similar principles can be applied to piezometers in a non-pumping aquifer. In general it is undesirable to use production wells for water-table observations as their large diameters and pumping effects may distort the fluctuation in, or the shape of, the groundwater body, resulting in erroneous interpretations of the areal conditions (UNESCO 1975). Observation wells are commonly of small diameter primarily because the drilling of a small-diameter well is cheaper (Monkhouse and Phillips 1978). They must however be of large enough diameter to permit a measurement of the water level without difficulty, and to use a water sampler If changes in groundwater chemical composition are to be studied. The wells also need to be large enough to allow the periodic removal of accumulated sediment and any rubbish Introduced at the well head. Considering these factors, UNESCO (1975) recommends observation wells should be constructed with inner diameters ranging from 5-15 cm. The number of observation wells necessary to study the groundwater regime, and their areal distribution is dependant on the hydrogeological features of the area to be Investigated and the type of study planned (Dlxey 1950). The number of observation wells utilised vary in the literature, as can be seen from Table 3:1 which illustrates documented well densities. In practice the well densities tend to be a function of finance, aquifer material, and the type of investigation being undertaken. 3:2 The Observation Network in the Narrator Catchment In addition to the previously discussed factors concerning the choice of observation wells, the following specific considerations were taken into account in the setting up of the network in the Narrator Brook catchment: 1. Good areal distribution 2. Accessibility 3. Hydrogeological variability.

109 Table 3:1 Well Densities

Author Type of Aquifer Location { Well Density

Rasmussen & Sands, silts and Beaverdam 0.5 wells/km2 Andreasen gravels over• Creek, U.S.A. 1959 lying crystalline rocks

Schlct and Panther Creek 0.0859 Walton •* Hadley Creek 0.156 1961 Goose Creek 0.0742 Illinois, U.S.A.

De Ridder and Silts and clays Estuary Polder, 3.81 Wit overlying well S.W. Netherlands 1965 sorted coarse- medium sands, underlain by Ollgocene clay

Smith Millstone Grit, Yorkshire, U.K. 0.0082 1966 Magnesium lime• stone and Bunter sst.

Bonell Glacial clay Holderness 2.27 1971 and sands Yorkshire, U.K.

Desmedt, Peat and Belgium 28 Van der Beken alluvium and Demarre 1977

Uhl Sapolite, basalt Satpura Hills 0.0163 1979 sandstone and Central India granite

Granneman and River valley Missouri 0.312 Sharp infill, sands, River Valley 1979 gravels and U.S.A. alluvial

This study Weathered granite Narrator Brook 2.77 Head deposits, U.K. alluvium over• lying granite

1. The data collection network for groundwater consists of thirteen wells, astride the principal directions of surface water flow in this part of the catchment. The wells are located on both sides

110 of the Narrator Brook and Sheepstor Beck, and near the reservoir. The requirement of a good areal distribution was the main consideration in the choice of well positions, but financial constraints restricted the number of wells installed. The thirteen observation wells are identified by site numbers 1, 2, 3, 4, 6, 7, 9, 10, 11, 12, 13, 14 and 15. Well 5 was abandoned after severe caving during drilling and Well 8 was not attempted due to problems with the drill's compression unit. 2. The location of individual sites were chosen to facilitate drilling operations in terms of accessibility, and to be in close association with the principal surface drainage lines. This latter feature was included to enable an assessment of the degree of hydraulic continuity between the stream and the aquifer. The wells were located to take into the network a variety of hydrogeological conditions with respect to altitude, slope and position relative to the reservoir and surface.drainage. Table 3:2 summarises the topographical positions of the wells.

Table 3:2 Topographical Position of Wells

Position in Catchment Well Numbers

High-Middle Slope 1, 2, 14, 13 Base of Slope 9, 12, 15 Valley Floor 10, 11, 7 Reservoir Level 6. 7 Banks, Streams and Springs 3, 4, 11

In the first instance, in the planning of the network, electrical resistivity and geoseismic methods were utilised in the hope of positioning the wells in more representative sites, with regard to the hydrogeological conditions present in the catchment. The results of these geophysical surveys were generally inconclusive. This was thought to be due to peaty and unconsolidated subsurface conditions.

Ill - Thus the resultant network of observation wells was chosen mainly with regard to topographic considerations. 3:2:1 Installation of Observation Wells Observation wells may be drilled by several different methods using drill rigs of various designs, the most common being the rotary and the cable-tool percussion types. According to Cruse (1979) there is, however, no water-well drilling technique applicable to all conditions found in'the field. It Is therefore important to use the system appropriate to site conditions (Stow 1962). Primarily because of economic considerations, an air rotary system for drilling was utilised for this investigation. It was anticipated that massive granite (devoid of Joint systems), fractured granite (traversed by joints and cracks), or weathered granite (from rocks exhibiting slightly altered minerals to incoherent material), would be met soon after penetration of superficial materials and soils. This type of drilling is known as the down-the-hole-hammer. The technique was originally designed in Sweden and developed in the U.S.A. (Johnson 1975). It combines the percussion effect of cable- tool drilling and the rotary action of rotary drilling. Essentially a pneumatic hammer is operated at the lower end of a drill-pipe, and compressed air Is transported down the drill pipe and back up between the pipe and the borehole wall, thus removing the chlppings. The drilling bit consists of an alloy steel hammer with heavy tungsten-carbide inserts which chip and crush the rock. It is a particularly suitable method In drilling for groundwater in crystalline rocks, or other materials with high compressive strength, (Beyer 1966). The rate of penetration in several types of rock is well documented as being faster than other methods and tool types ' (Cruse 1979). Campbell and Lehr (1973) cited that down-the-hole- hammer is most efficient in consolidated rock formations which do not require casing. They suggested that the technique is not usually satisfactory for use on boulders or unconsolidated formations, although it does achieve some success in these materials. As a result of a survey dealing with water-well design, drilling specifications and well development in a variety of Igneous and sedimentary rocks in the U.S.A., Ahrens (1970) noted that in the case of specialised drilling techniques when the optimum operating

112 conditions are not met the drilling success rate falls off markedly. In the light of experience of the drilling operations in the Narrator Brook the down-the-hole-hammer method proved not entirely suitable for the ground conditions encountered. The hammer did not operate efficiently in the unconsolidated material, which also contained large quantities of clay. Such conditions were experienced at Sites 5, 6, 11, 13 and 14 as Illustrated by the drilling logs in Appendix 1. Observations of water being blown out of the hole with cuttings during drilling, indicated the position of the watertable. Such water issuing from the ground, was formed by the action of the compressed air rising up the annulus between the pipe and borehole wall, and was experienced at all sites, except Site 14 (where the watertable was not penetrated) during the drilling operations in the catchment. The borehole logging carried out in the catchment only gives an indication of the disposition, depth and likely nature of the materials in the Narrator Brook. The borehole logs for all drilled sites are in Appendix 1 and are described in detail in Chapter 2. Samples of drilled material were taken at 3 m intervals, corresponding to the length of individual drill strings. It was considered that particle size analysis of these samples would not be useful, since materials from varying depths had been mixed up during transit to the surface via the compressed air column. 3:2:2 Well Dimensions Economic considerations tend to be the deciding factor in the selection of drilling methods, depths of penetration,. diameter of the borehole, and casing characteristics. In this investigation a maximum depth of 20 m was decided upon, but if bedrock was reached beforehand then that depth would be the finished depth of the borehole. This depth was determined mainly from the evidence of weathered depths from other valleys on Dartmoor, as outlined in Chapter 2. In the Narrator Brook, the base of the valley aquifer is assumed to be coincident with the solid granite basement. As mentioned in Chapter 2, water movement through the granite mass is likely to only take place via localised joint systems, whose subsurface presence or disposition are unknown in the Narrator Catchment. Consequently the granite boundary is taken to be an impermeable layer. If this is the case, drilling further would not be likely to penetrate a water-bearing horizon.

113 The diameter of the holes drilled was 37.5 mm (Figure 3:1). A 32 mm casing was Installed and the 5.5 mm annular ring was backfilled as outlined in Section 3:2:7. Table 3:3 gives the depths of the observation holes. Some wells were drilled to greater depths than their cased dimensions, but caving took place after withdrawal of the drill rods, before the casing could be installed, resulting in a

Table 3:3 Observation Borehole Depths Sites 5 and 8 were abandoned

Site Drilled Depth Casing Depth Comments

1 19.5 19.25 - 2 19.0 18.79 - 3 3 2.29 caving 4 3 1.30 caving 6 7 5.63 caving 7 13.50 13.50 - 9 15.0 14.72 - 10 12.0 9.67 - 11 12.0 4.80 badly caving 12 12.0 9.39 - 13 16.0 10.33 - 14 15.0 12.76 water table not registered 15 15.0 14.45 —

shallower observation well. This occurred at Sites 4, 5, 6, 11, 12 and 13. 3:2:3 Casing Materials These consisted of 3 m lengths of white polypropylene tubes with an internal diameter (I.D.) of 32 mm. They were joined together with straight couplings of the same material, glued Into position, and lowered down the drilled holes.

114 FIGURE 3:1 OBSERVATION WELLS:-DIMENSIONS

PLAN VIEW

Scale: Actual size

37-5mm

ELEVATION

Rubber bung

Observation well

i 1 Cylindrical outer cover

32 mm Backfill material to support the Ground observation well .O^"^ . leveI _il _ CsS and outer cover

^^S^^i^'^ —Formation stabilizer

115- Previously constructed I m lengths of well screen were attached at the lower ends of the casing by another straight coupling. A rubber bung, dimensions: 35 mm diameter at its top; 30 ram diameter at its base, were utilised to seal off the lower end of the screen. Another bung of the same dimensions was used as a stopper at the surface end (well head) to allow both access for measurements of the watertable, and to prevent direct accretion of rainwater to the groundwater. 3:2:4 Well Screens A well screen serves as the intake section of a well that obtains water from an unconsolidated aquifer. The well screen allows water to flow freely into the well from the saturated horizons, prevents material entering with water, and serves as a structural retainer to support the borehole in an unconsolidated aquifer (Todd 1959). Desirable features of a properly designed well screen are reviewed by Johnson (1975), Campbell and Lehr (1973) and Ahrens (1970). These include: (1) Openings should be in the form of slots which are continuous and uninterrupted around the circumference of the screen, (11) A close spacing of the slot opening should be adopted to provide the maximum percentage of open area, which should be consistent with an adequate strength of the screen, (ill) The screen should be strong enough to resist the forces to which it may be subjected during and after installation, (iv) Screens should be built from suitable materials to avoid corrosion. In the place of commercial well screens 'make-shift' substitutes are often employed. Slotted casing can be made by sawing slots ranging from 0.025-0.635 cm in width, giving a maximum open area of 12% for the larger slots (U.S. Department of the Interior, 1977). Perforated casing can be made by punching holes into the screen section, but these are more difficult to develop than continuous slot screen (U.S. Dept. Int., 1977). This is because irregular edges Impede water movement and cause excessive head loss. The desirable open area in a well should be at least equal to the porosity of the aquifer material (Walton 1970).

- 116 Whatever design is chosen the slot size should be such that the entrance velocity of water should not normally exceed 0.03 m sec"^ , (Jeffcoate 1977). This ensures laminar flow of water at the well interface during pumping, (Walton 1970). Higher velocities increase the liability to blockage due to chemical scale. If the percentage open area is smaller than aquifer porosity, constriction of flow occurs as the water enters the well. In a production well this means more drawdown because additional head loss occurs in the movement of water through the screen openings (Campbell and Lehr 1973). A well screen is particularly susceptible to corrosion attack and incrustation by mineral deposits. The many perforations expose more surface area to a reactive environment in the water-bearing strata. Plastic is practically immune to corrosion attack, but plastic screens commonly have relatively low percentage of open areas (U.S. Dept. of Interior 1977). Non-reinforced plastics are subject to creep under sustained load with resultant changes in slot size- The collapse resistance of plastic screens in unconsolidated material, particularly in wells deeper than 45 m is questionable, unless wall thicknesses are properly sized to resist stresses (Johnson 1975). Increasing wall thicknesses increases the stress-resistant nature of the casing but increases the cost (Campbell and Lehr 1973). In conclusion, plastic casing has high corrosion resistance but their use is limited to shallow small diameter wells of low capacity. A well screen is adequate only when it is capable of letting sediment- free water flow into a well in ample quantity and with the minimum of head loss. All screens are subject to some extent to plugging through silting by mechanical processes, which lowers their water-transmitting capacity. Screen silting processes appear less significant in polyethylene pipes (UNESCO 1975). In the observation wells of the type used in this investigation, factors of corrosion, strength, costs, slot size and the percentage of open area of the screen, were taken into consideration. The overall design of the Narrator Brook observation wells satisfied the previously discussed design considerations. It should be pointed out that since these wells were not constructed for water supply purposes, well entrance velocities were not ascertained. It was decided that a

117 - well screen was an integral part of the network design, particularly as it was intended that pump tests were to be conducted on selected sites. With these aims in mind, and given the financial and time constraints, an appropriate well screen design was chosen, 3:2:5 Slot Size Determination Samples of a variety of granitic materials were taken. Soils along with gravel mixtures, sands and clays, were collected from localities within the catchment upstream of the Narrator Brook bridge, (N.G.R. S.X. 569690). Materials from hillslope exposures, banks, pits and ditches, along with river bed samples were taken. From this variety of catchment material the d50 grain size, likely to be representative of the water-bearing formations was obtained. Particle size analysis placed the majority of these samples in the fine sand- gravel grade (0.06 mm-2.0 mm dia.) (Wentworth 1922) with a median value of approximately 1.8 mm. Stow (1962) suggested that slots in the screen should be sufficiently narrow to retain 40-60% of the aquifer material. Jeffcoate (1977) recommended that for deep boreholes the slot size is usually selected to retain about 40% of the sample, but with shallow formations a much wider slot may be justified, since rapid development of the whole length of screen can be achieved without excessive collapse of the surrounding material. For the materials thought to be representative of subsurface conditions in the Narrator Catchment the appropriate size of the slot opening was determined as 2.55 mm. However, due to the rigidity and low strength of the casing material, a 2 mm diameter (d60) size was selected. The slots were 2 cm in length around the circumference of the screen, with 1 cm between the end of one slot and the beginning of another and a 1 cm width between offset rows In the screen, as illustrated in Figure 3:2. This offset arrangement was used because of the lower strength of the polypropylene sections when perforated. 3:2:6 Screen Length, Position and Effective Open Area Screen length selection is often a compromise between factors of cost and lithological conditions (Walton 1970). The screen length is based in part on the effective open area of the screen and on the optimum screen entrance velocity. When a screen is placed in an

118 aquifer sediment will settle around it and partially block the slot openings. The amount depends on the shape and type of slot, and on the shape, size and sorting of the surrounding water-bearing materials. On average half of the open area of screen will be blocked by aquifer material, therefore the effective open area is usually 50% of the actual open area (Monkhouse and Phillips 1978). The open area for the Narrator Catchment well screens was 9.52% as compared to 12% as recommended by U.S. Department of the Interior (1977). This gives an effective open area of less than 5%. If these were production wells they would be very Inefficient (Monkhouse 1975). The optimum length of well screen should be chosen with relation to the thickness of the aquifer, available drawdown and stratification of the aquifer. An example cited by Johnson (1975) gives a screen length equal to about one third of the saturated thickness of a homogeneous watertable aquifer. Walton (1970) suggests that as long a screen as possible should be used to reduce entrance velocities and the effects of partial penetration of the aquifer by a well. He also suggests that under watertable conditions, optimum production, well specific capacity and yield are obtained by screening the lower one third to one half of the aquifer. Prior to drilling in the Narrator Valley relatively little specific information was available concerning subsurface llthologies and the disposition of the water-bearing horizons. Consequently the relative positioning of the well screen in the most permeable section could not be predetermined. According to UNESCO (1975) the screen for an observation well Is commonly Installed at a depth that will ensure its remaining below the lowest anticipated water level. Generally screen lengths for such observation wells need not exceed 2 m, a length that ensures a good response to the well water level fluctuations. Prior to drilling, the assumption was made that lower down in the aquifer Intergranular flow predominates, so the lower portions of the boreholes were likely to provide the most water. In accordance with this supposition, one metre lengths of slotted-screens were utilised. In accordance with recommendations from the U.S. Dept. Interior Bureau of Reclamation (1977) that screens set in unconsolidated materials should have a bottom seal or plug, rubber bungs were placed into the lower ends of the screens at all sites (Figure 3:2). Such a

119 FIGURE 3:2 WELL SCREEN AND CASING ARRANGEMENT

(not to scale)

Rubber Bung V Ground Surface

Annulus backfilled with drilled material and formation stabilizer around well screen

Coupling glued into position

Polypropylene casing

•1 metre length of slotted well screen

V X Rubber Bung seal precludes heaving of materials up Into the well and provides a bearing area for support of the screen assembly* The rest of the borehole was lined with unperforated casing, 3:2:7 Formation Stabiliser A sand and gravel mixture, the formation stabiliser, was utilised to fill the annular space around the well screen and the lower part of the casing* The principal functions of a formation stabiliser as suggested by Ahrens (1970) and Agerstand (1966) are outlined below: (1) to stabilise the aquifer and minimise sediment Ingress (2) to permit the use of the largest possible slot size with a resultant maximum effective open area (3) to provide an annular zone of high permeability thereby increasing the effective radius of the well (4) to support the screen against unbalanced forces which might arise during development of the well. Formation stabiliser Is a less uniform mixture of grain sizes than a conventional gravel pack, whose grain size is carefully selected to match the aquifer material, Johnson (1977) suggested a mixture of materials about the same or slightly coarser than the water-bearing formation as being the most suitable for the formation stabiliser. When the well is pumped this mixture will then retain all the aquifer material that would otherwise enter the well. The U.S. Dept. Interior Bureau of Reclamation (1977) recommended that as long as the smaller grains are larger than the scree slot size and the largest are 0.95 cm in diameter or less, the formation stabiliser will achieve its purpose. The formation stabiliser utilised in this investigation were stream deposits from the Narrator Brook's river bed and banks, comprising sands and fine-medium gravels, 0* 2 mm 6.0 mm in diameter. 3:2:8 Installation of Casing Components The borehole casing, with its screen at the base, was centred in the drill-hole and then the annulus was backfilled to a depth of approximately 4 m above the top of the screen, to allow for settlement. The rest of the annulus above this level was backfilled to the surface with the previously drilled materials.

121 The extent of the lining projecting above the ground level was measured at the time of installation, after which a small mound of material was arranged around it, forming a base for the outer sleeving. Breather holes were cut In the lining above the ground to permit normal groundwater fluctuations. A rubber bung was used as a well stopper, and a length of black P.V.C. tubing was slipped over the installation to protect it. 3:2:9 Well Development Development work, according to Monkhouse and Phillips (197A), is an essential operation in the proper completion of a water well. Although development of wells is appropriate for production wells only, in the case of the Narrator Valley where pump tests were to be carried out in the observation wells, it was considered to be a necessity. Drilling operations can cause some plugging of openings and compact previously loosely consolidated materials. Developing such wells will eliminate the 'skin-effect' of mud caking the borehole walls and dislodge the unconsolidated material around the screen to recover lost porosity, and hence enhance water flow at the well/aquifer Interface. An effective means of development is to surge the water in a well. Water is forced Into and out of the aquifer by the operation of a plunger up and down in a cylinder. This repeated motion leaves the aquifer materials graded radially from the screen centre with an average grain size getting smaller further into the aquifer (Cullen 1968). Should the aquifer contain streaks of clay in the vicinity of the screen, then the surging action may well cause clay to plaster over the screen surface and thereby reduce well yield. According to Cullen (1968), care must be taken not to develop the well too violently or excessive grading of the surrounding formation may cause subsidence of materials above the well screen. This will effectively block the previously graded zone. Should a screen become wholly or partially plugged by clay of mud, the surging action may produce high differential pressures which can cause the screen to collapse*

122 - The tool normally used Is a surge plunger or block. This is one of the most effective devices for development (Walton 1970), A hand operated surge block was constructed suitable for use la these small diameter wells, (Figure 3:3). The surge block consists of a solid stainless-steel cylinder through which four holes have been drilled. A system of rubber '0' rings» housed In notches around the circumference, aided the snug fit inside the well casing. A thin rubber skirt was attached to the top of the block which acted as a one-way valve on the upstroke. Individual 2-meter lengths of steel rods could be screwed together and attached by a thread housed centrally on top of the surge block as illustrated in Figure 3:3. Observation wells 13, 12, 10, 7 and 2 were chosen to be developed after installation. Wells 3, 4, 6 and 11 were located in badly caving materials, and because of their shallow penetration depth, the well screen was too near the surface to enable satisfactory use of the surge-block. Such development may have reduced their usefulness as wells for water level observations, since the risk of damaging the screens was high. Wells 9, 14 and 15 lost their screen plugs during Installation into caving sands and clays, and development of these sites was not undertaken since it may have induced further 'heaving up* of material into the wells. The wells were developed in the following manner: The surge block was lowered 1-2 m below the watertable, but always above the top of the screen. The water column transmits the action of the plunger to the screen section. An upward movement produces an inward flow of water and fine materials into the well, and a downward movement reverses the flow and causes a surging effect, disrupting the bridging of grains across the screen slots (Cullen 1968). This motion was continued smoothly for 3-4 minutes. The surge block was then removed and a bailer lowered down to ball out the water and any particulate material. The surging and bailing was carried out three times, or until the water being removed was clear. It was thought inadvisable to carry development on for longer periods as the casing may not have stood the strain. A maximum of 15 minutes on each well was used although total time for development may range from about 2 hours on small production wells to 20 days on large wells with long screens (Monkhouse and Phillips 1978). The bailers

123 - FIGURE 3-3 SURGE BLOCK-HAND-OPERATED MODEL (Not to scale)

Screwthread for rod attachment Ring to hold down rubber skirt Thin rubber skirt which acts as one-way valve on up-draught

Rubber 'O^ rings 30mm diameter

Solid stainless steel cylinder with four valve-holes drilled through

Water enters valve-holes on down-draught

Dimensions:- 20cm long 28mm diameter

Design am ended after "Groundwater and wells Johnson Division U.S.A. 1975 used for the well development are Illustrated in Figure 3:4. The effects that such well development may have had on the hydrologlcal response of these wells is discussed in Chapter 5. Three months after development all wells were tested with a plumb-weight and tape to estimate depths of sediment accumulation and infilling of the screens. Table 3:4. As expected, those without screen plugs, wells 9, 14, and particularly well 13, showed maximum depths of infilling of 0.62 m, 0.81 m and 6.88 m respectively. The casing in well 13 at a depth of approximately 6 m below the surface appeared to have fractured, resulting in the large amount of infilling, some 2.49 m at this site. Sites 3, 4, 6 and U also showed infilling depths of 0.62 m, 0.27 m, 1.16 m and 1.32 m respectively. This was to be expected, since the water-bearing horizons at these localities proved unstable during drilling. Sites 1, 2 and 7 showed little evidence of ingress of aquifer materials.

Table 3:4 Depths of Infilling of Observation Wells

Well Depth of Infilling

1 0.0 2 0.11 3 0.62 4 0.27 6 1.16 7 0.0 9 0.62 10 0.71 11 1.32 12 0.69 13 2.49 14 0.81 15 6.88

125 FIGURE 3i4 GROUNDWATER THERMOMETER (Diagrams not to scale) Tape Measure WELL BAILERS

Thermometer in brass casing Nylon cord

I String handles Hollow stainless steel tube

Plastic 10ml cylinder PLAN VIEW cord handles Top of brass tubing

close fitting ring holds cord handles in place 3:3 Water Levels and Discharge Measurement Water-table fluctuations were recorded weekly on a routine basis between 0900-1300 hours, and on other occasions when fieldwork was undertaken. Levels of the water-table below the ground were recorded with a purpose-built water level recorder whose main components are Illustrated in Figure 3:5. Two of these recorders were built, and recordings were accurate to ±1 cm. The altitude of the observation wells, measured to the top of the casing were 'levelled-in'. The heights are above O.D. (Newljm) and are illustrated in Table 3:5.

Table 3:5 'Levelled-in' Altitudes of the Observation Boreholes

Well Topographic Elevation

1 239.51 m 2 230.31 3 223.09 4 223.13 6 220.24 7 228.33 9 237.62 10 234.62 11 230.86 12 242.44 13 248.11 14 248.09 15 240.89

3:3:1 Spring Network The location and magnitude of springs give a good indication of the general hydrologic conditions in a region. Abundant small springs on valley sides and the hillslopes Indicate a shallow watertable with a shallow circulation of subsurface waters in aquifers of poor permeability (Davis and De Wiest 1966). The appearance of such springs may also indicate areas of multilayered soils supplied by discrete drainage systems. In such a case saturated conditions may build up from the bases of several soil layers giving rise to saturated lateral drainage (Reeve and Kirkham 1951).

127 - FIGURE 3:5 WATER LEVEL RECORDER (not to scale) A twin electrode battery operated device (T.E.B.O.D)

wooden cable spool housing the battery and speaker The cable is wound aroundl the spool for storage

spool handle

14m length of two conductor electric cable permanently marked in metres and centimetres

10cm

two electrodes housed in a brass tube In contrast large springs confined to valley bottoms indicate high permeability and greater depths to the watertable. The presence of springs may also indicate the position of the regional watertable and give some indication of the depth of well which may be necessary (Toth 1971). On the valley slopes, where the horizontal permeability of the saturated soil is greater than the vertical permeability, a series of 'perched* soil watertables may persist under suitable antecedent moisture conditions. These will normally be located or 'perched' above the regional watertable in an area. The natural discharge is some indication of the maximum quantity of recoverable water in an area, and Indirect point measurements of groundwater can be made by measuring spring flow (Gray 1969). This method of determining groundwater discharge is limited by the distribution of springs, and there may be a considerable time lag between watertable change and spring flow response, especially If the recharge and discharge points are some distance apart. Since one of the aims of this investigation is to delineate areas of recharge and discharge, some measurements on selected springs was felt to be appropriate. Two springs were selected from within the groundwater observation network area: (1) Spring (12/13); located between wells 12 and 13 (li) Spring (11); adjacent to well 11. Both springs are centrally placed in respect of the well observation network and are easily accessible for sampling. Both springs are of a permanent nature in contrast to the large number of intermittent springs which issue periodically within the catchment. Spring (12/13) possesses a 45** *V' notch weir and discharge was recorded on a weekly basis (Chapter 5). It also contributes significantly all year round to Sheepstor Beck's flow via an old mine adit (N.G.R. S.X. 569 686). Spring (11) Issues strongly from the ground, is perennial in nature and is a significant contributor to a tributary of the Narrator Brook. As the intention was to study the relationship between ground and surface waters, 'thermometric observations', (Schneider 1962), as outlined in section 3:5, were extended to the surface waters In the catchment. Water samples and temperature recordings of the two springs were taken weekly as outlined in section 3:4.

- 129 3:3:2 Surface Water Measurements Stream discharge is the sum of surface runoff and groundwater flow that reaches the river. Such discharge is a function of precipitation duration, intensity and distribution; permeability of the ground surface; area of the drainage basin; type of vegetation; stream channel geometry; depth to the watertable and the slope of the land surface (Davis and De Weist 1966). Within the framework of the h^rological cycle, the components of the surface and groundwater regimes are interlinked through the unsaturated zone. Hence it is Important that the relative effects of any interaction of components are determined for a hydrogeological investigation. With this in mind, stream discharge was monitored in the Narrator Brook catchment.

The data from the main discharge measuring stations. Station Cubt (St. Cutt) and Station^^(St. 11) (whose locations are Illustrated in

Figure 1.:^., Chapter I) were used for the analysis of stream flow characteristics. Station 11 is situated at the junction of old farmland and the Sitka Spruce plantation, and monitors stage with an Ott Horizontal Type X chart recorder. At Station Cutt, the downstream end of the spruce plantation, an Ott Strip Chart Type XX, monitors stream stage. During previous stream discharge determinations on the Narrator Brook, (Sims pers. comm. 1978) it was observed that discharge decreased in a downstream direction* Visual assessment of this stretch of the Narrator Brook, which falls within the observation network area, proved a number of springs and tributaries which supplement the runoff in the main channel. The number of contributing sources increase towards the reservoir end of the Narrator Brook, being substantially increased in times of heavy precipitation via artificially constructed drainage channels on the south side of the river, as Illustrated in Figure 3:6. Taking into consideration the above factors, one would expect discharge to increase in a downstream direction, unless the stream was influent in some sectors. In order to verify this point, a further three discharge measuring sites in addition to St. Cutt and St. 11 were set up at the locations illustrated in Figure 3:6. A discussion of the stream flow component and its relationship to the groundwater regime Is presented in Chapter 4.

- 130 - 1—^ / FIGURE 3:6 ' SURFACE WATER DISTRIBUTION

ST. CUT

®

Drainage ditches V'notch weir Spring sampling sites Stream gauging \ sections r^-^ Permanent marshy areas -I-Z-l: Intermittent <^Scale 22cm-^1 km marshy areas * Perennial spring positions ++ Position of inter• mittent springs 3:3:3 Supplementary Discharge Data Daily figures of reservoir water levels were obtained from the S.W.W. warden at Burrator Reservoir. A correlation of these figures with groundwater levels in the wells nearest to the reservoir was attempted to explain some unusual features of their water level fluctuations. The analyses of such fluctuations are presented in Chapter 5. 3:4 Geochemical Sampling A study of the differences and changes In the chemical content of groundwater may be useful in determining the source of recharge, direction of flow and presence of boundaries in a groundwater system (Todd 1959, Ward 1975). An increase in a specific ionic component in the groundwater at a given location may suggest a recharge area, while a decrease in an ionic component with time may suggest absorption by the aquifer materials or discharge from the system. A time lag in the relative amounts of a specific element recorded- at different sites may indicate the residence time of water in the aquifer. Eriksson and Khunakasem (1968), on their work on the chemistry of groundwater in Swedish eskers, came to the conclusion that the chemical composition of newly formed groundwater Is initially determined by the chemical composition of precipitation. As the eight major ions calcium, magnesium, sodium, potassium, bicarbonate, chloride, nitrate, and sulphate, constitute around 98% or more of the total solutes of most groundwaters, the determination of these provides the basic data on groundwater geochemistry (Cook eX al. 1979). Partial analysis to determine the concentrations of some principal chemical constituents in a water may provide sufficient data for many investigations (UNESCO 1975). The chemical composition of natural waters in rivers, lakes and wells depends upon many interrelated factors including geology, soils, climate, topography, biological processes and time (Gorham 1961). Rodda et al. (1976) and Lahermo (1970) consider rock type to be a fundamental or even a dominant factor controlling groundwater chemistry. In the Narrator Valley, some of the major cations and anions present in the groundwaters were used as 'tracers* to attempt to determine sources and pathways of water movement. This was achieved

132 by chemical monitoring of both the inputs (rainfall) and groundwater in the system. An attempt was also made to correlate changes in water chemistry with periods of precipitation, with a view to estimating possible residence times of water in the aquifer* The choice of the elements for analysis for this investigation were based upon considerations of the geochemistry of the Dartmoor granite, and data from past analysis of water from the Narrator Brook catchment (Williams 1978, pers. comm.). Information regarding the geochemistry of the Dartmoor granite has been derived from work by Brammall and Harwood (1932) and Al-Saleh et al. (1977), as outlined in Chapter 2. Table 2:1 in that chapter summarises the geochemical data. Ions which form minerals with relatively high solubilities and which are least likely to be removed in water-rock reactions, are referred to as having high geochemical mobility. Sodium and chloride are considered as potentially useful natural tracers in view of their relatively high geochemical mobility (Edmunds et al. 1976). Gorham (1961), on a geochemical survey of Finnish bedrocks came to the conclusion that most igneous rocks are very deficient in carbonates, sulphates and chloride, which were found in large amounts in the natural waters of Finland, suggesting that these constituents had sources elsewhere. Cleaves et al. (1970) suggested that chloride is the only ion likely to be contributed entirely by precipitation because it is not present in sufficient quantities In the bedrock of an igneous catchment. Chloride, potassium, sodium and silica were the elements chosen for investigation in the Narrator Brook catchment. The silica content of natural water is less variable than any of the other major dissolved constituents (Davis 1964) and consequently maintains a constant background level in groundwater. Silica represents the by-products of weathering of the silicate minerals present in the granites and concentrations in groundwater may Indicate rates of weathering and residence time of water In the aquifer. Recent investigations on granite weathering on Dartmoor, (Ternan and Williams 1979), used silica concentrations in stemflow, throughflow, litter-runoff, interflow and sprlngflow waters from the Narrator Valley to suggest the importance of contemporary weathering processes in operation. It seemed relevant, in regards to water

133 movement and weathering rates, to determine the range of silica concentrations in the groundwater and to see If any seasonal variations were evident, and to attempt to interpret these in terms of groundwater movement and potential recharge. Both sodium and potassium are present in the Dartmoor granite (Table 2:1, Chapter 2) and are by-products of weathering. Additional sources of these elements seems to be from a maritime origin (Stevenson 1968). Like chloride they appear to be suitable natural tracers in an attempt to locate sources of water and periods of recharge, in the Narrator Brook catchment. Past chemical determinations on water samples from the Narrator Brook, (Williams pers. comm. 1978) showed only trace amounts of nitrate, sulphate and bicarbonate. Consequently these anions were considered to have too low a concentration to be useful as natural tracers, while chloride, sodium, potassium^and silica appeared to be appropriate for this type of Investigation. 3:4:1 Field Techniques; The Collection of Water Samples Water samples from the wells, springs, rain gauges, stemflow and the Narrator Brook were collected weekly and analysed for potassium, sodium, chloride and silica. pH and electrical conductivity were also recorded. All bottles were first rinsed in some of the collected sample from each site, before the sample to be analysed was taken. (1) Groundwater Samples Water samples from the wells were extracted using a purpose built water sampler consisting of a hollow brass tube with a ball-valve in its lower end, as illustrated in Figure 3:7. This was lowered 2-6 m beneath the water's surface and upon uplift sealed the water sample in until at the surface it was transferred to a chemically inert polyethylene bottle. The water sampler was a dual-purpose piece of equipment designed for use in the slug-tests as discussed in Chapter 6. The water sampler holds 50 ml. The first sample at each site was discarded, then approximately 150 mis were taken for analysis.

134 FIGURE 3:7 WATER SAMPLER (Not to scale)

Nylon cord

Hollow brass tube

Rim to prevent ball-valve from drifting upwards Perspex ball-valve

Points of entry of water into sampler

Dimensions 30cm long 20mm diameter 50ml volume

I3S- (11) Surface Water Samples Spring and river samples were taken by submersing a bottle below the water's surface, thereby allowing the water to flow directly into the container* (ill) Other Water Samples As a tracer-technique was used in conjunction with later chemical analysis of groundwater in an attempt to define recharge and discharge points in the catchment, then some idea of the initial chemical condition of the inputs was required. With this in view, samples of rainwater and stemflow water were taken. The relative volumes of each were recorded in the first Instance. Any insects, leaves and other debris present in the rain gauge were skimmed off the surface before the samples were taken. 3:4:2 Laboratory Techniques; Storage of Samples Within two hours of collection the samples were taken to the laboratory where pH, electrical conductivity, sodium and potassium values were determined, as described below. After these Initial determinations were made, the samples, following recommendations of Golterman (1971) and Madtereth (1978), were stored in the dark to A inhibit any biological activity, at a temperature of S^C, to await analysis for silica and chloride. 3:4:3 Determination of the Properties of Water Samples Certain properties of water, especially its pH, are so closely related to the environment of the water that they are likely to be altered by sampling and storage procedures. The pH of a water represents the Interrelated result of a number of chemical equilibria which is altered when water Is brought to the surface, sampled and stored. The pH value of most groundwaters is controlled by the amount of dissolved carbon dioxide gas and the dissolved carbonates and blcarbonates (Freeze and Cherry 1981). Since pH Is liable to be modified by biological activity or by CO^ exchange with the air, long Intervals exceeding a few hours between collections and measurement should be avoided (MacJCereth et_ al. 1978). 3:4:4 pH Measurements A pH meter, model E.I.L. 7050 was used throughout this research period to Investigate pH values in all water samples collected in the Narrator Brook catchment. The meter and glass-electrode system were

- 136 standardised before sample measurement using commercially available pH powders dissolved in the requisite amount of distilled water. pH values of pH4 and pH7 were used being those suggested by the manufacturers and recommended by Cook et al. (1979). The sample pH determinations were recorded at the actual temperature of the sample at the time of measurement in the laboratory. The electrode was agitated in the sample and the reading at which the meter stabilised was taken as the representative pH value. 3:4:5 Specific Electrical Conductivity The specific conductance (conductivity) of a water Is a measure of its ability to convey an electric current, and varies positively with the concentrations of ions present and the temperature at which the measurement Is made* The relationship is to some extent dependant on the nature of the major ions in solution so that waters of different ionic composition will display a different relationship between ionic concentration and conductivity (Macteereth et al. 1978). The conductivity of most fresh waters is low and is reported as micromhos cm"^ or in SI units microsiemen, |iS, at 25*'C. Single distilled water in the laboratory generally has a conductivity of 1-5 ^S. Carbon dioxide from the air dissolves Into the distilled water and the resulting bicarbonate and hydrogen ions Impart most of the observed conductivity. For most groundwaters the conductivity value multiplied by a factor of 0.55-0.75 will give a good estimate for total dissolved solids (T.D.S.) which include all solid material in solution, whether ionised or not. Estimating T.D.S. by measuring conductivity is convenient because it can be determined quickly (Johnson 1975). Hem (1970) argued that specific conductance determinations when applied to natural waters cannot be expected to be simply related to ion concentrations or T.D.S. but agreed they are useful measurements In a practical way, but only as a general indication of T.D.S. concentrations. Electrical conductivity was measured using an Electronic Swltchgear battery operated conductivity meter. Readings are standardised on this equipment at 25**C.

- 137 3:4:6 Sodium and Potassium Determinations An E.E.L. Flame Photometer Mark II was used to determine levels of sodium and potassium In the water samples collected. The instrument was calibrated using known standards. Initially a check program was instigated in which samples were analysed over varying time periods after collection: (I) One half of one month's samples were filtered and both sets were analysed. The results showed little significant difference so further filtration of samples prior to analysis was deemed unnecessary. In addition there is a suggestion from the geochemical literature that filtration can interfere with resultant element values, particularly in the case of chloride (Golterman and Clymo 1971). (11) Sodium and potassium levels were prone to deterioration after two to four weeks of storage, so after the first month's samples, immediate analysis was decided upon for the rest of the research period. 3:4:7 Silica and Chloride Determinations The analyses for chloride and silicon were carried out on a fortnightly basis using an Auto-Analyser, Technicon Mark II. Silicon values were then converted to silica values. This instrument is designed to investigate these elements in the range 2-100 ppm although concentrations of less than 1.0 ppm can be determined but with less accuracy. The instrument was calibrated using standard solutions. To check the consistency of standards and any decomposition of samples due to storage effects a check procedure was initially carried out: (1) The first four months' samples were analysed weekly to determine any variations due to age and storage. No significant differences were obtained. (II) The first month's samples were acidified with HNO3 (nitric acid) in proportions 1 ml HNO3 : 100 ml water. This gave some dilution error in the results but these were found to be very small (Williams pers. comm. 1978), This was a check to see if any deterioration of samples occurred, if so, values for silica and chloride would be different, however

138 no significant change was apparent between acidified and non-acidlfled samples so acidification was discontinued. It seemed likely that since the water samples were fairly acid In the first Instance, pH 3-5, they tend to retain their original composition for longer periods* (111) Each week's results for chloride and silica represent an average figure for the number of Independent analyses carried out per sample. Differences ascertained between runs on Individual samples showed less than 1% variation for silica and chloride values. 3:5 Supplementary Measurements These Include recordings of water temperature, rainfall and stemflow measurements, and soli moisture data. 3:5:1 Temperature Temperature Is a parameter offering a means for observing changes in the state of the groundwater regime. In time and space, which relate to conditions of recharge, (UNESCO 1975). Generally groundwater temperatures correspond very closely to the yearly average air temperature, and those groundwaters nearer the surface reflect air temperature changes more rapidly (Walton 1970). Schneider (1962) used temperature fluctuations as a means to study rates and directions of groundwater movement In a glacial outwash aquifer In Minnesota, U.S.A. to evaluate recharge conditions. He concluded that with other hydrogeologlc and cllmatologlc data, groundwater and surface water thermographs may be useful In determining the following: Induced filtration from a body of surface water; to estimate the order of magnitude of travel time from the source of recharge; and the detection of direct Infiltration of rainfall to groundwater bodies. De Geer (1966) suggested that as part of a hydrogeologlc Inventory, It Is useful to determine the characteristic, groundwater temperature of an area. With such information it is usually easy to distinguish deep-seated springs from other groundwater horizons. Razmann (1947) utilised subsurface temperature anomalies to detect infiltration from surface water sources into a gravel aquifer, and Parsons (1970) investigated the factors influencing the groundwater

139 thermal regime and temperature distribution In a glacial complex, with an emphasis on the effect of groundwater flow. (1) Water Temperature Measurements Temperatures of groundwater were recorded weekly. Temperature In this Instance refers to that In the first two metres of water In the well below the watertable. A mercury filled thermometer with a range -10**C to 20*C, graduated In 0.1*0, In a metal housing case, was used for all water temperature measurements. A small cap. In the form of a 10 ml plastic measuring cylinder with string handles was attached to the bulb-end of the thermometer as Illustrated In Figure 3:4. The whole unit was lowered down the well by being attached to the end of a 30 m tape. This cap allowed the thermometer bulb to be kept Immersed, during withdrawal, in water from the original measuring depth In the well, thereby minimising temperature changes between the time of withdrawal and time of reading. Down the profile temperatures were taken at the end of each month to try to elucidate the stratlgraphlc changes of temperature down the length of the observation well. The aim here was to attempt to Identify sources of water and to locate areas of direct Infiltration of warmer precipitation derived water. Analysis of this data Is presented In Chapter 7. As recommended by UNESCO (1975) temperature measurements were collected through the full range of the wells. Suggested Intervals between readings are recommended as being no more than 2.5 m and In the case of the Narrator Valley wells a 2 m Interval was chosen. The temperature readings were taken from the water level downwards In order to reduce mixing and hence any disturbance of temperature stratification present. The thermometer was lowered to the required depth and left for a three-minute interval for equilibration of the thermometer with the surrounding groundwater. As the Intention was to study the relationship between ground and surface waters, the thermometrlc observations were extended to the surface waters In the catchment. Spring and river temperatures were also recorded weekly by placing the thermometer In the flowing water at specific sites.

140 3:5:2 Rainfall, Stemflow and Throughfall Measurements l^ainfall, stemflow and throughfall were to be correlated with observed water level fluctuations in an attempt to define the watertable response to recharge events. These measurements were also used to assess the chemical inputs of water reaching the ground, and hence infiltration to groundwater storage. (I) Rainfall An automatic weather recording station is situated near Site 6, as illustrated in Figure 3:9. It records at 5 minute intervals, rainfall, temperature, wind run and direction, net radiation and solar radiation. Rainfall and temperature observations have been extracted for this research period, and as other measurements were Inaccurate, evapotransplration was determined by the Thornthwaite method. The Thornthwaite method of calculating potential evapotransplration (P.E.) was developed from rainfall and runoff data for several drainage basins, (Thornthwaite 1948). The result is basically an empirical relationship between P.E. and mean air temperature. The empirical formulae can be used for any location at which daily maximum and minimum temperatures are recorded. It is this simple universal applicability rather than any claim to outstanding accuracy which has led to the widespread use of this method (Palmer and Havens 1958). For similar reasons the Thornthwaite method has been utilised for meteorological data collected in the Narrator Brook catchment, in order to estimate P.E. The monthly values were derived using graphical techniques as outlined by Palmer and Havens (1958). Four autographic raingauges are located in the Narrator Brook catchment and records from one of these, next to Site 6, have been used. This ralngauge (Head Weir) is the closest of the four to the groundwater observation network, and is the more appropriate one in respect of altitude (220.24 m). In addition check raingauges were Installed at Sites 9, 7 and 11. These utilised a plastic bottle with a 12.5 cm diameter funnel, housed in a black P.V.C. cylinder. (II) Throughfall Those raingauges in the forest Sites 9 and 7, monitored throughfall, i.e. rainfall reaching the ground after passing through the canopy layer. Each week the amount collected was measured in a

141 graduated beaker, and a sample was taken for chemical analysis. This rainfall volume was later converted to mm of rain over the catchment. (Ill) Stemflow: A stemflow collar was set up adjacent to Site H. For this device 2.4 m of 3 cm diameter bicycle tyre tubing, cut in half lengthways, was attached to the circumference of the tree by the use of nails. A 5-litre plastic container caught the stemflow via a piped section. The contents of this was measured weekly In a graduated beaker, and a sample taken for subsequent chemical analysis. 3:5:3 Soil Moisture Measurements The soil water zone, located In the uppermost layer of the earth's crust, constitutes a key zone in which incoming precipitation water is dlrectioned and partitioned in various ways. Accordingly, the nature of this zone will determine the extent of storage, runoff, downward percolation to the groundwater and the degree of upward return to the atmosphere through evapotransplratlon. All water held in the soil between the soil surface and the groundwater level is soil water, (Kristensen 1975). Since the distance between the soil surface and the groundwater table may be considerable in many locations, and thus not all the soil water Is influenced by the plant root activity, a subdivision of the soil water regime is desirable. According to traditional concepts (Melnzer 1923), the unsaturated zone of an unconfined aquifer is divided into three zones, the soil zone, the intermediate zone and the capillary fringe. The soil zone consists of the upper metre or two of the profile which is influenced by soil forming processes. It contains the roots of the vegetation and is Influenced largely by rapid moisture changes due to abstraction, rainfall and evaporation. The Intermediate zone lies beneath the soil zone and overlies the capillary fringe. It can vary in thickness from zero to many tens of metres, depending on the depth of the watertable. Most water movement is downwards, although some lateral movement or interflow may take place. The capillary fringe is the zone extending upwards from the watertable for a distance which might vary from a few centimetres to many metres depending on the pore size distribution of the material.

142 Bell and Wellings (1980) considered that the above described three-zone concept, while not invalid, tends to obscure the fact that the unsaturated zone is a continuum from the watertable to the soil surface, and that the system is a dynamic one in which change in any part influences subsequent events throughout the whole profile. Within the three zones the water content is divided into various categories which attempt to define a threshold water content which separates water than can drain (and hence contribute to recharge) from water that Is held in the smaller pores (conceived as being unable to drain). In the soil zone the threshold water content is called 'field capacity' being the greatest amount of water that the soil will hold under conditions of free drainage, (le against gravity). Excess water 'gravitational water' drains Into the intermediate zone, and the remaining water is divided into 'capillary water' available to plants) and 'hydroscopic water' (retained about individual soil particles by molecular attraction). In the intermediate zone the threshold is termed 'specific retention'. The retained water content is 'pellicular water' which cannot drain. Excess water is again 'gravitational water' which does drain. In the capillary zone the water is described as 'capillary water'. The zone is saturated and therefore (when static) is unchanging In water content, although (implicitly) acting as transmission path for drainage from above to the watertable below. The in^ial moisture content of a particular soil will influence the proportion of precipitation which infiltrates into the soil, and Its ability to serve as a temporary storage reservoir for downward percolating water (Pitty 1973). Painter (1971) on work associated with a hydrologlcal classification of soils in England and Wales, notes that after periods of drought infiltration of rain first permits any soil moisture deficit to be reduced, and downward percolation to replenish the groundwater only starts when the field capacity is exceeded. The same point was noted by Remson et al. (1959) whilst studying the relationship of the zone of aeration with climatic factors and groundwater recharge at Seabrook, New Jersey, U.S.A. Soil moisture measurements were conducted in the Narrator Valley, as part of the groundwater Investigation with the following alms:

- 143 - 1. To determine whether direct increments to the water levels are due to Infiltration through the unsaturated zone, and soil layers near some of the observation wells. 2. To determine the speed of moisture movement and hence recharge by correlation with water level changes in the observation wells. 3. To assess whether groundwater recharge takes place only after total soil saturation, by comparing soil moisture contents with increments in groundwater levels at chosen sites. Soil moisture was determined by the gravimetric method and by tensiometer studies as described below. 3:5:4 Gravimetric Soil Moisture Measurements Sampling of the soil moisture profiles were undertaken at Site 1 and Site 9 for the following reasons: (1) Site 1 - taken to be representative of the upslope position of the wells in the catchment. (11) Site 9 - considered as being representative of the base of slope position of wells in the catchment. Samples of soil were taken weekly using a one-meter soil auger. Samples were taken at 0.1 m intervals to a depth of one metre, and the material was transferred Immediately into pre-welghed screw-top sample tins. These were taken back to the laboratory, weighed, dried in an oven at 105'*C for 24 hours, reweighed and the soli moisture content expressed as a precentage of dry weight. J:5:5 Tensloraeter Studies Richards (1965) proposed that the term 'soil suction' should be used to specify the property of soil water measured with tenslometers, since 'suction* is a term used extensively to characterise the 'action' of water retention by soils. A porous ceramic cup is positioned in the soil where Information regarding soil water is desired. The cup and the sensing element of a vacuum indicator are all filled with water. Film water in the soil near the cup is in hydraulic contact with bulk water inside the cup through pores in the cup wall. Flow, in or out through the cup wall, tends to bring the cup water into hydraulic equilibrium with the soil water. As soil water is depleted by root action and percolation, or

- 144 - replenished by rainfall, corresponding changes in readings on the tensiometer gauge occur. Tensiometer readings plotted as a function of time provide a useful record of soil water conditions in the neighbourhood of the cup (Rose 1966). Tensiometers can be used to measure hydraulic head in the soil and can measure positive (+) and negative (-) pressures. They also have a shorter response time than wells or piezometers (Richard e£ al. 1973). 3:5:6 Tensiometer Installation In an attempt to determine moisture movement In the surface layers in more detail, seven tensiometers were installed In October 1979. The sites for installation, 1, 7 and 10, were chosen taking into account the following considerations: (1) Site 1 is representative of the upslope position of wells in the network. One tensiometer was Installed here (60 cm), (ii) Site 7 is a forested site not affected by tree felling operations. Two tensiometers were installed here (60 cm, 30 cm). (ill) Site 10 is centrally positioned in relation to the other wells. A nest of four tensiometers was installed here (105 cm, 60 cm, 45 cm, 30 cm), (iv) There was little evidence of Interference with previously installed equipment at these three sites. Access for the soil tensiometers was achieved by the recommended procedure of driving a hollow steel tube, 1.25 cm in diameter, into the ground to the required depth with a mallet. Upon removal this produced a hole into which the tensiometer fitted tightly. The tensiometer was then pushed down until the bottom of the dial-gauge was 5-7 cm above the ground surface, as Illustrated in Figure 3:8. The soil around the tube was then tamped at the surface to seal around the body tube and prevent surface water from running down the sides. In each case care was taken to ensure that the ceramic sensing tip was in contact with the soil in order for the tensiometer to function properly. As an extra precaution packing straw was placed around the tube and dial above ground level and an outer black sleeving placed over

145 - —Service Cap

Vacuum Dial Gauge

•Ground Surface Soil tamped at surface to seal around body tube Plastic Body Tube Tensiometer Bore filled with water and an algal inhibitor

Ceramic Sensing Tip

FIGURE 3:8 TENSIOMETER INSTALLATION PLAN VIEWS OF TENSIOMETER SITES

Site? N

observation Site 10 a well

observation 3^cm )6 0 cm well 60cm 42cnr—' 30cm _ — O N Site 1 150 \ W depths of N tensiometers O-^-50c^'^5cnr|belo ^ surface 45cm 60cm o observation well* FIGURE 3:9 TOTAL INSTRUMENTATION NARRATOR BROOK

WESTERN SECTOR 1978-1980 N

ST. CUT T

229-

236—

®

@

KEY ® observation bore holes check gauge \ T Tensiometers ,335-^ ^ [S Automatic weather \ recording station \ • V'notch weir Scale 19cm =1km SF Stem flow site \ Stream gauging section \ o soil moisture sites I (gravimetric) to serve as protection from frosts. The teasiometers were read weekly and the results are discussed in Chapter 5. 3:6 Summary In an attempt to assess Che potential of groundwater resources in a weathered granite aquifer, a data collection network was developed in the Narrator Brook experimental catchment, based around previous instrumentation. Figure 3:9 Illustrates the total instrumentation used in the western sector of the Narrator Brook catchment between 1978-1980. Table 3:6 summarises the type of data collected in the Narrator Brook catchment and the frequency of observation. The analysis of the data collected in the Narrator Valley jjan be found in Chapters A-7, which also include a discussion of pumping tests and slug tests carried out in the catchment.

Table 3:6 Summary of Data Collected in the Narrator Brook Catchment

Observation Site Type of Frequency of Period of Observation Observation Observation Well network 1,2,3, Water-levels, Weekly Nov 1978- Feb 1980 4,6,7, temperature, 9,10,11 12,13 groundwater Weekly Jan 1979- Feb 1980 13,15 samples Spring and Sp. 11 temperature, Weekly Jan 1979-Feb 1980 Eliver Sites Sp. 12/13 groundwater Narrator samples Brook Stream St. Cutt Stream Continuous Oct 1978-Feb 1980 Discharge St. 11 stage record measurement •* Sites Current Weekly Oct 1978-Feb 1980 C, D, metering and E Rainfall Head Weir Rainfall Continuous Oct 1978-Feb 1980 amount meteorological parameters Throughfall 7,9,11 Volume and Weekly Jan 1979-Feb 1980 chemistry Stemflow 11 Volume and Weekly Mar 1979-Feb 1980 chemistry Spring Sp. 12/13 Head Weekly Jan 1979-Feb 1980 discharge measurement at 'V notch Reservoir Head Weir Reservoir Daily Nov 1978-Feb 1980 level Soil sites Soil moisture Weekly Dec 1978-Feb 1980 gravimetric samples Tensiometers 1,7,10. „ Soil., tension^ ^--Weekly - Oct..1979r-Feb . 1980 Chapter 4 Discharge Characteristics in The Narrator Brook Catchment

4:1 Introduction The analysis of streamflow data is an important component of any hydrogeologlcal investigation, since streamflow hydrograph analysis provides information concerning the baseflow component and hence inferences concerning aquifer characteristics can be derived. In addition to maintaining surface water supplies, streamflow is important In terms of recharge to an aquifer. Spatial variations in discharge may be attributed to Influent and effluent zones along the stream channel and hence the degree of hydraulic continuity between the stream bed and underlying aquifer may be ascertained. The analysis of discharge characteristics in the Narrator Brook catchment has been undertaken with a view to emphasizing the potential of weathered granite areas as a groundwater source. 4:1:1 Sources of Streamflow Streamflow is derived from many sources which vary spatially within a catchment and over time. Three general components can be Identified (a) overland flow and channel precipitation (b) Interflow and (c) groundwater discharge. 4:1:2 Overland flow Two main types are described in the literature; infiltration excess overland flow occurs when rainfall intensity exceeds infiltration capacity (Horton 1933). This kind of overland flow is common where vegetation Is sparse and soils thin, but rare where there is a vegetative cover, (Chorley 1978). It has however been observed In the Narrator Brook Catchment on moorland and acid grassland vegetation (Williams pers. comm. 1979). Saturated overland flow, (Klrkby and Chorley 1967), occurs where the soil is saturated even though the local infiltration capacity has not been exceeded by the rainfall Intensity. Saturated overland flow constitutes an Important component to surface runoff in the stream channel. In the Narrator Brook Catchment saturated overland flow results from saturation upwards from an iron pan.

149 Direct precipitation also contributes to surface runoff. This may occur by direct precipitation onto areas of saturated overland flow, (Ternan and Williams 1979), and onto the stream channel itself. In general, direct precipitation onto the stream channel is only a small percentage of the total volume of water flowing In the channel, it may however be significant in catchments with large areas of saturated overland flow. Overland flow and channel precipitation may be regarded as water which enters the drainage system with little or no delay, and as such contributes to the characteristic peak on the discharge hydrograph as Illustrated In Section 4:2:1. 4:1:3 Interflow Some of the rainfall which Infiltrates through the surface layers passes down to recharge the groundwater, and some, usually the greater part according to Chorley (1969), flows down the hillside within soil layers as 'throughflow*, or Interflow. Interflow results when infiltrating waters meet a relatively Impermeable layer, the downward water movement is retarded in the vertical direction and Instead flows laterally at depths varying from a few centimetres to several metres below the surface towards the stream. Interflow ultimately contributes to streamflow, but the relative speed of response is determined by the depth of the interflow pathways. The contribution of Interflow to total runoff will depend largely on the soil characteristics of a catchment. Ward (1967) suggested that Interflow is likely to be considerable in areas where a thin soil overlies Impermeable rock, or where an iron pan occurs a short distance below the surface, but it is not llkley to be Important In areas where the soil is deep and has uniform hydraulic characteristics. Experimental evidence, (Hertzler 1939) suggested that in some areas Interflow may account for 85% of the total runoff. Interflow may occur at a shallow depth beneath the surface as a swift response to direct precipitation through the soil macro-pores and 'pseudopiping*, (Jones 1971). Where differing texural horizons providing varying hydraulic properties in the soil are present, soil piping may occur. The preferred locations of piping, according to Jones (1971), are either just above or within a horizon of low relative permeability. The development of 'piping' and 'pseudopiping' in a catchment increases the transmission properties of a horizon and

150 as a result, precipitation water infiltrating downwards is transferred laterally and relatively quickly towards the stream channel. Such lateral movement of infiltrating waters may occur at greater depths which results In a longer delay period before contributions to streamflow are received. The potential for Interflow at depth is realised in the Narrator Brook Catchment where variable thicknesses of weathered granite are present In the valley bottom and sides. Ternan and Williams (1979), investigating hydrological pathways in the Narrator Brook Catchment found that as a result of the variable nature of the soil, weathered horizons and slope conditions. Interflow frequency and amount was highly variable in both space and time. Locally this interflow was very significant with discharges up to 5 litres per hour per Im width of slope. 4:1:4 Groundwater Discharge Infiltrating precipitation that is not incorporated in overland flow, evapotranspiration and interflow processes, accrues in the deeper zones of saturation. This portion constitutes groundwater recharge. Baseflow, delivered to the stream by deeper groundwater flow and Interflow, maintains streamflow particularly during dry periods. As groundwater flow follows a more complex route than the more rapid surface runoff and Interflow components, its contribution to the baseflow component of the stream discharge is much delayed (figure 4:3). Rapid recharge to the stream channel produces a rise In the level of the stream surface, which results in influent seepage from the stream channel into the adjacent groundwater body. This results in an accumulation of a wedge of groundwater, known as bank storage on either side of the stream. As the stream level falls effluent seepage is resumed and the volume of bank storage is returned to the stream channel. The relative speed of this depends on the hydraulic properties and nature of the stream banks. 4:1:5 Stream Classification The long-term relationship between groundwater, interflow and surface runoff will obviously determine the main characteristics of a river, and provide a basis for the classification of streams. Streams can be classified into three main types:- ephemeral, intermittent and perennial, (Wisler and Brater 1959).

151 Ephemeral streams flow only during and immediately after rainfall or snowmelt. Normally there are no permanent or well defined channels and the watertable is usally below the bed of the stream. Ephemeral streams in the Narrator Brook Catchment are often fed by ephemeral springs, which have short-duration flows, and may be the result of a rapid saturation upwards process from an iron pan or lens of less permeable material. Streams which flow during wetter seasons but cease to flow in long dry periods are intermittent in nature. This type of stream usually possesses a well defined channel. Groundwater makes some contribution to streamflow during wetter periods, when the watertable rises above the base of the stream. Groundwater contributions to these streams may also be received from intermittent springs in the Narrator Brook Catchment. In the case of the Sheepstor Beck, during dry periods spring flow provides the only source of discharge in upstream sections. Such spring flow is insufficient In dry periods to sustain flow along the entire length without additional inputs from surface runoff or other springs. Perennial streams and springs flow throughout the year even during the most prolonged dry spell. In general the watertable is usually above the bed of the stream, although locally perched watertables may be traversed, which contribute significantly to discharge. The Narrator Brook is perennial in nature, and its major source Is from perennial springs issuing from tin-miners spoil heaps at the eastern end of the catchment. Intermittent and ephemeral flows also contribute to discharge in the Narrator Brook. 4:2 Analysis of Data from Continuous Recording Stations As described in Chapter 3 two main gauging stations located at Station Cutt (St. Cutt) and Station 11 (St. 11), together with data from the South West Water Station at Head Weir, provide the principal source of streamflow data. Streamflow data for the two Narrator Brook Stations is available covering the period 26th May 1975 to 4th March 1980. A total of some 18 months of data is not however, available from the stage recorder at Station Cutt. Station 11 records, with only 12 weeks of missing data, are of a more consistent nature, and consequently this station was chosen for the main discharge control throughout this present investigation.

152 Regression analysis between Station Cutt and Station 11 dally discharges was carried out to generate doubtful or missing discharges for Station Cutt. The regression Is Illustrated In figure A:l. A correlation coefficient of 0.87 (significant at 0.01 Level) was also derived. Using this equation missing Station Cutt data were generated. For the period examined Station Cutt's flows averaged 45% greater than those of Station 11. This Increase In streamflow between the two discharge measuring stations may be accounted for by artificial drainage channels, together with spring and seepage contributions along the banks downstream from Station 11. Such Inputs are discussed In section 4:3. As stre£uiiflow in the Narrator Brook Is being assessed with a view to determining the baseflow component» and thereby Indicating the potential of weathered granite aquifers as groundwater sources. It was considered Important that some estimate of past streamflow conditions should be derived. Such discharge records will provide Information concerning the variation of groundwater discharge from the weathered granite catchment, and the degree of constancy of the contributions to streamflow. Past Inflow data to Burrator Reservoir, as recorded at Head Weir are available since 1957. A further regression analysis was carried out using 3 6 months of flow data from the Narrator Brook (Station Cutt) and the corresponding period for Head Weir. This regression is Illustrated In figure 4:2. A correlation coefficient of 0.84, significant at 0.01 significance level was also derived. The derivation of past flows of the Narrator Brook from this regression equation was not successful. A comparison of derived Narrator Brook discharges with those actually measured in the Narrator Brook during 1977-1979 showed that the regression equation predicted much higher discharge values than those actually recorded. It was considered that the discrepancy In magnitude of estimated flows would Incur large errors In any catchment balance calculations prior to 1975, so the approach was abandoned for this current investigation. Measurement of streamflow at Head Weir Is taken downstream of the confluence of the River Meavy, Newleycombelake, and Narrator Brook. Variations in flow characteristics of the largest stream, the River Meavy, may partly explain the inability to derive more appropriate flows for the Narrator Brook from the regression equation. Other

153 OSr u Of 07 0=011 •0.67X

06 E 05

04 if) 03

0 01 02 03 OA 05 06 07 08 09 1 0 11 1 2 Discharge SMI m3/ sec

FIG 4:1 REGRESSION ST CUTT vs. ST 11 DISCHARGES. FIGURE 4:2 Regression:-

Head Weir (Burrator Reservoir inflows)

vs. Narrator Brook Discharge

10 1

0 9

0 0-05+ 0-22 X

0-7 H

I o 2 0 6 CD O 0-5 2 (Q 0-4 z

0-3

0-2 .

0»

-1 00 2-0 2-5 2-7 0-1 0-3 10

Head Weir (Burrator Inflow^s) Mean Doily F/ow M/s features which may affect such derived flows are discussed In section 4:3, For a more specific study In the Narrator Valley, four water years, where records are available for this catchment, have been selected. These water years span the period 1975-1979. 4:2:1 Hydrograph Analysis and Graphical Techniques The classification of streamflow according to the path followed by the water to the channel as outlined In section 4:1, Is somewhat arbitrary and Imprecise* Nevertheless this three stage distinction leads to useful methods of analysis and affords the means whereby Individual components may be quantified by the division of the storm hydrograph. From the previous discussion In section 4:1 It Is obvious that the stream hydrographs will reflect two very different types of contributions from the catchment, (figure 4:3). The peaks, which are delivered to stream by overland flow and interflow, and sometimes by bank storage, are the result of a fast response to short-terra changes In sub-surface flow systems In hill slopes adjacent to stream channels. The baseflow, which Is delivered to the stream by Interflow and deeper groundwater flow, Is the result of a slow response to long- term changes In the groundwater flow system. Time based definitions of the contribution of these different components to the discharge hydrograph, take Into account the variability of such sources. Weyman (1970) divided streamflow into quick flow and delayed flow (figure 4:3b). Quick flow accounts for overland flow, direct channel precipitation and shallow interflow, while delayed flow accounts for deep Interflow and deeper groundwater contributions to discharge. Hydrograph separation techniques provide an approach to the determination of the relative Importance of different flow origins in runoff prediction. For the surface water hydrologlsts. hydrograph separation is a means by which streamflow prediction models can be Improved, for both high and low flows, (Wright 1974), while for groundwater studies indirect evidence about the nature of the groundwater regime in the catchment Is provided. Many of the stream hydrograph separation techniques frequently used tend to overlook the variable components of discharge, both volumetrlcally and spatially, and rely solely on attributing discharge at any point in time to three distinct components (figure 4:3c).

- 156 01 Total Hydrograph u Of o Rising Limb Surface E Run-off u Surface e Run-off o Recession Limb Q. O > O Interflow Interflow 0.55C/sKm2 cr A Bank Storage per hour Baseflow

Groundwater Discharge Delayed Flow •> Time (Days) Time Time Semilogarithmic plot of a a. Components of a natural b. Time based hydrograph 3-component natural hydrograph separation hydrograph

Hibbert & Cunningham (1967) Barnes (1939)

FIG. 4.3 HYDROGRAPH SEPARATION TECHNIQUES AND COMPONENTS. Traditional hydrograph separation techniques are essentially empirical and are generally based upon the assumption that the beginning of the hydrograph rise is the result of storm runoff in the stream channel (Linsley et al. 1949). In the analysis of storm hydrographs one or more lines may be drawn through the hydrograph to distinguish runoff from interflow and groundwater contributions (figure 4:3)« The precise method of drawing the separation lines varies considerably (section 4:2:2). Two of the most widely used methods devised for hydrograph separation of storm hydrographs, are briefly described as these techniques can be extended to the annual hydrographs to provide estimates of the relative proportions of total runoff originating from surface, interflow and groundwater sources. Barnes (1939), used a procedure of semilogarithmic hydrograph plotting, based on the backward extension of straight line recession curve segments, to distinguish the individual flow components in a storm hydrograph (figure 4:3c). Hibbert and Cunningham (1967), advocated the time based separation, which avoids the controversy over the relative importance of overland flow and interflow in the form of the storm hydrograph, (Weyman 1970) (figure 4:3b). A storm hydrograph is divided into quick flow and delayed flow by a line drawn upwards from the point of the hydrograph rise at a gradient of 0.55 litres/sec per km^ per hr. Walling (1971), used this method on an annual basis to evaluate the relative contributions of quick flow and delayed flow from five small catchments in East Devon, from which it was readily apparent that the proportion of delayed flow was larger from more permeable catchments. The use of hydrograph separation techniques in determining the annual groundwater component in a catchment has not always been successful, (Freeze and Cherry 1979). The success that has been achieved has been based on the concept of a baseflow recession curve. The recession limb or depletion curve comprises the gradually falling trace of streamflow following a storm event when unaffected by rainfall. Investigators such as Horton (1933), Barnes (1939) and Grundy (1951) have suggested that either the entire recession curve.

158 or segments of it, can be expressed as a mathematical function. The simplest function is the basic exponential equation of the form

Qt = Q/'

Where Qo = initial discharge e = base of natural logarithms a = constant t = time Interval

Qt = discharge after time t K = recession or depletion constant representing

Other equations include the double exponential:-

-b Qt = Q^e or Qt = Q^ K*^"

where n = constant K = constant = (e ^)

proposed by Horton (1933) for a drainage basin comprising a series of sub-basins of differing characteristics and the hyperbola:-

(1 + ct)2 where c = constant, after Werner and Sundquist (1952). If the stream- flow recession curve can be fitted by one of these equations then its form can be described by the value of the constant or constants. The exponential function is often used for this purpose because it can be decribed by the K value, or recession constant, which represents the slope of a semilogarithmic plot of river flow against time, (Barnes 1939). Butler (1939) also noted that some groundwater flow recession curves plot as nearly straight lines on semilogarithmic paper, while Ineson and Downing (1964), mentioned that in simple geological conditions a linear semilogarithmic plot of groundwater discharge can be obtained. Because each streamflow component tends to overlap each other, such as groundwater discharge commencing before interflow ceases, then the recession curve of river flow, as defined by the semilogarithmic plot of Q. vs t., assume a curvilinear form. indicating a continuing change In conditions of discharge to river flow. Hall (1968) proposed that most, if not all, natural hydrologlc systems are nonlinear, and that the cause and extent of the non- linearity should be investigated. As a result of the tendency of the different components of stream hydrographs to overlap each other in volume and time, no one separation technique can adequately Isolate such components. On an annual basis the hydrograph of baseflow Integrated overtime can provide a measure of annual aquifer recharge rates, (Birtles 1978). These may then be used to evaluate groundwater resources and assist in the determination of the significance of the groundwater component In the water balance of an area. The Barnes method appears to be a standard technique in hydrogeological literature, for example Webber (1961), Kunkle (1962), Riggs (1964) and Headworth (1972), although Kulandaiswamy (1969) suggested that the method was not helpful for separating and identifying Interflow contributions. However, as a consequence of its simple exponential function, its description by a recession constant, and its well documented utilization, it provides a more readily available means of comparison for baseflow recession. On this basis it was adopted for use in the Narrator Brook Catchment. In hydrogeologically complex areas the derivation of individual components is difficult as each contributing aquifer makes up a varying proportion to river flow, so that this situation is best represented by a series of groundwater recession curves. In this investigation baseflow recession is regarded as being analogous to the recession of groundwater although Riggs (1964) and Ineson and Downing (1964) have discussed problems Involved with determining whether baseflow is exclusively from groundwater discharge. Rapid stream level rises associated with heavy precipitation and the subsequent generation of runoff may cause water to flow into the banks of the channel (Brater 1940). Work by Todd (1955) and Rorabough (1964) has confirmed this origin of bank storage, and in addition has shown that considerable time may be required for the resulting bank storage to drain back into the river channel. In this case the receding limb of the hydrograph, for sometime after the peak- event. Includes flow from bank storage and interflow sources as well as from the principal groundwater sources.

- 160 The presentation of the form of the groundwater recession curve and the identification of the groundwater components or river flow are subjective, (Ineson and Downing 1964), and no single solution may have general application. Hall (1968) in his review concluded that as far as hydrogeologlcal interpretations are concerned, most of the present approaches to hydrograph analysis are unsuccessful or inconclusive. From a practical viewpoint, procedures can be applied to different stream hydrographs in order that estimates of groundwater discharge, which Include bank storage and Interflow contributions, may be derived which are comparable if not absolutely accurate. Derivation of the annual baseflow component in the Narrator Brook and the recession curves and subsequent recession constants are achieved mainly by graphical techniques which are outlined in sections 4:2:2 to 4:2:5. However, in the last two decades there have been some severe criticisms concerning the appropriateness of using graphical techniques to derive recession flow in a catchment area. Laurenson (1961), studying recession curves from a catchment in New South Wales, Australia, referred to an aspect of the semi-logarithmic plots that tends to confuse their interpretation. This author found that breaks in the recession curve slopes occur at discharge values as high as 22.65 m^ sec~^, which he regarded as too high a value to be attributed solely to groundwater flow or groundwater plus interflow. In view of the clayey nature of the soil in the South Creek Catchment this figure seemed too high for groundwater flow over a two-day period, and it was concluded that the break in the recession curve, when plotted on semi- logarithmic paper does not necessarily indicate a change from surface runoff to groundwater flow. Some breaks in the slope may be attributable to the difference in a real distribution of storms which affect the distribution of channel storage in the stream system, and consequently the shape of the recession. Not all slope breaks observed by Laurenson (1961), could be explained in this way, but it was suggested that the temporal pattern of the storm may be a contributing factor. Anderson and Burt (1980) reached the conclusion that runoff generating recession flow cannot be Inferred from graphical plotting techniques. They found by comparing laboratory and field recessions that some graphical techniques produce breaks of slope, which could be

161 interpreted as indicating a change in flow characteristics, even though no change of flow processes has actually occurred under field conditions. A semilog plot was found by Anderson and Burt (1980) to correctly predict a single recession line for a laboratory slope drainage, but did not provide a single straight line for the field recession, even though this was generated by a single flow process* These authors conclude that any inference from graphical plots of recession flow must be treated with caution if unsupported by field evidence. The use of recession flow graphs may be very useful for low flow prediction purposes, but their use as indicators of the flow process operating during recession is probably very limited, (Anderson and Burt 1980). 4:2:2 Derivation of the Baseflow Component for the Narrator Brook Although there are difficulties, as outlined previously, the use of a recession curve technique allows a reasonable estimate of the appearance of groundwater discharge to be made. The main problem with the stream hydrograph analysis is the positioning of the groundwater flow hydrographs under high flow conditions. It is impractical and unrealistic to attempt to show the influence of each flood peak on the form of the underlying hydrograph, therefore several non-rigorous methods have to be adopted. The groundwater recession line recedes along, and is coincident with (or just below), a normal low flow recession until the line approaches a peak discharge. When the groundwater recession line has reached a point equivalent to the time of the peak of the peak streamflow, it then climbs gradually to the groundwater breakoff point on the next recession. Under many closely spaced flood peaks typical of flashy catchments, the groundwater discharge hydrograph is maintained as a smooth slope to the breakoff point on the next recession. This is illustrated for the Narrator Brook (Station 11) on the 1978-79 hydrograph in figure 4:4. Hydrographs utilizing mean dally discharges of the Narrator Brook (Station Cutt) for the period 1976-1979 are included in Appendix 3. The baseflow components for the water years 1975-1979 were derived using this technique and are discussed in section 4:2:5. By calculating the area under the derived groundwater hydrographs for the respective water years, the following factors were determined for the

162 Narrator Brook at Station Cutt and Station 11:- annual groundwater discharge, average daily groundwater discharge, annual baseflow discharge as a percentage of total streamflow, and annual effective infiltration, (section 4:2:5). The formulae used here for the derivation of effective Infiltration (le) expresses the groundwater flow (Qgw), as an equivalent depth of water over the catchment, and Is given by:-

S^^S^vrlOOO X 86400 X Qgw ^ - . = le (mm.) Ad

where le = average effective infiltration (mm.) Qgw = mean daily groundwater discharge (m^ sec*"^) Ad = Catchment area (m^)

This method is based on the assumption that groundwater flow represents that part of the precipitation that has reached the watertable, and that infiltration takes place over the whole catchment area. The formulae is therefore restricted and no account is taken of the differential infiltration into the more permeable or less permeable materials in the Narrator Brook Catchment. The average annual groundwater discharge may be regarded as the average annual replenishment resource since it represents water effectively infiltrating during times of the year when precipitation exceeds evapotranspiration, and the soils are at field capacity. 4:2:3 Baseflow Recession in the Narrator Brook Ideally a complete groundwater recession curve would be derived over the period following the main season of heavy rainfall, providing that no increment to surface flow results from intermittent storms during this period. Unfortunately meteorological conditions in this country rarely provide the opportunity to derive such a curve for a particular drainage basin. Barnes (1939), suggested that the point on the recession limb corresponding to the point at which streamflow firsts consists entirely of groundwater, can be approximated by plotting semilogarithmically post peak discharge against time. It is impractical to plot each individual recession in the manner suggested

- 163 FIGURE 4:4 Narrator Brook (ST 11 )

Mean Daily Discharge 1978-1979

Showing Derived Boseflow Component (.

9 J -P i

^ 2 5 o y

Jan 79 by Barnes (1939), but the technique can be modified by preparing a composite or synthetic recession curve. This master recession curve is produced by fitting together all the shorter Individual recessions from the stream hydrograph. Grundy (1951), used cardboard templates of his derived groundwater depletion curve, while Toebes and Strang (1964) used the matching strip method by preparing transparent overlays for all recession periods to obtain a synthetic master recession curve. In the case of the Narrator Brook, on an individual water year basis, all short recession periods were traced out from the hydrographs, and fitted together to give a recession curve for each water year Illustrated for Station 11 in figure 4:5. These four curves were then smoothed together to produce the master recession curve (fig 4:6b). Master recession curves for both Station Cutt (figure 4:6a) and Station 11 (figure 4:6b) were also produced. •Generally in the Narrator Brook good streamflow recessions are not apparent on the hydrograph plots due to the highly variable 'flashy' flow regime. This variability is Illustrated by the absolute maximum and minimum mean dally flows in table 4:1. Table 4:1 Range of Discharge in the Narrator Brook 1976-1979

Mean dally flow 1976 1977 1978 1979

Maximum 0.4 1.00 0.94 2.11 Minimum 0.01 0.04 0.04 0.10 / 3 _1 X (m sec ;

When overland flow occurs in the catchment, then frequent rapid increases in runoff results, giving a hydrograph with a fretted appearance. A smoother hydrograph indicates areas where groundwater runoff predominates, (Ward 1967). In the Narrator Valley this feature may be overshadowed due to Interflow contributions and runoff through fissured granites higher up in the catchment. The recession curve by Itself tells, in a general way, the amount of flow from baseflow sources in a catchment, and the time required to substantially deplete the source, (Rlggs 1964). The master recession curve for the Narrator Brook (Station Cutt) as illustrated in figure

165 FIGURE 4:5

Groundwater Depletion Curves(ST11 ) 0 75 H

Narrator Brook 1975-79 a-

0-50 H

o 0-25 H

00 -r -I 1 1 1 r 1— 1 1 1 r 10 20 7^ 60 70 80 90 100 no 120 130 140 150 160 170 180

7"//ne in Days FIGURE 4:6a

Master Recession Curve

1-0 Narrator Brook ST CUTT 1975-1979

OB 1

06

0-4 E

O 0 3

700 •220 0 0

Time in Days FIGURE 4:6b

Master Recession Curve (SITl) 005

0 25 o

0 0 -f —I— -i— -T 1 1 \ 1 ^ T 120 130 140 150 160 170 180 190 200 0 10 20 30 40 50 60 70 80 90 100 110

Time in Days 4:6a has a time axis of greater than 210 days, while that for Station 11 has one of 190 days (figure 4:6b). By comparison with other recession times relative to the catchment area, groundwater In the Narrator Valley Is a significant component of streamflow In a small weathered granite catchment on Dartmoor (table 4:2).

Table 4:2 Comparison of Recession Times and Catchment Areas

River Catchment Area Recession Time Authors km^ (days)

Test 1043.8 270 Webber (1961) Hampshire UK

Itchen 3 60 270 Webber (1961) Hampshire UK

Meon 72.8 260 Webber (1961) Hampshire UK

Narrator )Statlon 11 3.44 190 Alexander Brook, UK)Statlon Cutt 4.75 210

Makuyu, 77.7 440 Grundy (1951) Kenya

South Creek 88.0 25 Lauren8on(1961) New South Wales Australia

4:2:4 Recession-Constants Under Ideal hydrogeologlcal conditions a semllogarlthmlc plot of a 83rnthetlc recession curve would enable three straight lines of best- fit to be fitted to the data points. These would represent, surface flow. Interflow, and groundwater flow with typical Barnes constants of 0.329, 0.694 and 0.980 respectively. Under field conditions, Ineson and Downing (1964), noted that the recession constant K Increases from values as low as 0.10 for the recession of surface runoff, to values as high as 0.99 for groundwater recession.

169 - A seml-logarithmic plot of the master curves of figures 4:6a, 4:6b were produced (figs 4:7, 4:8) and the Barnes recession constants obtained by the relationship t Qt = Q^K

log Q - log Qt o using - log K

By substitution from values on the graph of figures 4:7, 4:8 Barnes* recession constants are derived as

= 0.9502

K2 = 0.9908 for Station 11

and = 0.9462

K2 = 0.9937 for Station Cutt

Previous recession values derived for the Narrator Brook, Sims (1975) obtained values of 0.9759 and 0.9984 which were interpreted as repre• senting Interflow and baseflow respectively. In a hydrogeologically complex area like the Narrator Brook Catchment the distinction between interflow and baseflow components Is an arbitrary decision since contributions are of a multicomponent nature. The recession constants produced show a two-component nature which is due primarily to the delay time of the discharge components.

170 FIGURE 4:7

Semi-Log Plot of Master Recession Curve

Narrator Brook sr. CUTT 1975-1979 10 3 Barnes. % 09462 0 5 H Morton t 0 055

0-2 Barnes i 0 9937 Morton i 0 0062 010

o-osH

002

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220

Time In Days RGURE 4:8

Semi-Log Plot of the Master Recession Curve

Narrator Brook 1975-1979 (ST 11) 10 i

Barnes m 0-9502 0*3

Norton n 0 051

-4 1 4 0! O Barnes • 0 9908

0-03 Hortonm 0 0092

0O2

00 n -T— 1— -I 170 20 30 40 50 60 70 60 90 lo: no 120 130 UO ISO 160 I BO

Time in Days In the Narrator Brook Catchment, horizons with widely different properties drain at distinctly different rates. This Is a likely situation in the upper reaches of the catchment where more permeable areas of weathered granite, clitter slopes and mining spoil heaps, drain more quickly towards the stream. Lenses of clay materials and large scale textural variations in the weathered granite may retard water movement downwards towards the stream channel, and zones of deep-seated interflow may contribute to groundwater recharge after a considerable time period has elapsed since the recharge event. The joints and fissures within the granite bedrock of the catchment may give rise to a large number of springs and seepages which contribute to the total groundwater component of the Narrator Brook Catchment. Such contributions to the main stream channel will have very different decay or recession rates, which could account for the two-component nature of the groundwater recession curve. Baseflow discharge into a river is influenced by evapotrans- piration losses which are particularly important during the dryer months, when recession takes place over a longer time period. Such losses, along with other minor fluctuations in river flow, result In an Imperfect plot of the logarithm of discharge against time. The fact that the straight line segments rarely yield typical Barnes recession constant values is^^expected in hydrogeologlcally complex areas. In such situations composite groundwater flow from different aquifers of different gross yields and storage properties, cause variations in the logarithms of discharge versus time plots, and result in multlcomponent recessions, (Nutbrown and Downing 1964). Because of the multicomponent nature of the contributions, both volumetrically and spatially, to streamflow in the Narrator Brook, hydrograph separation techniques will not be able to easily differentiate between the Individual components. In broad terms a division between surface and sub-surface components may be approxi• mated which gives an indication of the magnitude of groundwater contribution to streamflow, 4:2:5 Active storage in the Narrator Brook Catchment The quantity of water in active storage Is derived from the baseflow recession constant and is quantified in tables 4:3, 4:4.

- 173 - Active storage values in this case represent the total volume of water availabley for streamflow within the aquifer above the elevation of the gauging station. Figures in tables 4:3, 4:4 are the actual quantity of water in temporary storage above Station Cutt and Station 11. for a given time, for any mean daily groundwater discharge.

Table 4:3 Groundwater, Effective Infiltration and Storage Components, Station Cutt

Year Q(m^/sec) Qgw Qgw as le(mm) Daily Active Storage (m /sec) % of Q Mean 3 ^^""^ 3 Qgw (m^'/day x lO")

1975-76 53.92 48.56 90.05 883 0.147 2048

1976-77 81.61 69.16 84.74 1257 0.223 3107

1977-78 81.01 63.90 78;87 1162 0.221 3079

1978-79 81.92 67.08 81.88 1220 0.224 3121

Mean 74.61 62.17 83.88 1130 0.203 2838

Station Cutt catchment area ° 4.75 km^ Q total discharge m^/sec groundwater component of discharge m^/sec, from Qgw separated baseflow hydrograph Qgw expressed in millimetres over the catchment area le (effective infiltration) active storage for the catchment above Station Cutt (St)

174 - Table 4:4 Groundwater, Effective Infiltration and Storage Components, Station 11

Year Q(m^/sec) Qgw Qgw as le(mm) Dally Active Storage (m /sec) % of Q Mean Qgw (mVday x 10^)

1975-76 24.78 16.06 64.81 403 0.044 413

1976-77 65.22 49.83 76.40 1251 0.136 1277

1977-78 63.94 47.70 74.60 1198 0.130 1220

1978-79 62.95 44.62 70.88 U20 0.122 1145

Mean 54.22 39.55 71.67 993 0.108 1014

Station 11 catchment area = 3.44 km^ Q = total discharge m^/sec Qgw = groundwater component of discharge m^/sec, from separated baseflow hydrograph le = Qgw expressed In millimetres over the catchment area (effective infiltration) (St) = active storage for the catchment above Station 11

The calculation of active storage is based on the Horton recession equatlon:-

(St)

where (St) =» the amount of water in storage available for discharge above the gauging station (m^/sec x 10^)

Q^ = baseflow discharge at time (t) after the beginning of the recession (days)

Horton's groundwater recession constant

- 175 This equation assumes that groundwater and surface water components of discharge can be Isolated and that the beginning of the groundwater recession can be easily Identified, which has been shown In section 4:2:2 not to be the case In the Narrator Valley. Since hydrograph separation techniques adopted In this Instance are considered to approximate rapid and delayed contributions to stream discharge, the use of the Horton recession equation may be Justified. The Barnes recession constants for the Narrator Brook were converted to equivalent Horton recession constants, figure 4:7. 4:8 by:-

K^ = 2.303 log . K* where K* = Barnes recession constant

From a comparison of tables 4:3 and 4:4 It Is obvious that the lower part of the Narrator Brook Catchment Is behaving In a different fashion to the upstream catchment above Station 11. Station Cutt possesses a much larger groundwater component of discharge (84%) than Station 11 (72%). This can be attributed possibly to the greater thicknesses of weathered granite in the lower catchment with varying hydraulic properties, and to the greater contributions to streamflow lower down the valley from springs and seeps. The variations in total discharge per unit area of catchment are discussed in section 4:3. The greater amounts of groundwater available in this downstream area are substantiated by the dominance of springs and seepages between Station Cutt and Station 11 In comparison to those above Station 11. Figure 2:18 in Chapter 2 and figure 3:6 In Chapter 3 illustrate the distribution of springs, seeps and marshy areas over the catchment. Further hydrogeological differences between the upper and lower catchment are summarised in section 4:4:3. The groundwater recession line. Illustrated In figures 4:7, 4:8 as a dotted line, was extrapolated back to where t = 0 in order to obtain the theoretical maximum groundwater storage. The values achieved for Station 11 and Station Cutt are presented in table 4:5.

176 Table 4:5 Theoretical Maximum Groundwater Storage at Station Cutt and Station 11

Theoretical Average Groundwater Storage Mean Active Maximum per Water Year m^^yr x 10^ Storage Groundwater m/day x 10^ Storage /sec

0.24 7568 2838 Station Cutt

0.10 3153 1014 Station 11

It can be seen that the annual active storage represents the smaller component of groundwater recharge. Storage factors in the Narrator Brook Catchment are outlined in table 4:6. The mean stream surplus in table 4:6 represents that volume of water which could be abstracted without depletion of storage or baseflow contribution to the natural discharge of the Narrator Brook as measured at Station Cutt. The Narrator Brook discharges into the Reservoir downstream, which supplies drinking water to Plymouth and its environs, so the potential for abstraction from the Narrator Valley is curtailed.

Table 4:6 Storage Factors in the Narrator Brook Catchment

Component per day per Water Year

Maximum Groundwater Storage 0. J Zf m^/sec 7.562x10^ m3/yr Average Active Storage 28.38 xlO^ m^/d 1.03 xlO^ m^/yr Mean Stream Discharge 74.61 m/^sec 2.35 xlO^ m^/yr Mean Groundwater Discharge 62.17 m/^sec 1.96 xlO^ m^/yr Mean Stream Surplus 12.44 m/^sec 3.92 xlO^ m^/yr

During the droughtoll976, a large proportion of springs feeding streams in the South West region dried up, (Brooker and Mildren 1977) The lowest flow of 0.006 m^ sec"^ was recorded in the Narrator Brook during this period, suggesting that groundwater makes a sustained

177 - contribution to streamflow in this valley. Despite the importance of the groundwater component in the Narrator Brook, the general hydrograph shape, as illustrated in figure 4:4-, is somewhat fretted. This feature has already been discussed in section 4:2, but is worthy of further comment in the light of the foregoing discussion. Large tracts of the Narrator Brook Catchment are characterised by sequences of varying thicknesses and textures of weathered granite, fractured granite blocks and soil horizons. Each horizon possesses varying hydraulic properties, and Isolated pockets of permeable materials of limited temporary storage capacity, may well ensure the 'flashy* characteristics of the hydrograph, particulary in winter months. Flashy hydrographs of the Narrator Brook and other Dartmoor streams may also be attributed to saturated overland flow from the moorland. Where an iron-pan Is present saturation occurs upwards resulting in moorland which is more or less permanently saturated in the winter months. Because of the impeding effect of the iron-pan, Infiltration is minimal, so saturated overland flow contributes to moorland tributaries, resulting in flashy hydrographs. Saturated areas in the Narrator Brook Catchment (illustrated in figure 2:17 in Chapter 2), have been observed in wet periods (by the author) to expand markedly, thereby effectively expanding the direct contributory regions to the channel network. This also results in •flashy* features in the hydrograph shape. Whlpkey and Kirkby (1978), have outlined the importance of soil and weathered zones on the stream hydrograph shape. Once sub-surface flow has drained into the impeding soil layer, it drains out very slowly with a response time of several weeks. As a result the soil hydrograph is likely to be a major determinant of drainage basin recession curves where a suitable impeding layer is present, but only where groundwater does not play a major role, (Whipkey and Kirkby 1978). It can be seen that rapid surface runoff effects in the Narrator Brook Catchment may mask the relative importance of groundwater flow contributions to streamflow, but in dry periods the significance of baseflow contributions to stream discharge Is more apparent.

178 - 4:3 Analysis of Data from Nor^-Contlnuous Streamflow Gauging Sites It had been observed in previous years, during undergraduate streamflow measurement exercises, that the flow characteristics of the Narrator Brook showed some anomalies, (pers. comm. Sims, 1978). The flow appeared both to Increase, and in some sections, decrease in a downstream direction. This suggested that the Narrator Brook exhibited at certain periods, both Influent and effluent sections along its course, indicating some hydraulic continuity with the valley aquifer. This characteristic may also affect the hydrograph shape. During this investigation, a study stretch of some 600 metres of the Narrator Brook, between the discharge measuring stations, was gauged. An attempt was made to determine the presence or absence of hydraulic continuity between the channel and weathered granite aquifer, along this stretch. Three stream gauging sites C, D and E were chosen between Station Cutt and Station 11 (figure 4:9). As can be seen from this figure several inputs to the Narrator Brook are evident along this 600 metre stretch of the stream channel, none of which were monitored during this stream gauging exercise. Current metering was carried out during the water year 1978-79 on this study section. Such velocity measurements were taken from 1-4 times in any month and at periods considered to correspond with the rising limb, peak and recession of any given storm hydrograph. Velocity measurements were converted to discharge measurements taking into account the cross-sectional area of the stream at an individual gauging site. These derived flows were compiled, together with flows at Station Cutt and Station 11 on the same day, and compared with a precipitation plot for the Narrator Brook Catchment. The comparison of rainfall events with individual records for stream discharge at each station and site, provided a complex view of seasonal variation of influent and effluent conditions in the Narrator Brook. An approach which considered contributions to streamflow per unit area, taking Independently the individual basin areas between stream discharge stations and sites, was adopted in an attempt to identify Influent and effluent sections of the Narrator Brook. This method minimises the effect of ephemeral tributary and spring contributions to the main channel (which were not measured Independently) as discharge in the stream channel becomes proportional to a particular stream-gauging site or discharge station's drainage area. Table 4:7 illustrates the catchment drainage area between stations and sites, and the distances along the stream between each of the monitored locations. - 179 - KEY

A Stream discharge Newleyco^ measuring stations / E Stream discharge measuring sites

J effluent sections of Channel

Roughtor Influent sections Plantation \ of Channel

Boundary of

\ Forest Plantation

FIG. 4:9 STREAM GAUGING STUDY SECTION WITH EFFLUENT AND INFLUENT REGIONS IN THE NARRATOR BROOK CATCHMENT Table 4:7 Area of Contributory Surfaces to Discharge Sites and Stations, Narrator Brook

Station or Site Area (km ) Distance along stream between sites (m)

St. 11 and 3.44 - catchment boundary

St. 11 and C 0.288 150

C and D 0.233 130

D and E 0.133 80

E and St. Cutt 0.65 240 Total Total Catchment 4.75 length 600 Area

Moving downstream from Station 11 the following discharge variations are apparent:- (a) between Station U and C effluent section (b) between C and D slightly effluent (c) between D and E strongly effluent (d) between E and Station Cutt influent section and are illustrated in Figure 4:9. Table 4:8 illustrates Q m^ sec"^ per km^ of contributing catchment area. The contributing area is that measured between two sites (eg. site D and C); between a station and a site (eg. Station Cutt and Site E); and between Station 11 and the catchment boundary. Figure 4:10 illustrates these discharges variations per unit drainage area graphically. Generally over most of the catchment, groundwater contributes to streamflow, except In the lower portions of the Narrator Brook where influent conditions are present from Site E, down towards Burrator Reservoir. Table 4:8 tabulates the average percentage of streamflow loss between discharge at Site E and Station Cutt. This constitutes nearly 30% of total discharge measured at Site E. These Influent conditions in the lower reaches of the Narrator Brook may be due partly to the greater transmlssivity of the coarser valley Infill In

- 181 Table 4:8 Streamflow Per Unit Drainage Area, at Discharge Sites and Stations, Along a Study Section of the Narrator Brook Channel (m^ sec- 1 km-2)

St.11 Site C Site D Site E St Cutt (m 3 sec-^ km-2) (Catchment (St 11 (C-D) (D-E) (fi• Loss of Flow at Date Boundary - C) st Cutt St Cutt = - St 11) St Cutt as % E flow

20.10.78 0.013 0.138 0.214 0.526 0.230 43.72 7.12.78 0.045 0.315 0.351 0.481 0.323 67.15 12. 1.79 0.049 0.847 1.304 2.075 0.338 16.28 12. 2.79 0.095 1.458 1.888 2.932 0.476 16.23 20. 2.79 0.055 0.902 1.158 1.954 0.369 18.88 27. 2.79 0.055 0.756 0.931 1.639 0.369 22.51 19. 3.79 0.078 0.666 1.540 2.526 0.338 13.38 9 4.79 0.049 0.656 1.000 1.571 0.338 21.51 19. 4.79 0.046 0.548 0.755 1.368 0.338 24.70 24. 4.79 0.052 0.833 1.060 1.511 0.353 23.36 18 5.79 0.034 0.375 0.386 0.924 0.307 33.22 25. 5.79 0.037 0.427 0.420 1.015 0.307 30.24 5. 6.79 0.049 0.552 0.866 1.308 0.338 25.84 20. 6.79 0.037 0.444 0.751 1.030 0.307 29.80 3. 7.79 0.026 0.312 0.450 0.872 0.276 31.66 10. 7.79 0.020 0.260 0.484 0.721 0.261 36.20 24. 7.79 0.017 0.222 0.343 0.466 0.246 52.79 31. 7.79 0.058 0.354 0.622 1.300 0.369 28.39 7. 8.79 0.026 0.288 0.369 0.593 0.276 46.55 14. 8.79 0.130 0.750 1.918 2.421 0.584 24.13 21. 8.79 0.034 0.496 0.480 0.932 0.307 32.94 4. 9.79 0.029 0.357 0.399 1.105 0.292 26.43 11. 9.79 0.026 0.291 0.420 0.744 0.276 37.10 25. 9.79 0.026 0.243 0.493 0.804 0.276 34.33 2.10.79 0.026 0.229 0.309 0.586 0.277 47.27 9.10.79 0.026 0.264 0.356 0.797 0.276 34.63 16.10.79 0.023 0.232 0.317 0.676 0.261 38.61 16.11.79 0.055 0.545 0.914 1.924 0.369 19.18 20.11.79 0.063 0.465 1.167 1.872 0.384 20.52 18.12.79 0.110 1.149 1.763 3.248 0.523 16.11 22. 1.80 0.043 0.923 1.055 2.150 0.323 15.03 5. 2.80 0.075 1.381 1.768 3.255 0.415 12.75

Average % of stream flow lost between Site E and Station Cutt = 29.43%

182 FIGURE 4:10

Contributions to Streamflow per unit Drainage Area, of Discharge Sites and Stations^ along a Study Section of the Narrator Brook

« 0-1

^0-09

aoo7

0-06

ST. CUTT ' 100 200 300 400 Distance between discharge sites and stations, metres this section, particularly in the vicinity of well 6, and to the draining of groundwaters from weathered materials towards the lower elevation of the water-level in the reservoir. Figure 4:11 illustrates the relationship between the percentage loss of flow to the aquifer between Site E and Station Cutt, and the absolute discharge at Station Cutt giving a significant negative correlation coefficient - 0.58 (0.001 significance level). From this figure it is apparent that the larger the flow is in the stream channel the smaller is the loss to the valley aquifer through the Influent section. This again relates to the greater transmissivity of some materials in the lower valley section. In wet periods the rapidly moving water table is raised sufficiently in order to reduce the hydraulic gradient from the channel to the aquifer and hence reduce the volume of water transmitted. 4:4 The Stability of Groundwater Discharge in tlie Narrator Brook Catchment Water measured at a gauging station is the surface outflow of the drainage basin above a specified point on a stream, and the discharge record integrates the effects of climate, topography and geology, giving the distribution of runoff in time and magnitude. When the flows are arranged according to frequency of occurrence and a flow- duration curve is plotted, the resulting curve shows the integrated effect of the various factors that affect runoff throughout a range of discharges, and provides some indication as to the stability of groundwater discharge. 4:4:1 Flow Duration Analysis in the Narrator Brook Catchment Using the method described by Searcy (1959), flow duration curves were plotted for Station Cutt and Station 11 utilising mean daily flows at each location over a four year period, (figures 4:12a, 4:12b). Tabulations of this data are found in Appendix 3. Details concerning the extreme ends of the discharge range are somewhat obscured, so to overcome this disadvantage a log-normal probability scale was employed on all individual four years data at each station, figures 4:13, 4:14, and on the compounded 4-year plot for Station 11 and Station Cutt (figure 4:15). The shape of the flow duration curve is determined by the hydrogeologlcal characteristics of a drainage area, (Searcy 1959). A curve with a steep slope throughout denotes a highly variable stream whose flow is largely from direct runoff, where

- 184 FIGURE 4i11 LOSS OF FLOW TO AQUIFER BETWEEN SITE E AND ST. CUTT, AGAINST DISCHARGE AT ST CUTT

Ya a • bi 70 aa - 130-99

bs 57-97 60 J correlation coofflclent • -0-58 u 50

40 H

30 A

20

10

01 0-2 0-3 0-4 05

Q ST CUTT m^aec"^ 0-76 • 0-76i 0754 FIGURE 4t12a 0-72. FIGURE 4:12b FLOW DURATION CURVE 0-69 FLOW DURATION CURVE ST CUTT 1976-1979 0-66 ST 11 1976-1979 0-63- 0-60- 0-57 - 0-54-1 0-51- a48. 0-45- 0-45

039 0-39

0-33 033

0-27 0-27

0-21 0-21

0-15 o 0-15

009 009H

003 003H \ T 1 1 1 1 1 r 1 1 1 0 -i r——I 1 1 1 1 r i i i 0 10 20 30 40 50 6 0 70 8 0 90 100 0 10 20 30 40 50 60 70 80 90 100

PERCENT OF TIME EXCEEDED 1977 1979

a*

n E

\ 005 h 1977 1978

001 0.1 05 1 2 5 10 20 30 t*0 50 60 70 80 90 95 98 99 99 8 999 99 99 % time exceeded FIG. 4:13 LOG-PROBABILITY PLOTS FLOW DURATION CURVES ST 11 NARRATOR BROOK 1976-1979. ror

0.5 h

01

op

\ 0.05

001 I I J L J I I L J L J L 0 01 0.05 0 2 0.5 1 2 5 10 20 30 40 50 60 70 80 90 95 98 99 99.8 999 9999 % time exceeded FIG. 4:14 LOG-PROBABILITY PLOTS FLOW DURATION CURVES ST CUTT NARRATOR BROOK 1976-1979. u St.Cutt cn E

005 h

0 01 0 05 0 2 0 5 1 2 5 10 20 30 40 50 60 70 80 90 95 98 99 99 8 999 99-99 % time exceeded FIG. 4:15 LOG-PROBABILITY PLOT 4-YEAR FLOW DURATION CURVE NARRATOR BROOK 1976-1979 a curve with a flat slope reveals the presence of surface or groundwater storage which tends to equalise the flow. The slope of the lower end of the flow duration curve shows the characteristics of the perennial storage in the drainage basin; a steep slope at the lower end indicates a negligible amount. A steep curve which covers a wide range of flows is characteristic of a flashy river where direct surface runoff predominates owing to the largely impermeable nature of catchment geology. This steeper portion is illustrated in figures 4:12a, 4:12b, of the flow duration curve for both stations on the Narrator Brook and results from peak flows due to high-intensity long-duration winter precipitation. Contributions to these peak discharges are likely to be provided by fissure flow from granite outcrops and clitter slopes higher up in the catchment, along with saturated overland flow facilitated by iron-pan horizons in the Narrator Valley weathered horizons. From figures 4:12a and 4:12b it is evident that the upper and lower catchments behave in hydrogeologically different manners. Both flow duration curves begin to flatten out towards their middle and lower ends indicating that groundwater contribution to streamflow is significant. Station Cutt has a much flatter curve (fig 4:12a) than Station 11 (fig 4:12b) indicating that the groundwater component is larger in the lower catchment than in the higher catchment area; a fact that is discerned by active storage calculations in section 4:2:5. As can be seen by comparing figures 4:13, 4:14 and 4:15, all Station Cutt's and all of Station 11 individual year plots show great similarities. From a comparison of figures 4:15 50% and 90% percentiles, it can be seen (table 4:9) that the baseflow component at Station Cutt is higher than that of Station 11 as supported by analysis (section 4:2). The high values at 90% percentile for both stations further suggests that there is significant base flow which is ascribed to groundwater contributions.

Table 4:9 Flow Duration Analysis Narrator Brook

Station Cutt m^ sec"^ Station 11 m^ sec"^ % of flow exceeded

0.23 0.13 50% percentile 0.18 0.04 90% percentile

190 - As an estimate of bascflow contributions to streamflow from the four year average flow duration curve, a line joining the flatter regions of both curves*^was extrapolated backwards to intersect the ordinate. In the case of Station 11 this value Is 0,30 sec~^ and Station Cutt 0,33 sec"^, which occurs for 86% and 88% of the four year time period, respectively. One may infer that the Narrator Brook is fed entirely by baseflow at all measured flows less than 0*30 and 0.33 m^sec"^. These figures compare well with those of Table 4:4 and Table 4:3 for Station 11 and Station Cutt giving the average groundwater component as a percentage of streamflow as 71.67% and 83.88% respectively. These figures demonstrate conclusively the importance of the groundwater contribution to the Narrator Brook, and reflect favourably on the potential storage properties of weathered granite valley aquifers. 4:4:2 Flow Duration Curve Indices Indices have been developed to quantify the slope of a flow duration curve, and Hall (1967) used the ratio of the flow level exceeded 30% of the time to that exceeded 70% of the time as a means of curve quantification. Lane and Lei (1950) proposed the variability Index as a means of quantifying the flow duration curve. This is the standard deviation of the flow data and is derived by reading off the values of discharge at 10% duration intervals between 5 and 95% and calculating the standard deviation of the logarithms of these values. The variability index has been adopted here as a means of comparison between flow duration curves constructed for the Narrator Brook at both Station 11 and Station Cutt. Where the index value is high the variation is great, and conversely where the index Is low, the variation is small. As can be seen from Table 4:10 Station Cutt has a low variability index while that of Station U Is slightly higher. The variability index for an average river is approximately 0.6, (Gregory and Walling 1973), but Lane and Lei found a range between 0.14 and 1.17 in their study of rivers in the eastern United States, which suggests the Narrator Brook has overall a low variability index. In the case of the Narrator Brook the differences in variability Indices between Station Cutt and Station 11 again suggest hydrogeological differences between the upper and lower parts of this drainage basin. The higher indices 0.25 - 0.43 may be explained partly by the variety of flow contributing units in the upper

191 catchment area, along with their spatial distribution and frequency of flow. These comprise of rapid recharge via the clltter slopes and massive jointed granite tors» saturated overland flow due to Impedence of Infiltrating waters by iron pan accumulations, and spring and seepage contributions. The lower variability of flow between Station 11 and Station Cutt can be attributed to the storage capacity of weathered granite materials in these locations. This results in greater uniformity of flow on a yearly basis in this part of the weathered granite aquifer. Springs and seepages are also noted in this middle section of the Narrator Valley, but these are observed to be of a more consistent nature in this area, being mostly Intermittent and perennial, compared to the empheral, shorter duration flows higher upstream.

Table 4:10 Variability Indices for the Narrator Brook, after Lane and Lei (1950)

Mean Log of Q Sum of Squares (a) or Variability Index

St. Cutt 1976 0.70 0.172 0.138 1977 - 0.63 0.157 0.132 1978 - 0.62 0.162 0.145 1979 -— 0.61 0.099 0.105

St. 11 1976 - 1.15 - 1.715 0.436 1977 0.82 - 0.852 0.307 1978 - 0.92 - 1.08 0.346 1979 —- 0.80 - 0.60 0.258

The two-component nature of the groundwater recession curves has been referred to previously in section 4:2:4. Recourse is now made to Station 11 and Station Cutt's recession curves by way of illustrating the differences in the overall hydrogeological properties of the weathered granite aquifer between Station Cutt and Station 11, and upstream of Station 11. Using the mean groundwater discharge, (Qgw), and the Horton recession constants, active storage (St) is calculated using and K^. The active storage of the early part of the groundwater recession is likely to be more readily available for streamflow and is expressed

192 as a percentage of the total active storage for the catchment at each station (table 4:U). By reference to this table It can be seen that above Station Cutt 11.0% of the active storage Is released In the first 20 days of the recession (figure 4:7), whilst 18.0% of the available active storage is released in the first 20 days of the recession above Station 11 (figure 4:8). This further suggests that the storage capacity of the weathered granite aquifer is greater in the downstream region, and responds in a more uniform manner to recharge than the Isolated and widely distributed zones upstream*

Table 4:11 Two-Component Storage Factors

Mean Mean Active Qgw Storage

PI Sec m^d-l

Station Cutt

0.203 0.055 319 X 10^

0.203 0.0062 28J8 X 10^

11.24% of total St. is released to the stream in the first 20 days of the recession

Station 11

0. 08 0.051 135 X 10^

0.08 0.00*12

17.97% of total active storage is released to the stream in the first 20 days of the recession

4:4:3 Water Balance in the Narrator Brook Catchment Streamflow gauging stations measure catchment runoff In the stream channel at any given moment in time only. A component of discharge from effluent sections of the channel, unsaturated lateral

- 193 - flow, and a flow component parallel with the river channel may be undetected. This underflow fraction may be a significant amount, particularly In areas with a variety of permeable weathered products in the vicinity of the river. Substantial underflow from the catchment upstream of Station 11 may account for the higher proportion of groundwater discharge in streamflow in the lower catchment area in the vicinity of Station Cutt. To obtain an approximation of the amount of underflow passing out through the Narrator Brook Catchment beneath the gauging station at Station Cutt, Darcy's equation (Q KiA) was utilised. The basic velocity equation: V = ~- is substituted into Darcy's equation giving vA = KAl, and v => Kl. A porosity correction is applied because groundwater only moves through the pores giving V = where V = average bulk velocity (m/d) K = hydraulic conductivity (m/d) 0 = porosity (%) 1 = hydraulic gradient Q = underflow (m^/d) A = area of valley cross-section (m^)

In conjunction with the basic velocity equation re-written as Q = vA, underflow estimates In the Narrator Valley were determined.

A cross-valley section of the aquifer of 4530 m^ in area,for potential underflow, was calculated in the vicinity of Station Cutt. The dimensions used are listed in Table 4:12. The mean hydraulic conductivity (K), is derived from measurements in the catchment as outlined In Chapter 6. An average porosity of 25% for the valley materials were utilised after Freeze and Cherry's (1979) recommendations for porosities of silts, sands and gravels being between 25-50%.

194 Table 4:12 Components for the Calculation of Cross Underflow at Station Cutt

Mean Hydraulic Gradient Mean Valley Width Mean Aquifer K Underflow 1 m Thickness m mday"^ m^day"^

0.03 302 15 1.106 601.22

By substition into the Darcy equation a gross value for underflow of 601 m^ day"^ (table 4:12) was derived. This results in underflow quantities below the gauging station being on average some 5% of a given discharge in the stre£un channel at any time. In view of the varying influent and effluent sections along the stream channel, the assumptions made in order to determine potential underflow by the Darcy equation are gross approximations only. Averaged hydraulic gradient values determined for the lower part of the catchment are considered not to be appropriate for the higher catchment location, and consequently an estimate for underflow at Station 11 was not derived. Since no definitive depths for thickness of the weathered horizons in the Narrator Valley are currently available, particularly for higher areas In the catchment, this too would lead to large errors in the determination of underflow volumes. As a consequence underflow values are not definitive and may be orders of magnitude larger in some sections of the valley and much smaller at upstream locations. A gross catchment balance was undertaken using the precipitation and streamflow data for the four water years 1975-79, in order to ascertain the potential groundwater recharge to the Narrator Brook Catchment. A catchment balance was determined for both Station Cutt and Station 11 which again emphasised the hydrogeological differences between the higher and lower catchment regions. Tables 4:13 and 4:14 attempt to quantify the amount of rainfall infiltrating to groundwater and feeding baseflow. The residue columns refer to the percentage of effective rainfall left over after the streamflow component has been subtracted. This value is available within the catchment to be utilised for soil moisture deficit, under• flow and recharge to the weathered aquifer in addition to deeper

195 percolation to the granite bedrock via joint and fissure systems. Column ERF refers to the effective rainfall in the catchment after evapotranspiration losses have been deducted. Column Qmm refers to the discharge at the Individual stations expressed as millimeters over the catchment area.

Table 4:13 Water Balance in the Narrator Brook Catchment: St. Cutt

Year ERF Qmm ERF le.mm le le Residue Residue % of Q % of Q % of ERF % of ERF (mm)

1975-76 1535 654 - 425 64.98 27.68 72.32 1110

1976-77 1407 1720 81.8 1314 76.39 93.39 6.61 93

1977-78 1303 1686 77.28 1258 74.61 95.52 4.48 58

1978-79 1364 1660 81.16 1177 70.90 86.29 13.71 187

3 Year 1358 1689 80.0 1250 73.96 92.0 8.00 113 Mean

The higher residue figure for 1976 for both stations in table 4:13 and table 4:14 reflects the fact that significant rainfall only occurred in the closing months of the water year 1975-76. Such large quantities of rainfall were experienced then, so any soil moisture deficit was quickly satisfied, and as the growing season was nearing its end, little moisture was required for vegetation demands, hence a large proportion was available for recharge. As 1975-76 was somewhat anomalous only the last three water years data have been used to derive the average figures illustrated in tables 4:13 and 4:14.

196 - Table 4:14 Water Balance in the Narrator Brook Catchment: St. 11

Year ERF Qmm ERF le.mm le le Residue Residue %ofQ %ofQ %of ERF %of ERF (mm)

1975-76 1535 622 - 403 64.79 26.25 73.75 1132

1976-77 1407 1638 85.89 1251 76.37 88.91 11.09 156

1977-78 1303 1606 81.13 1198 74.59 91.94 8.06 105

1978-79 1364 1581 86.27 1120 70.84 82.11 17.89 244

3 Year 1358 1608.3 84.43 1189.67 73.93 87.65 12.34 168 Mean

A comparison of tables 4:13 and 4:14 again illustrates that the lower and upper reaches of the Narrator Brook Catchment behave in quite different ways. This distinction is reiterated in both the analysis of streamflow to determine hydraulic continuity and the flow duration curve analysis undertaken in sections 4:3 and 4:4 respectively. At Station 11 a greater average percentage (84.0%) of effective rainfall contributes to stream discharge than at Station Cutt (80.0%) This is liable to be caused by overland flow processes and direct rainfall onto larger expanses of wet areas. Shallow interflow in the weathered areas and pipe flow in the fractured granite may also contribute a larger percentage of the effective rainfall rapidly to the stream channel. At Station Cutt the potential for overland flow processes are reduced and since infiltration appears to be greater the baseflow component is a larger proportion of flow in the lower catchment. These high percentages of effective rainfall contributing to stream discharge reflects the fact that in areas of weathered granite aquifers the condition of the surface horizons may detract from recharge reaching the deeper groundwater locations. The effective Infiltration appears to be less in the higher regions of the catchment, which may be a direct result of the more widespread development of ironpan horizons in the soil. The weathered granite cover may be shallower, and because of ironpan and fragipan

197 horizons, less permeable than the lower catchment materials. The efficiency of the effective Infiltration in the locality of Station Cutt may be greatly aided by the root systems of the forest areas. Trees facilitate water movement down towards groundwater by forming flow channels contiguous with the tree roots. Rough grazing lands higher up in the catchment are subjected to intermittent burning, a practice which may improve the quality of the pasture but can be detrimental with regards to potential infiltration. The burning of peat areas alters the loose textured surface areas into a substance which is nearly amorphous and highly resistant to water movement through it, (Conway and Miller 1960). As a consequence this will further reduce groundwater recharge In some areas. The average residue at Station Cutt (113mm) is less than at Station 11 (168mm). As a result there is a larger proportion of infiltrating water to contribute to underflow and the feeding of fissure-systems at Station 11. Because effective Infiltration into the weathered materials Is greater at Station Cutt and consequently forms a larger proportion of baseflow, less recharge is available for deeper percolation and underflow components. On average effective rainfall constitutes 82% of stream discharge in the catchment. Baseflow represents 90% of the effective rainfall, and 73% of stream discharge in the channel is from groundwater sources* A residue of some 10% of effective rainfall in a given year is available for deeper groundwater recharge (which is equivalent to 557x10^ m^ year"^). Obviously these values vary from year to year dependent upon catchment antecedent conditions. The fact that on average 82% of effective rainfall available for infiltration in any given year, finds its way to becoming a component of streamflow, suggests a rapid overall catchment throughput time. This is corroborated by estimates of groundwater velocities (2.5x10"** m sec"M and travel times (3 months) through selected flowpaths in the Narrator Valley, as outlined in Chapter 2. 4:5 Conclusions

From the foregoing presentation of streamflow characteristics, it has been shown conclusively that there is a significant groundwater component in the Narrator Brook. Influent and effluent discharges in the catchment account for the variations in flow volume

198 observed between the stream discharge measuring stations and the gauging sites. In addition underflow at the stations may approach an average value of some 5% of the discharge in the channel at any point In time. These figures however, are tentative due to the gross assumptions made in their estimation. The significanctof groundwater discharge in the Narrator Brook Catchment is well illustrated by the hydrograph analysis, baseflow recession and flow duration curve techniques utilised. Contributions to streamflow from groundwater sources constitutes on average 73% of total stream discharge. Baseflow recession in the Narrator Brook, derived from four water years data, can be greater than 200 days in length, showing substantial groundwater contribution from the weathered granite aquifer. From a consideration of a gross water balance some 557x10^ m^ year"^ appears to be available in an average year for recharge in the Narrator Brook Catchment. Flow duration analysis also suggests from the low flow indices, that discharge varies little indicating a more or less constant source or sources over a water year in the catchment.. Many of the stream hydrograph techniques frequently used in the hydrologlc literature, tend to overlook the variable components of • discharge and their tendency to overlap each other in volume and time. Techniques employed such as the Barnes (1939) method rely solely on attributing discharge at any point in time to three distinct components of the discharge hydrograph. This categoric distinction between individual stream components based on the shape of the discharge hydrograph in the case of the Narrator Brook is unrealistic. The Barnes (1939) technique is used in the Narrator Brook to broadly isolate surface discharge (channel precipitation, overland flow) from subsurface (bank storage, interflow and groundwater flow) discharge, and no other distinction into the contribution of the various streamflow components is undertaken. Separation of the hydrograph to determine baseflow, groundwater recession and flow duration analysis was carried out at both Station 11 and Station Cutt. It was considered, due to previous observations on streamflow in the catchment, that this two-fold analysis was warranted, and that the two sections of the catchment upstream from the gauging stations exhibited differing characteristics. This proved

199 to be the case and is amply demonstrated by the groundwater recession, storage values, and variations In Influent and effluent sections of the lower stream channel. The two quite different responses between the upstream and downstream sections of the valley are important features with respect to the assessment of the potential of the weathered granite aquifer of the Narrator Brook Catchment. Such features have great hydrogeological significance, and may be a widespread occurrence in other weathered granite areas.

Igneous bedrocks are not usually considered to make significant groundwater contributions to river flow, but in the case of the Narrator Brook Catchment several features persist which enhance the water-bearing properties of a granite based catchment. Dartmoor In general, and the Narrator Brook in particular, have extensive deposits of weathered granite insltu, overlain by subsequent perlglaclal deposits and river alluvium. These con^boe to provide small isolated but locally significant aquifers and sustain stream discharge.

200 Chapter 5 Groundwater Fluctuations In the Narrator Brook Catchment

5:1 Introduction This chapter considers groundwater level fluctuations In the Narrator Brook Catchment, their response to precipitation Inputs» and their magnitude In relation to variations In hydrogeologlcal proper• ties of the weathered material of the Narrator Valley* Groundwater fluctuations observed during the research period are analysed with the aim of Identifying and characterising the response of the Incoherent granitic materials of the valley aquifer, to Inputs. Statistical approaches are examined and their usefulness discussed In an attempt to Isolate the most Important parameters Influencing water level move• ments In the Narrator Brook Valley, Groundwater contour maps are produced and contrlbutary parameters to the overall groundwater balance, such as sprlngflow and soil moisture content are outlined. 5:1:2 Controls on Watertable Fluctuations The watertable Is the surface of a water body In saturated rock which Is constantly adjusting Itself towards an equilibrium condition. Few groundwater basins have uniform recharge conditions. As a result of Intermittent recharge, mounds and ridges form In the watertable under areas of greatest recharge. Superimposed upon these conditions are variations In the permeability of aquifer materials. The Influence of Impermeable strata, streams, wells and lakes In a region result In variations In the watertable elevation. In the case of the Narrator Brook Valley textural variations In the weathered materials along with Iron-pan development and the presence of discontinuous clay lenses affect both the rate of recharge and the magnitude of groundwater level fluctuations. Several factors have a strong Influence on natural groundwater levels; these are rainfall, evapo- transplratlon, vegetative cover, soil type, depth to the watertable and aquifer properties (Phillips 1978). 5:1:3 Meteorological Factors The position of the watertable Is dependent on the continued accretion to groundwater by recharge through a zone of Intermittent saturation. Seasonal variations in precipitation (modified by

201 evapotransplratlon processes) are transmitted through this zone and give rise to seasonable watertable fluctuations. In many areas of the UK watertable fluctuations tend to follow a seasonal pattern with high water-levels occurring during the winter months and low levels during the summer. This Is also evident from the water-levels observed in the Narrator Brook Valley as outlined in Section 5:3:1. 5:1:4 Topographic Factors A number of authors (Buchan 1966; Binnie and Partners 1971; Howcroft 1977) have noted that the watertable tends to be a subdued reflection of surface topography. Assuming a similar amount of infiltration from rainfall over both high and low ground, the variations in the amplitude of the watertable surface depends largely upon the texture of the material comprising the upper layers of the aquifer. In the case of a very open-textured rock, such as a coarse sandstone, groundwater will move through at such a rate that it Is able to rapidly find Its own level, and thus form a more or less horizontal surface. In close textured material, or aquifers with varying ranges of permeability, groundwater movement will be slower, so that water will still be draining towards the valley bottom from beneath higher ground, when additional infiltration from subsequent precipitation occurs. As a consequence the height of the watertable is built up in the higher areas of the catchment. This tendency is magnified by the fact that precipitation normally increases with relief. The topographic location of a well in which water-level fluc• tuations are monitored Is then an important factor. In a valley bottom the more rapid lateral movement of groundwater will tend to flatten out the seasonal rise and fall of a watertable (Smith 1972). The effects of micro-topography also result in a contrast between the movement of the watertable in wells located In the bottoms of depres• sions, and those in wells sited at higher elevations. Ward (1963) during hydrological investigation of the Thames flood plain, proposed that accumulation of surface and sub-surface runoff in the depressions after heavy rainfall ensures a higher moisture content, and thereby promotes percolation of water towards the watertable. This situation promotes the rapid response of the watertable to rainfall, but also means that a larger amount of water is available for percolation in

202 - the depressions than in higher areas. This appears to be the situation in the lower areas of the Narrator Brook Valley in the vicinity of the stream channel; the shallow wells 3 and 4 maintain a more or less constant level throughout the year suggesting sustained infiltration in comparison to topographically higher wells all of which exhibit clear weekly and seasonal variations in water-level amplitude (section 5:3). In the lower regions of the catchment the continual presence of surface water ponded above more impermeable layers is marked by marshy areas (as illustrated in Chapter 2 figure 2:18). This accumulation ensures a higher soil moisture content in the soil layers which facilitates percolation and maintains a constant but shallow water- table elevation. 5:1:4 Soil and Infiltration Factors Infiltration to groundwater occurs when the overlying soil Is at field capacity, as is usually the case during winter months in this country. When evapotranspiratlon starts to exceed rainfall during the spring, a soil moisture deficit is built up. As a rule any rain which falls during summer months replenishes soil moisture and Is consumed by evapotranspiratlon requirements so percolation to groundwater effectively ceases. During the autumn., rainfall starts to exceed evapotranspiration and infiltration Increases. This is an idealised case which, with the exception of drought years, rarely occurs in this country. Some water may percolate through the soil towards the watertable even if there is a soil moisture deficit, as has been demonstrated by Coleman and Farrar (1966). Such water movement occurs through fissures in the soil profile, and once the fissure-surfaces become wetted, water movement is facilitated as frictional forces are reduced, and subse• quently effective infiltration may occur before soil moisture deficits are made up. The blocky structures of the Brown Earths/Brown Podsols in the Narrator Brook Valley (Chapter 2) may be favourable locations for this process. Bonell (1978) reported that wells intersecting clay, sand and gravel lenses in the Holderness Catchment Humberside, responded prematurely to rainfall, well before the calculated soil moisture deficit had been erased. This was attributed to infiltrating water by-passing areas of soil moisture deficit via cracks, or to

203 - downward seepages of rainfall from a clay lens which effectively Increases the water-level height in a well. 5:1:5 Hydrogeological Factors The total seasonal range of restwater levels is determined partially by the composition of the aquifer, and by the topographic location of the well. When watertables are at shallow depths then annual and long term fluctuations are insignificant. Conversely the amplitude Increases with increasing thickness of the unsaturated zone, (UNESCO 1975). Larger fluctuations occur in fine-grained material, than in coarse-grained rock for an equivalent amount of percolation, (Smith 1972). The size of the annual amplitude of groundwater levels in relation to lithologlcal controls can be illustrated by comparing groundwater fluctuations for four aquifers in the UK during the water year 1971-1972, as tabulated in table 5:1 (data collated from the Groundwater in the UK Year 1971-72).

Table 5:1 Annual Groundwater Level Fluctuations In Four Aquifers for the Water Year 1971-72

Aquifer Region Range Annual Amplitude

Chalk Narborough 14 - 28m 14m Norfolk

Chalk Dolton Holme, 14 - 22m 8m Humberside

Bunter St. Georges Hill 79 - 85m 6m (Breccia) , Devon

Millstone Big Moor, 250.5 - 251.5m Im Grit Holmesfield

In general, as can be seen from table 5:1 the finer-textured chalks show the largest fluctuations, and the more permeable grits and Breccias show the smallest change. This assumes that in chalk and similar materials 'piston-flow' or displacement flow predominates while intergranular flow occurs in the sandstones and grits. Regional variations in the amplitude of water-levels In the same geological

- 204 formation would suggest however that fissure-flow mechanisms may predominate in these lithologies as well (Fox, Rushton 1976). In the Narrator Brook Valley the wells in clay areas show the largest fluctuation range, while those In sandy deposits show a smaller water-level amplitude. This is further discussed In section 5:3:2. 5:1:6 Miscellaneous Factors Barometric Pressure Changes in atmospheric pressure produce sizeable fluctuations in wells penetrating confined aquifers, and have little effect on water- table aquifers (neglecting entrapped air below the watertable), (Chow 1964). Peck (1960) outlined a quantitative theory relating the position of a watertable to atmospheric pressure, which suggests that when air is entrapped in water in an aquifer the watertable height will vary with atmospheric pressure. Peck (1960) predicted that the watertable will fall when the external air pressure increases, and the maximum rate of change of watertable height with air pressure occurs when the watertable is at the surface of the soil. Turk (1975), investigating shallow watertable movements in the Bonneville Salt Flats, Utah, compared continuous hydrographs of well fluctuations with barometric pressure graphs, and deduced that temper• ature changes reinforced by barometric pressure changes gave rise to diurnal watertable fluctuations of 1.5 to 6cm in the Salt Flats. According to Turk (1975), many of the small-scale fluctuations of shallow watertables that have been documented in the literature, eg. (Peck 1960, Headworth 1972), are responses to atmospheric pressure changes. These effects are most apparent In finer-grained aquifer materials and probably are of very small magnitude in sand or gravel aquifers which have a small capillary fringe, and are probably smaller than typical field measurements would detect (Turk 1975). Any day to day atmospheric pressure changes would require contin• uous monitoring with a well-recorder to assess its effects. In the case of the Narrator Brook Catchment only weekly spot measurements of groundwater levels were taken and as such were deemed unsuitable to ascertain effects of atmospheric pressure changes. For the purposes of this present investigation changes in atmospheric pressure were

205 - assumed to have minimal effect on the watertable fluctuations in the aquifer. Loading Effects Water-levels In an aquifer may be affected by fluctuations In the level of an adjacent surface water body, because of a loading effect, or because of hydraulic connection between the aquifer and surface water (Van Everdingen 1968). A load applied at the surface will be transmitted partly to the water in the aquifer. Thus a rise in the stage of the surface water produces a rise in head level, and a drop in surface stage produces a drop in head level. This effect has been taken into consideration in this investigation since fluctuations in Burrator reservoir stage levels may be transmitted to nearby wells, as it is assumed that the weathered materials are in hydraulic continuity with the water of the reservoir. 5:1:7 Air Entrapment during Groundwater Recharge Some workers have observed an anomalously large rise in water- levels in observation wells in shallow unconfined aquifers during heavy rainstorms (Kruel and Liefrinlck 1946). It is now recognised by many investigators that this type of water-level fluctuation is the results of air entrapment in the unsaturated zone (Bianchi and Haskell 1966, Klkkawa and Kawanlshi 1967, Bonell 1978). If the rainfall is sufficiently intense, an inverted zone of saturation is created at the ground surface, and the advancing wet fronts traps air between itself and the watertable. Air pressures in this zone may build up to values much greater than atmospheric. An increase in entrapped air pressure leads to a rise in water-levels in an observation well open to the atmosphere. This type of fluctuation bears no relation to groundwater recharge, but, because it is associated with rainfall events, can be easily mistaken for recharge. The most diagnostic feature of this entrapped air effect is the magnitude of the ratio of water-level rise to rainfall depth. Godwin (1931) reported a range of values of 7:1 to 12:1, while Meyboom (1967) reported values as high as 20:1. The anomalous rise usually dissi• pates within a few hours, or at the most a few days, owing to the lateral escape of entrapped air to the atmosphere outside the area of surface saturation.

206 - Ratios of water-level rise to rainfall depth for the Narrator Brook Valley were determined for 31st December 1979, a period of intense rainfall in the Catchment. These are illustrated in table 5:2. Table 5:2 Ratio of Water-level Rise: Rainfall Depth

Wells 1,2,7,12,13 3.05 : 1 Wells 9.14,15 3.58 1 Wells 3.4 0.32 1 Wells 6,10,11 1.23 1

The generally small magnitude of the responses suggest that changes in the water-levels in the Narrator Brook Valley may be attributed to changes in infiltration and storage, and are unlikely to be caused by air entrapment. 5:1:8 Effect of the Unsaturated Zone or Watertable Fluctuations In some areas a lack of coincidence between water-levels and rainfall suggest that there is a significant time lag in the rates at which deficits are satisfied before the water-level can rise. Such variations may be caused by the interaction of vegetatlonal, hydro- geological and meteorological parameters. When the watertable is only a few meters below the land surface, and where the unsaturated zone is moderately permeable, only a few days are required for water to reach the watertable. For watertables Im below ground level in sand and gravel deposits in the Thames Flood Plain, Ward (1962) quoted percola• tion rates from 24 hours to 2 days. Bonell (1978) noted that in observation wells in a boulder clay catchment, the lag time was a function of watertable depth. The greater the depth of the watertable below the ground surface the longer the time lag between water-level fluctuations and rainfall input response. Where the depth to the watertable Is great (15-30m), Heath and Trainer (1968) suggested that several months to a year or more may be required for all the water to reach the watertable. In chalk and limestone aquifers fissure flow gives rise to rapid recharge, and hence large plezometrlc head variations which can be in excess of 10m over 6 months of a water year (Fox and Rushton 1976).

- 207 Atkinson and Smith (1974) noted that groundwater flow through fissures and swallow holes in the chalk of South Hampshire, can be in excess of 2 km per day. As precipitation is discontinuous, infiltration of water at the land's surface is also discontinuous. Precipitation inputs may be visualised as occurring in 'slugs' which are Irregular. Water in such slugs follow diverse paths through the unsaturated zone. In areas where the depth to the watertable Is relatively the same throughout the catchment area, and by inference the aquifer is isotropic, the flow paths from the land's surface to the watertable are all about the same length. In such areas water which left the land surface as a slug arrives at the watertable as a slug, although probably somewhat attentuated. A more complicated situation exists in areas where the aquifer is anisotropic and the depth to the watertable Is consequently highly variable within short distances. A new slug of infiltrating water may reach the watertable in the valley area before the preceding slug reached the deeper watertable under the hillside. This portrayal takes no account of vertical or areal variations in permeability, nor considers lateral-flow conditions but may account for lag times recorded in some observations wells in catchments possessing a variety of geologic materials. This model has been utilised in this investigations along with chemical variations in the groundwaters with some success, as outlined in Chapter 7. 5:2 Contributary Factors to the Groundwater Regime in the Narrator Brook Catchment General soli moisture conditions in the Narrator Brook Catchment, are considered to have an effect on the rate of recharge to the water- table. In view of this, soil moisture content, as measured in the Narrator Brook Valley, merits further attention. Saturation upwards from an iron-pan or a relatively impermeable textural horizon in the soil and weathered granite matrix, may give rise to seeps or springs in the Narrator Brook Catchment which on aggregate contribute to the groundwater flow regime in weathered granite areas. 5:2:1 Soil Moisture Conditions Soil water conditions exert a strong influence on the acceptance of rain for infiltration. A watertable at shallow depth reduces the soils ability to accept heavy rain simply because the storage space

208 - above the watertable is restricted. Percolating water in such an instance has a short pathway to a saturated horizon, and consequently surface runoff is rapid in these regions (Thomasson 1978). With a greater depth to the watertable the pathway to the unsaturated zone is greater, and vertical movement of soil water is likely to predominate. However, the physical properties of the soil are important controls on moisture movement at a site, (Whipkey & Kirkby 1978). If the texture is coarse, with a predominance of sand and stones, vertical flow predominates and when the soil is deep, sub• surface flow response may be delayed. If the texture is fine, resistance to vertical flow results and lateral or shallow sub-surface flows sometimes occurs quickly. The soil moisture content of the two sites as measured in the Narrator Valley throughout a 14 months monitoring period are characterised by marked shallow-surface changes on a week to week basis (figure 5:1 and 5:2). The marked variation in the surface layers of the soil at Site 1 and Site 9 are dependent upon antecedent precipitation conditions. This weekly effect tends to be counteracted at depth by an Increment to the soil moisture deficit from following rainfall periods. Below a depth of 50cm, the range in fluctuation of soil moisture content down the profiles at both sites is noticeably less. This may reflect similarities in the structure and disposition of varying textural horizons down the profile at both sites, and as a consequence the moisture content becomes constant with depth. At Site 1, as further illustrated by the two typical soil moisture profiles with depth at Site 1 and Site 9 (figure 5:3), the profile shows a drop to a more or less constant level, of 20-30%, at depths greater than 50cm below the surface. This is apparent even during the drier months indicating little overall change in the moisture content with the soil profile at Site 1 over long-term intervals. At Site 9 however, greater variation within the soil moisture profile determined on a weekly basis was apparent. This is illus• trated in figure 5:2. Figure 5:3 depicts the moisture profile to depths of llOcms and 90cms. The primary controls on such variability are thought to be related to the position on the slope; the soil type and the soil aspect. Newson (1976), discussing the Plynllmon

209 - FIGURE 5:1 SOIL MOISTURE VARIATIONS WITH DEPTH SITE 1

Sampio 40 >. Depth cms

10cm

•tOcm

60cm

80 cm

IGSO WITH DEPTH Somple Depth FIGURE 5i2 SOIL MOISTURE VARIATIONS cms. SITE 9 lOcm

20cm

lOOcm

1979 S/7£ ;

I 0 10 20 30 40 50 60 70

Moisture Content Z

0 SITE 9

10 X

20

30

40

50

60

70

I 80

90 0 10 20 30 40 5 0 60 70 Moisture Content %

FIGURE 5 3 Typical Soil Moisture Profiles in the Narrator Brook as Determined by the Gravimetric Method - 'X\^ - Catchment in Wales, mentioned that there is little correlation between slope position and the range in soil moisture changes, but that the effects of mlcrotopography and soil variability grossly exceed larger scale systematic effects. The soil at Site 9 had a cover of organic material which is thought to be capable of absorbing a large proportion of incoming precipitation, and to release this slowly to the underlying soil layers.with time. If the hydraulic properties of the soil profile at Site 9 were uniform, this would result in a more even distribution of soil moisture down the profile. From figure 5:3, the non-uniform conditions in the soil profile at Site 9 are readily apparent. Averaged soil moisture values over the two sets of profiles for Site 1 and Site 9, show that Site 9 tends to retain moisture longer In the profile than at Site 1, suggesting that Site 1 has more uniform soil properties. Whilst soil moisture determinations indicate the moisture contents of a soil at a particular location for a given time, they do not provide continuity of measurement with regards to determining moisture movement through a soil profile. Tenslometer measurements indicate moisture tension gradients in the soil and hence the direction of water movement. Water in unsaturated zones is held in the soil-pores under surface tension forces or 'soil-suction'. This suction characterises the action of water retention by soil and regolith materials (Richards 1965). Hydraulic head measurements above a watertable are obtainable through the use of tensiometers which measure the pressure head in a soil at a specific depth. The choice of the soil suface as a zero reference in effect makes all hydraulic head readings on a tenslometer negative, and consequently the direction of moisture movement is from smaller to larger absolute numbers, (Richard et al. 1973). The three months data from tenslometer readings in the Narrator Brook Catchment show that soil moisture movement for the most part is down towards the watertable.

- 213 - At Site 10, where four tenslometers were located in a 'nest', tenslometer readings during the three months showed a downward movement of water from the surface towards the tensiometer cup at a depth of 45cm, but an upward movement from 60cm to 45cm. Water movement was downwards from 60cm - 105cm, at the deepest part of the 'nest'. This variation can be acounted for by the distribution of clay and sand constituents in this region, and it Is likely that the entire soil profile in the Narrator Valley shows a heterogeneity akin to the underlying weathered granite materials in the aquifer. 5:2:2 Spring Discharge Springs and seeps are amongst the hydrogeological features that may aid in evaluating the groundwater potential of certain areas. A few large springs may Indicate a thick transmlsslve aquifer, whereas frequent small springs tend to indicate thin aquifers of low trans- mlssivlty (Bouwer 1978). The principal variables which determine sprlngflow are aquifer permeability, contributary recharge area, and the quantity of recharge. Many aquifers have considerable discharge in the form of springs, but the permeability of some aquifers with depth is so low that water is forced to the surface over a large area (Davis and De Welst 1966). The development of resources in such areas would be inappropriate since the potential for deep groundwater supplies are restricted by the low permeability of the formations with depth. In fractured granitic rocks, fissures can fill with rainwater which then flows through the same fissure system to form spring developments at lower topographic locations. Saturation upwards from an iron-pan or relatively impermeable textural horizon, may give rise to small perched waterbodies in the soil and weathered granite horizons in the Narrator Brook Catchment. The flow rate from springs depends on the size of their recharge area, the amount of rainfall and the transmlssivlty of materials in the neighbourhood. Springs which only flow during and immediately after rainfall events have been observed in the groundwater observation network area during the research period. This type of Issue is ephemeral in nature (as defined in Chapter 4), and the duration of flow la from a few hours up to four days. The position in the groundwater observation network of these ephemeral issues are illustrated in figure 2:17 Chapter 2.

214 - Perennial springs which discharge throughout the year usually drain extensive permeable zones, whereas intermittent issues discharge only during portions of the year when sufficient groundwater is recharged to the malnflow zones. The headwaters of the Narrator Brook Is a perennial spring and parts of the Sheepstor Beck are fed by Intermittent springs as previously outlined in Chapter 4. The chemical parameters of two smaller perennial springs in the Narrator Valley groundwater observation network area, were measured weekly during this investigation, but only the flow of one of these springs (Spring 12/13) was monitored by a 'V* notch. The analysis of this springflow forms the basis of this discussion, although it must be pointed out that this spring is one of at least 30 springs in the Narrator Brook Catchment. Ternan and Williams (1979) noted the presence of 23 springs in the Narrator Brook Valley, and seven are recorded at the downstream end of the catchment (by the author in Chapter 2 figure 2:17). Most springs show measurable fluctuations of discharge in response to seasonal fluctuations in precipitation. The measured spring discharge of Spring 12/13 in the Narrator Valley is no exception. Correlation of springflow discharge with rainfall measured two weeks prior to the discharge measurement, produced a correlation coefficient of 0.A66, significant at the 0.01 level. This suggests that Spring 12/13 has a two-week response period. Discharge In springs during periods of no or little recharge can be expressed by an equation given by Brown et al. (1975). A laminar flow is assumed: ^ e - at Qt = Qo where Qt = discharge at time t Qo = discharge at time o a = recession constant obtained by plotting Q vs t on semi• log paper as shown in figure 5:4. From the above equation: log Qo - log Qt ^ ^ 0.4343 t

- 215 FIGURE 5:4 Recession Constant for Spring 12/13

0']

Q)

0-01- August September Ju/y

Time in Days

2th " For Spring 12/13, inspection of the discharge and rainfall data allowed the choice of a relatively dry period in July and August 1979. The observed discharge was plotted against time to derive the recession constant a ^ 0.020, as illustrated in figure 5:4. A regression analysis was carried out to fit the line to the data plot more accurately giving an r^ (coefficient of determination) value of 0.78. By converting this ratio to a percentage, it can be said that 78% of the variance of the dependent variable, spring discharge, is accounted for by the regression. However due to the small amount of data pairs available for this analysis the regression is not statistically significant. Strongly fractured limestones often show a values in the range of 0.0025-0.05, while sandstones with minor fractures gives values of 0.001-0.0024, (Englund and Meyer 1980). With a for the spring in the Narrator Valley the time needed for reducing the recharge rate Qo = 0.99 1 sec~^ to an arbitrary value of 0.1 1 sec^^ can be obtained by substitution into equation (1) above. This method as outlined by Englund and Meyer (1980), gives a time of 102 days (14i weeks) during which this spring will flow without recharge, and discharge will recede to 0.1 1 sec"^. This limited analysis gives a general idea as to the constancy of the spring's groundwater resources on the south side of the Narrator Brook Valley. Spring 12/13 Issues from an old mine adit whose sub• surface extent is unknown, and in addition It may well be fed from sources in the fractured granite bedrock at depth. Water stored in the weathered granite area of the catchment in the vicinity of Spring 12/13, in association with that draining mine workings and contri• butions via fractured granite, may be responsible for the variability seen in the springflow discharges at this location. If the ground• water feeding Spring 12/13 is solely from the weathered granite aquifer in the near vicinity, the length of recession indicates a considerable stored volume of water held in the aquifer at this location. Discharge from Spring 12/13 may be visualised as a constant slow draining of varying thicknesses of weathered materials on the south side of the valley, whose flow is supplemented in high rainfall periods as a response to recharge from neighbouring parts of the

217 - granite catchment. The rapid responses of sprlngflow to rainfall input, within two weeks, is within the 1-4 weeks response time of water-levels in the observation wells in the catchment as outlined in section 5:4. This analysis of springflow variations in response to rainfall inputs in the Narrator Valley, provides an indication of the stability of groundwater resources in this southern part of the valley aquifer. However, conclusions concerning springflow volumes and variations in discharge in the Narrator Brook Valley, must be restricted by virtue of the limited amount of data collected. The characteristics of Spring 12/13 are not thought to be representative of all the springs of perennial, intermittent and emphemeral natures, present In the Narrator Brook Catchment, but they do illustrate the importance of groundwater contributions to the drainage system in weathered granite areas. 5:3 Analysis of Groundwater Data; Graphical Techniques 5:3:1 Well Hydrographs Water-level fluctuation data for the observation wells obtained during the research period are presented as well hydrographs in figures 5:5a - 5:5c. Generally the fluctuations follow a rhythmic seasonal pattern with high water-levels occurring during the winter months and lower levels during the summer. An exception to this general pattern is apparent for August 1979 when a peak in water- levels occurred due to heavy rainfall. This pattern is also exhibited by water-level fluctuations in the Bunter sandstone aquifer of East Devon for the same time period. Table 5:3 tabulates the highest and lowest waterlevels recorded at each site.

218 Table 5:3 Extremes in Water-level Data In the Narrator Brook

Well Highest Date Lowest Date Range (m) Water Level Water Level

1 10.02 31.12.79 12.90 6. 8.79 2.88

2 4.19 31.12.79 6.81 7. 8.69 2.62

3 +0.03 31.12.79 0.31 22.10.77 0.34

4 0.11 19. 3.79 0.37 22.10.77 0.26

6 0.26 17.12.79 1.01 24. 7.79 0.75

7 3.59 31.12.79 6.46 13. 8.79 2.87

9 0.77 31.12.79 4.81 7. 8.79 4.04

10 2.31 31.12.79 3.58 6. 8.79 1.27

11 +0.05 31.12.79 0.96 7. 8.79 0.91

12 0.65 31.12.79 3.67 13. 8.79 3.02

13 4.29 18. 2.80 6.18 7. 8.79 1.89

14 3.59 31.12.80 7.39 17. 8.79 3.80

15 1.72 31.12.80 7.09 13. 8.79 5.37

The dates of observation of the highest water-levels correspond to the winter months when rainfall and hence recharge are expected to be higher In the absence of soil moisture deficit and high evapo- transpiration rates. In contrast the lowest water-levels are recorded towards the end of the water year when evapotransplration rates are higher due to the growing season, and recharge rates are lower or absent. This further demonstrates that groundwater behaves In accordance to seasonal trends of inputs and deficits In the catchment.

219 - Site No.

I * o 0)

HO-5 i i! HOO

Ho

j I9tt0 1 D J IV79

FIGURE 5:5a Groundwater Level Fluctuations 1978-1980 Site No

I 3

4

D 71979 ' f • M ' M ' jUN J IV80

FIGURE &5b Groundwater Level Fluctuations 1978-1980 Site No

1

I Q

FIGURE 5:5c Groundwater Level Fluctuations 1978-1980 The wells in the observation network may be divided into groups based on a visual assessment of the variation in amplitude of the well hydrographs, and these are as follows:

Group A 3,A,6,11 Group B 1.2.7.10.12.13 Group C 9.14.15

Group A is composed of wells with very little annual variation; Group B with intermediate amplitude; and Group C those wells with a large range in watertable fluctuations. This preliminary grouping can be attributed to topographic locations, those of Group A in the valley bottom. Group B in middle to upper slope positions and Group C wells in the higher reaches of the groundwater study area* This distribution is illustrated in figure 5:8 section 5:3:2, and other group features are described in this section. 5:3:2 Groundwater Contour Maps The groundwater contour maps of the southwest portion of the Narrator Valley are based on information collected during July 1979 (figure 5:6) and December 1979 (figure 5:7). These are representative periods illustrating the driest month (July) and the wettest month (December). Although the watertable rises and falls seasonally as illustrated by the hydrographs in figures 5:5a - 5:5c, the groundwater contours preserve their general form throughout the year as indicated in figures 5:6 and 5:7. Figure 5:6 and figure 5:7 were superimposed, and the waterlevel changes at the contour intersections were transferred onto a third map. figure 5:8, on which lines of equal water-level change were drawn in. Figure 5:8 illustrates features regarding the aquifer responses which are difficult to detect from a comparison of successive water- level maps. Two zones of very low change are apparent; (1) the area around the reservoir and (11) areas on the higher valley sides. An intermediate belt at the mid-slope position in the vicinity of wells 9, 14 and 15 show the most marked fluctuations of up to 5m.

223 0 22d-6

® 223-3

® 229-9

® 233 0

242-2

-245 _ _

KEY ©222 WeiUond height of R.WL.aboveQD

"230^ Groundwater Contours metres

FIGURE 5:6 GROUNDWATER CONTOURS NARRATOR VALLEY JULY 1979 ©229'*

230 —

232-3

236-3

2*27

'723 Weils and heighf of RWL. above QD.

220 ^ Groundwater Contours metres

FIGURE 5:7 GROUNDWATER CONTOURS NARRATOR VALLEY DECEMBER 1979

-225 - KEY (J) Welh

Change in /eve/ contour metres

/nter^ection of j^o July and Decembflfj contours metres

125 m

FIGURE 5:8 CHANGE IN GROUNDWATER LEVELS NARRATOR VALLEY 1979 As groundwater moves downslope towards the reservoir the Input is more or less constant and as a result levels do not vary greatly from 0-lm. Hydraulic continuity is assumed between the reservoir and the weathered materials and variations in the reservoir level may moderate the range of fluctuations experienced in nearby wells. Contributions from springs and seeps to the north of the reservoir which drain southwards towards the reservoir basin, may also have a moderating influence by evenly distributing recharge to the lower valley area and hence diminish the range of water-level fluctuations experienced. Groundwater movement downslope from the higher areas of the catchment may feed the system in discrete units which bypass wells at higher topographic locations, to coalesce towards the middle and base of slopes* This is indicated by the larger variations In the water- level change map at Sites 9, 12, 14 and 15. Drainage towards the reservoir and in a lateral direction parallel to the Narrator Brook may be Inferred from the trough in the water-level change map (figure 5:8). Spring 12/13 near Site 12 reaches the surface in this trough area, and discharges in this direction too. This region of greatest water-level change is In the locality where observation wells penetrate notable clay lenses, which are regarded as the chief influences determining the route and direction of drainage downslope. In general, fine-grained clay materials give rise to larger amplitudes In groundwater levels, wells 9, 14 and 15, whereas coarser materials around the well screens, as in the case of wells 2, 3, 4, 6 and 10, results in smaller amplitude variations. The fact that the change in groundwater contour levels do not vary greatly on the higher slopes may also indicate that recharge is fairly constant, or that the higher hydraulic conductivity of the aquifer material allows more rapid dissipation of recharge volumes. Over the majority of the rest of the topographically lower groundwater observation network in the Narrator Brook Valley, fluctuations are between 0-3m annually. Based on the change in water-level map figure 5:8 the wells can be assigned to the following three groups:

227 Group Wells Range of Water-level Change (m) a 3,4,6,11 < 1 b 1,2,7,10,12,13 l-<3 c 9,14,15 3-5 These groups correspond exactly with the previous division based on a visual assessment of well hydrographs. These variations in the annual watertable amplitude in the Narrator Brook Valley, reflect a feature which Uhl et^ al^. (1979) had observed in weathered granite aquifer, in the Satpura Hills, India. Here, watertable depths had an annual range of generally less than l-15m and water-level fluctuations were found to be more pronounced in groundwater recharge areas (6-15m) than in groundwater discharge areas (0-3m). By comparison with the work of Uhl^et^. (1979), the higher regions of the Narrator Brook Catchment are recharge areas for the valley aquifer. 5:4 Analysis of Groundwater Data; Statistical Techniques 5:4:1 Standard Deviation of Water-level Fluctuations The main features of contrast between the water-levels in the observation wells in the Narrator Brook Catchment appear to be in the magnitude of response of the watertable to rainfall events. Based on these observations a simple but more quantitative analysis of the watertable fluctuations was carried out, utilising the standard deviation of the water-levels measured at each site over the research period. The distribution of the standard deviation of water-levels over the catchment area is illustrated in figure 5:9. Smallest values of standard deviation, indicating little variability in water-level fluc• tuations, are present in the valley bottom along the river channels and around the reservoir. Values of the standard deviation of water- levels increase progressively upslope. There is also a mid-slope zone of rapid change in the standard deviation of water-levels. The presence of larger variations in water-levels at higher elevations is contrary to the conclusions drawn In section 5:3 where the change in water-level maps suggested little variation in water-level fluctuations. This may be explained by taking into account the data utilised in the two methods; the change in waterlevel variations is measured by the difference between extreme conditions; while the distribution at the standard deviations of water-levels takes Into

- 228 account all the data collected. On this basis the change in water- level map, although it provides a means of comparison between two extreme events» is not truly representative of all the water-level fluctuations experienced. In view of this the standard deviation technique is a more useful form of analysis. The variations in water-levels in the observation wells may be interpreted as being the result of vertical and lateral movement of water towards the watertable through materials of varying hydraulic properties in the valley. Such features have been documented in the hydrogeological literature. Sloan (1970), whilst undertaking investigations into water-level fluctuations in Praire Potholes, and those water-level movements In nearby observation wells, noted features in the glacial outwash materials, which are relevant in the context of the Narrator Brook Valley. Water-levels in the potholes and those in nearby observation wells were often different by greater than 0.30m. Such differences are characteristic of glacial till and result from low permeability till material and the large hydraulic gradient necessary to move groundwater through it. Many of the test holes drilled into poorly permeable glacial till appeared to be dry, and the holes did not fill up quickly due to the comparatively slow release of water-from the material, (Sloan 1970). This was noted in the Narrator Valley during the drilling of Well 12 and Well 14. Well 14 remained 'dry' until after the well completion procedures were finished, in contrast to Well 12 where the watertable was intersected during the drilling process. Sloan (1970), noted that water-levels in closely spaced wells differed particularly if the wells were cased to different depths. The cased wells acted as piezometers, such that water-levels represent the fluid pressure at the bottom or screened part of the well. According to Sloan (1970) large vertical fluid pressures are common in glacial till because of its low permeability, and If such water-levels are misinterpreted as points on the watertable, there would seem to be a series of perched watertables.

229 - KEV 00 >b wells and Standard Deviations Standard Deviation Contours

\ \\® 0-79

125m

\ \

FIGURE 5:9 DISTRIBUTION OF STANDARD DEVIATIONS OF WATER LEVELS NARRATOR BROOK CATCHMENT

- 23,0 - In the case of the Narrator Brook Valley aquifer the borehole logs (Chapter 3) show a series of clay materials with intervening zones of higher permeability between the clay layers. The vertical and lateral discontinuity of these clay layers result in 'perched' watertables at different elevations. The observation wells Intersect a number of these clay layers and their associated 'perched' water- tables. The variation in water-level fluctuations in the Narrator Brook Catchment may therefore be attributed to the differing hydraulic properties, and hence response to input, of materials in the neigh• bourhood of the wells. 5:4:2 Bivariate Analysis The well hydrographs of the Narrator Valley are illustrated in figures 5:5a - 5:5c and are summarised along with total rainfall values over the observation period in figure 5:10. The water-level change through time represents a time series In which three components are present:- (i) the overall long-term trend (secular trend) (11) periodic fluctuations of a rhythmic nature associated with daily or seasonal variations (ill) irregular or random variations A way of reducing the irregular fluctuations, and thereby highlighting those that are regular, is by the use of moving averages or running means (Hammond and McCullagh 1978). A simple moving average smoothing method, Davis (1973), has been used for Narrator Valley water-level data, using three weeks as the smoothing interval. The resultant hydrographs are Illustrated in figure 5:10. By graphical Inspection of the well hydrographs in association with total rainfall recorded, (figure 5:10), it can be seen that on average an increase in rainfall is marked by a rise in the watertable lagged from between one and four weeks over the catchment. This time lag is likely to be influenced by antecedent conditions. As an approach to an assessment of groundwater recharge in the Narrator Brook Valley aquifer the degree of relationship between three pairs of variables was determined using a Pearson product-moment correlation technique. Groundwater levels were correlated with both the effective rainfall in the preceding seven days and soil moisture deficit.

231 FIGURE 5:10 Smoothed Well-Hydrographs and Rainfall Values in the Narrator Brook, 1978-1980

WELL RANGE (M. 1 12-91 • 10-85

6-94 50

13 6- 32 4.28 15 7- 72- 2-55

14 8 09--4-4 7 7-23-3-83

//V-

6 0-87- 0-29 4 0-26- 0-21 3 0-30- 0-07 11 0-90- .0-13 12 4- 28- 1-82 10 3-83-•2-49

9 5-25--1-51

60

50

40 c c 30 •<•

20- <: (5 Q: 10

0 •

7/me The Pearson product-moment correlation technique is the most powerful test of correlation, (Hammond and McCullagh 1978). It Is a parametric measure of the relationship between a pair of variables, which assumes that they both have normally distributed populations and are measured on the Interval scale. The parameters correlated in this investigation meet both these criteria, and the use of this more rigorous statistic is appropriate. Correlation coefficients obtained using the Pearson product-moment correlation coefficient are listed in table 5:4.

Table 5:4 Correlation Coefficients for Preceding 7 Days Rainfall vs Water-level, and Soil Moisture Deficit with Water-level in the Narrator Brook Catchment

Water-level Significance Soil Moisture Deficit Significance Well vs Rainfall Level vs Water-level Level

1 0.45 0.05 0.37 0.05

2 0.54 0.05 0.32 0.01

- 3 0.55. . ,0.05. . 0.44. . . . 0.05

4 0.14 Not Sign 0.10 Not Sign

6 0.04 Not Sign 0.55 0.05

7 0.30 0.01 0.14 Not Sign

9 0.35 0.05 0.36 0.05

10 0.36 0.05 0.36 0.05

11 0.57 0.05 0.48 0.05

12 0.36 0.05 0.33 0.01

13 0.22 Not Sign 0.28 0.01

14 0.33 0.05 0.20 Not Sign

15 0.39 0.05 0.33 0.05

With the exception of wells 4, 6 and 13 all wells showed a significant correlation either at the 0.05 level or the 0.01 level between groundwater levels and and the seven day antecedent rainfall

- 233 - (table 5:4). Similarly with the exception of wells 4, 7 and 14 all wells showed a significant correlation at either the 0.05 or the 0.01 level between groundwater level and the soil moisture deficit. Dudarenko (1981) in a 3 month mlcrostudy on factors responsible for water-level and groundwater temperature variations in the Narrator Brook Valleyy used a daily correlation technique and producted the following lag times between rainfall and watertable response for selected wells in the catchment:- Well Lag in Days 1 12 2 20 6 13 7 11 9 20 10 25 12 20 14 26 15 19 which illustrate that the average of a one month's lag period, as ascertained by graphical inspection, is of the right order of magnitude. 5:4:3 Analysis of Variance in the Narrator Brook An analysis of variance was carried out on the water-level fluctuation data for the Narrator Brook Catchment. The aim of this was to determine if the wells responded in a similar fashion to recharge events. Such a condition would be expected to occur In a single homogenous aquifer, whilst a significant difference in water- level fluctuations between wells would suggest a heterogeneous water• bearing formation. Essentially an analysis of variance procedure compares sample mean values between groups within a body of data, and is a comprehensive and powerful set of procedures based on the F-test devised by Fisher (1956). A null hypothesis (HQ) is erected that the sample means represent population means that are equal:-

= ^2 = This hypothesis assumes that all sample variances are equal, an important criteria which must be met otherwise the power of the test Is reduced. Till (1974) suggested that if the population variances are grossly unequal a non-parametric test should be used in place of an F-test.

234 - The standard deviation of water-levels and the variance deter• mined at each well site In the Narrator Brook Valley are illustrated In table 5:3* Homogeneity of variance was established from the body of data and three distinct well groupings possessing similar variances were Identified. The analysis of variance was carrried out on the three individual well groupings from table 5:5 and the results are tabulated in the analysis of variance tables 5;6 and 5:7.

Table 5:5 Variance and Standard Deviation for Wells in the Narrator Brook Catchment

Group Wells Variance Standard Deviation

1 0.57 0.76 2 0.31 0.56 7 0.77 0.88 1 12 0.88 0.93 13 0.60 0.77 10 0.15 0.38

11 0.06 0.25 6 0.03 0.18 2 3 0.007 0.08 4 0.003 0.06

9 1.37 1.17 3 14 1.34 1.16 15 2.18 1.47

The total variance is the variation of all measurements about their mean. The wlthin-samples variance is the total variation of each sample about its own mean groundwater level, and the between- samples variance is the variation of the sample means about the grand mean. The F-test looks at the ratio of between-samples variance to wlthin-samples variance. Intuitively if a number of means are not significantly different there will be as much variation within samples and between samples, giving a small value of F. Conversely if the means are significantly different the variation betwep samples should be greater than the variation within samples, and this gives a larger value of F.

- 235 The F-value calculated In all three cases for wells in the Narrator Brook Valley are greater than the F-values from statistical tables, so Ho (that all water-level means are equal) Is rejected, (because calculated values of F would arise far less than 1 out of 100 times If the samples all belonged to the same population). Therefore in this case the alternative hypothesis H| (that some water-level means are not equal) is accepted, as the water-levels in the wells are not the same at 1% confidence level.

Table 5:6 Analysis of Variance (One Way)

Degrees Sum of Mean F Source of Variation of Squares Square Value Decision Freedom

Between Wells 2 225.76 112.88 32.61 Reject HQ 9, 14, 15

Within Wells 46 159.22 3.461 9, 14, 15 F (0.01, ;I, 46) - 4.98

Between Wells 3 4.62 2.31 Reject HQ 3, 4, 6, 11

Within Wells 46 0.16 0.08 3, 4, 6, 11 F (0.01, 2, 46) = '4.9 8

Between Wells 5 2685.32 537.06 4.98 Reject Ho 1,2,7,10,12,13

Within Wells 46 97.62 2.12 1,2,7,10,12,13 F (0.01, 2, 46) = 4.98

A more complex 2-way analysis of variance was utilised in which the data was analysed by rows and columns in a block. This method attempts to determine if there is a significant difference between the wells (columns) and the water-level data (rows) recorded. Two null hypothesis were erected:- (1) Ho^ all wells have similar water-level fluctuations versus H some wells have different water-level fluctuations A (11) H0O'2 all individual well water-level fluctuations are 4 similar

- 236 versus all individual well water-level fluctuations are different Again the analysis was carried out on each of the three groups and the results are tabulated in table 5:7. For all well groups Bo^ and H02 were rejected at the 1% level. This Indicates that there is a significant difference between the wells and the water-levels experienced in these wells.

Table 5:7 Analysis of Variance (2-Way)

Degrees Sum of Mean F Source of Variation of Squares Square Value Decision Freedom

Water-levels (Rows) 46 159.22 3.46 37.136 Reject Hoj

Wells 9,14,15 2 225.76 112.88 1211.09 Reject Ho2 (Columns)

Error/Residual 92 8.57 0.093

Fi (0.01, 46, 92) = 1.66 F2 (0.01, 2, 92) = 4.79

Water-levels 46 3.73 0.08 7.49 Reject Hoi

Wells 3,4,6,11 3 4.62 2.31 213.51 Reject Hoi

Residual 92 0.99 0.01

Fi (0.01, 46, 92) = 1.66 F2 (0.01, 5, 92) = 3.95

Water-levels 46 97.62 2.12 33.88 Reject Hoi

Wells 1,2,7,10, 5 2685.31 53.06 42868.76 Reject H02 12,13

Residual 230 14.40 0.062

Fl (0.01, 46, 230) = 1.66 F2 (0.01, 5, 230) = 3.02

From the analysis of variance it can be seen that water-levels experienced at each location are significantly different in that their

237 mean levels bear little relationship to each other. If the wells penetrated a homogeneous aquifer, individual well responses could be expected to be of a similar nature. Such an aquifer would be expected to respond in a similar manner to meteorological inputs, and fluc• tuations at any point in such a water body would be expected to have similar characteristics. It Is proposed that the wells in the obser• vation network in the Narrator Brook Valley penetrate aquifer horizons at different locations within the catchment, whose individual responses do not suggest a homogenous aquifer. Water-level fluc• tuations in the Narrator Brook Valley vary widely, as does the llthology of neighbouring sites (Chapter 2) suggesting that the wells Intersect discrete groundwater units within the weathered granite matrix. 5:5 Multivariate Analysis The increased use of statistical methods in hydrology over the last two decades has been most apparent in research papers reporting the use of multivariate methods and multiple regression techniques. Snyder (1962) attributed the widespread use of multiple regression to the fact that it is one of the few numerical methods which can be used to evaluate simultaneously the effect of several causative factors. The uses of niultivariate methods in hydrology are outlined by Rice (1967), who advocated their use since each parameter, in a given situation, can be described by multiple measurements obviating the need to describe a phenomenon by a single number. According to Rice (1967) multivariate methods can be used for the following purposes: (1) to estimate the probability of group membership based on multiple measurements (11) assess the position of a multiple-measurement observation with respect to two or more populations known to be different (111) estimate the minimum number of underlying factors present in a larger set of related measurable variables (iv) measure the correlation between two sets of variables or predict one set from another. As a consequence, multivariate methods allow for a flexibility of approach which Is more in harmony with the nature of hydrological and hydrogeological problems. The three most widely quoted techniques

238 utilised in the literature are principal components analysis, factor analysis, and multiple regression techniques including the use of stepwise regression procedures. Factor analysis and principal components analysis are referred to briefly below by way of developing the reasons for the choice of the stepwise regression procedure in the Narrator Brook. Factor analysis has been used extensively in the study of ground• water chemistry. Dawdy and Feth (1967) used factor analysis in a study of groundwater quality in the Mojave river valley, California, while Ashley and Lloyd (1978) used this technique to display major features of chemical variation and groundwater flow in Chile and Derbyshire. Factor analysis was originally devised to explain the interrelationships in large numbers of variables by the presence of a few factors. The original applications were accompanied by theory which specified the expected nature of the factors and thus allowed their interpretation. When factor analysis is applied to problems in areas where no theories of structure exist, it is necessary to deduce the meaning of the factors. At times this is not possible, because no pattern emerges in the factor loadings, or because the theoretical framework of the problem is too poorly developed for adequate under• standing (Davis 1973). Wallls (1965) suggested that as hydrologlcal data are so rarely a random sample of a homogeneous population, the classical factor analysis will most likely be unproductive, although he does advocate its use as a numerical procedure for screening variables. The major use of principal components analysis are usually related to the simplification of large data sets with upwards of 50 cases and 20 variables, for the purpose of generalising, organisation and understanding (Kent, 1979). Bonell (1978) used principal components analysis to evaluate shallow groundwater movement in a small boulder clay catchment. Diaz et al. (1968) used principal component analysis and factor analysis to identify the factors affecting the water yield from 21 watersheds in Ohio and Texas. These authors also used a stepwise regression procedure to substantiate the variables determined by principal components analysis and factor analysis.

- 239 There have been however, criticisms of principal components analysis as a technique, since Davis (1973) noted that it is not strictly speaking a statistical procedure, but rather a mathematical manipulation in which utility is judged by performance and not by theoretical considerations. Bearing the above features in mind, and taking into consideration that the parameters required to be derived were variables which had the most effect on water-levels, a stepwise regression procedure was adopted. This follows recommendations by Krumbein and Graybill (1965), who noted that when evaluating the importance of a set of independent variables for the prediction of a dependent variable, a stepwise regression technique is important in exploring the behaviour In a process-response model. 5:5:1 Stepwise Regression Although the stepwise regression technique has been widely used in water quality studies, (Keller 1970 and Foster 1978, 1979), its use in hydrogeologlcal work is minimal and has not been documented in the literature. The most common stepwise procedure Is forward selection, which adds variables on the basis of their partial correlations with the dependent variable, such that at each stage the variable with the highest partial correlation is added to the equation. The procedure continues until no further significant variables can be added to the equation. Draper and Smith (1980) recommend the forward selection, stepwise regression procedure as one of the best variable selection procedures. Within the "Geog-Stats Package" available on the Prime Dual 550 system at Plymouth Polytechnic, a stepwise regression program, as devised by Mather (1969) was available, and utilised for the analysis of data collected in the Narrator Brook Catchment. It was considered that the best course of action was to examine as many facets of the situation as possible, and ascertain 'a posteriori' the major factors influencing groundwater fluctuations in the valley aquifer. A multi• variate statistical method, such as a stepwise regression procedure, allows consideration of changes in several properties simultaneously, and hence enables deductions to be made for each well in the catchment, concerning the relative importance of the independent variables.

- 240 As each variable is entered Into the regression the effects on the coefficient of determination (r^) are determined and are printed, along with the partial correlation coefficients and standardised regression coefficients

J.2 variance of predicted values of dependent variable variance of observed values of dependent variable

When the fit is perfect, predicted values equal observed values so r2=1.0. The r^ ratio, converted to a percentage figure, equals the percentage of the variance explained by the regression equation. The appropriate fitted regression equation chosen is finally computed for all the remaining variables still in the model, along with residual values. 5:5:2 General Problems of the use of a Stepwise Regression Procedure In practice research rarely follows the standard method of hypo• thesis generation, data collection, testing and evaluation, so it is common for a range of alternative equations to be explored. In this context stepwise regression procedure search methods may be defended as a form of research strategy since resulting models provide guide• lines for further analysis. Sensible judgement is required in the initial selection of variables, and in the critical examination of the model through the examination of the residuals. Draper and Smith (1980) suggested that it is easy to rely too heavily on the automatic selection performed in the computer. Wallis (1968) suggested that given a wide choice of variables a stepwise regression procedure tends to capitalise on specific errors in the initial data, and to favour composite variables that may introduce an additional source of prediction error when used with control observations. Its chief advantage seems to be that it produces an equation that uses a small number of predictor variables and has a comparatively high r^ value (that percentage of the variance of the dependent variable which is accounted for by the chosen regression equation). Variables selected by the stepwise regression procedure and their coefficients leave much to be desired in ease of understanding of the underlying system, acording to Wallis (1965), but

241 Draper and Smith (1980) suggested that while stepwise regression procedure techniques do not necessarily select the absolute best model they usually select an acceptable one. When difficulties arise in stepwise regression procedure they are primarily due to the presence of high correlations among independent variables (Hauser 1974). When one independent variable correlates highly with another, the variables are said to be colllnear, and when one such variable correlates highly with a combination of the remaining variables there Is multicolllnearlty. Multicollinearity can be found from inspection of the correlation matrix produced for all the variables included in the stepwise regression procedure. The regression coefficients will tend to have high standard errors, and the degree to which individual variables contribute to the regression becomes difficult to ascertain. In the case of two variables palrwlse correlations in excess of about 0.80 are usually taken as evidence of serious collinearity (Wallls 1965). Where such interdependence, or multicollinearity exists, a number of different combinations of independent variables will give almost equally good fits to the empirical data, and it is likely that selec• tion procedures will not produce the best equation. Further, different equations all with similar values of r^, may have very different theoretical implications. Whereas such high correlations may not be a serious problem for the derivation of prediction equations, they can have far reaching Implications for explanatory equations (Hauser 1974). In general, because of the problem of collinearity, care needs to be taken in the interpretation of the role of significant variables chosen by the stepwise regression procedure. Ideally in the case of the Narrator Brook aquifer, the objective was to derive an estimating equation with the emphasis accordingly on maximising the amount of variation in the dependent variable accounted for by the Independent variable set (maximising r^). It was also hoped that an explanatory regression model could be achieved for the Narrator Valley system. Here the emphasis was on Individual regression coefficients and on establishing significant relationships, so that the objective was to maxlmuse r^ subject to significant regression coefficients.

242 - 5:5:3 Variables used in the Stepwise Regression Procedure The stepwise regression procedure was run for each of the 13 observation wells in the Narrator Brook Catchment. The procedure was run twice on each well using two slightly different arrangements of independent variables in an attempt to narrow down possible time lags by using shorter antecedent rainfall periods. Run A used 23 variables and run B used 8 variables. The independent variables chosen for the analysis are listed in Table 5:8. The dependent variable was the water-level observation for each well using 47 weeks of data over the period April 1979 to February 1981.

Table 5:8 Use of the Variables in the Stepwise Regression

Variable Abbreviation Variables used In Runs

A B

1 Rainfall (RF) (RF1-RF15) (RF) or Rainfall periods 1-15 (RF1-RF15)

2 Evapotranspiration (ET) (ET) (ET)

3 Soil moisture as (Site 1) (Site 1) (Site 1) measured at Site 1

4 Soil moisture as (Site 9) (Site 9) (Site 9) measured at Site 9

5 Sine Index (SI) (SI) (SI)

6 Antecedadt precipitation (API 30) (API 30) (API 30) Index for 30 days

7 Anteced^^nt precipitation (API 5) (API 5) (API 5) Index for 5 days

8 Soil moisture deficit (SMD) (SMD) (SMD)

9 Reservoir levels (RSLEV) (RSLEV) -

1. Rainfall The correlation of total catchment rainfall with the response of well-levels in the observation network in the catchment have already been referred to in Section 5:4. The rainfall parameter utilised in

243 the stepwise regression procedure refers to the total rainfall for the seven days prior to the water-level observation. In an attempt to narrow down the possible time lag between rainfall and watertable response, shorter rainfall periods were divided up into distributions overtime with the following characteristics:

Table 5:9 Rainfall Periods

Total Rainfall Period ° antecedent rainfall over a t

RF 1 1-12 hrs RF 2 12-24 RF 3 24-36 RF 4 36-48 RF 5 48-60 RF 6 60-72 RF 7 72-84 RF 8 84-96 RF 9 96-108 RF 10 108-120 RF 11 120-132 RF 12 132-144 RF 13 144-156 RF 14 156-168 RF 15 168-180m hours

The use of antecedent climatic data in conjunction with the statistical technique of correlation was originally developed by Pltty (1966), who correlated springwater hardness fluctuations with antecedent climatic data. This method was adapted by Bonell (1973) who correlated well level data with soil moisture deficit and rainfall In an attempt to determine lag time responses of wells in a boulder clay catchment. The adoption of this statistical approach in the Narrator Brook Valley attributed the variation in well response to rainfall Inputs over time, which by inspection of the well hydrographs suggests a variation in lag time for different wells in the catchment. Run A of the stepwise regression procedure was carried out using RFP1-RFP15 as indicated in Table 5:8. The unproductiveness of this approach (Section 5:6:2) suggests that lag times are not so widely different as originally hypothesised and as a consequence only one value of total rainfall, for 7 days prior to well level observation, was used in Run B.

- 244 2. Potential Evapotranspiration The amount of moisture that a land surface loses by evapo• transpiration depends primarily on incident precipitation, climatic factors of temperature and humidity, along with the extent and type of vegetation cover. The amount may be increased, by large trees whose roots penetrate deeply into the soil bringing up and transpiring water which would otherwise be far beyond the Influence of surface evaporation (Wilson 1974). The potential values of evapotranspiration used are a theoretical amount of loss that would occur without limitations of soil moisture availability, and as such were considered necessary to include in the stepwise regression together with measured soil moisture content in the catchment. Bonell (1973) used potential evapotranspiration in conjunction with rainfall totals to investigate the causal factors responsible for well lag effects in a boulder clay catchment, and Foster (1979) also used potential evapotranspiration as one of his meteorological variables In modelling catchment solute response. The value utilised In this present study was the mean value, in millimeters for the week prior to the water-level observation. The derivation of potential evapotranspiration data from the catchment is described in Chapter 3 and these monthly values were compared with South West Water's monthly estimates for North Hessary Tor 1978-1979. As can be seen from Table 5:10 these monthly figures were similar, and as such the mean weekly value calculated for use in the stepwise regression procedure, were considered to be appropriate.

Table 5:10 Mean Monthly Potential Evapotranspiration Values for North Hessary Tor and Head Weir, Narrator Brook (P.E. ram/month)

Month North Hessary Tor Head Weir

January 1979 12.3 11.5 February 15.2 19.5 March 42.8 34.0 April 52.0 37.5 May 78.0 71.2 June 96.4 97.5 July 116.4 135.5 August 89.7 101.5 September 61.0 72.7 October 33.0 60.2 November 18.8 28.5 December 17.4 16.2 January 1980 11.0 11.5 February 15-6 25.2

TOTAL 659.6 722.5 3 & 4 Soil Moisture Measurements Water in the soil moisture zone is in a continual state of flux, as It moves in response to a number of forces, such as gravity along with vapour pressure and temperature gradients. As a consequence soil water movement can be in any direction. The pattern of change in the distribution of water in the soil mass during infiltration from water on the surface, and during the movement of water upwards from a groundwater system has been treated by many workers (Carson 1969). The similarities between the two processes have been noted by Liakopoulos (1965) and are controlled predominantly by movement in the vertical direction. Movement of soil moisture, in the unsaturated zone, is normally from moist to dry regions, that is, from areas of low soil suction to those of high soil suction. Where the nature of the soil profile results in the Juxtaposition of soils of markedly different textures having the same soil moisture content, there will usually be a movement of moisture from coarse sandy materials to finer clayey soils, because for a given moisture content the soil water will be held at lower suctions in sand than clay. Soil moisture movement will be more rapid in moist soils than dry soils, and that the drier the soil becomes the steeper will be the soil suction gradient necessary to induce any appreciable moisture movement (Ward 1967). The main implications of the above discussion is that, soil moisture measurements in the Narrator Valley are considered to have an effect on the rate of recharge to the watertable in the catchment, and as such are important parameters for inclusion in the stepwise regression. Soil moisture values were determined by the gravimetric method as outlined in Chapter 3, for Sites 1 and 9. Site 1 soil moisture variations are taken to be representative of the free draining soil moisture conditions on the north side of the valley, while Site 9 soil moisture conditions reflect those of the sites on the lower more forested south side of the valley. 5. Sine Index Seasonal indices are frequently incorporated into multivariate statistical models for the development of predictive equations (Keller 1970). Following Walling (1974) a seasonal index, the sine index (SI) was utilised where:

246 SI = sin [(2nD/365), radians] where D = day of year numbered from 1 January. Keller (1970) used his season factor (S) to estimate soil moisture conditions during the course of a year, showing maximum soil moisture taking place at the end of snowmelt period and the beginning of the vegetation period. Depletion would take place until the end of the vegetation growth period in the autumn. The sine index approximates the distribution of the water year with a maximum (dimenslonless) value 0.99 at the end of March, and a minimum of -0.99 at the end of September (Figure 5:11). Foster (1979) noted that the sine index also appears to represent the seasonal biological and hydrologlcal rhythms of the catchment, reflected in solute responses in streamwater from a woodland catchment in east Devon. 6 & 7 Antecedent Precipitation Indices Antecedent precipitation indices provide an indication, for a specific moment in time, of the catchment wetness factor. Such Indices are primarily concerned with an estimate of the associated soil moisture conditions. A decay function Is normally employed in the calculation so that past rainfall exerts progressively less Influence on the value of the index as the time span Increases, (Gregory and Walling 1973). The antecedent precipitation index used in the stepwise regression procedure was calculated after Finlayson (1978) who used a reciprocal decay function to evaluate API for a rainfall total (Pt) occurring (t) days previously. This is given by:

APIs = 1'=' - X P t = l t where APIg = antecedent precipitation index for five days P = precipitation on a day (t) before the calculation. A 30 day antecedent precipitation index was also calculated for use In the stepwise regression procedure, to take account of the

247 FIGURE 5:11

I Distribution of the Sine Index in the

Narrator Brook Catchment

10

X

-00 4

5)

-10 M A M A J 1980 1979 TIME variable nature of the Narrator Brook valley materials, where Infil• trating rainfall may be in some areas, delayed for greater periods than 5-10 days. 8, Soil Moisture Deficit Soil variability imposes serious problems in sampling techniques. The validity of the method depends on how far the soil hydraulic conductivity, soil moisture and tension gradient measured at one location, can be matched by measurements taken at another site, (McGowan, 1975). According to this author, considerable lateral variation in the hydrologically significant properties may occur in soils sampled at sites no more than Im apart, and even in soils which look relatively uniform. Coleman and Farrar (1966) investigating the relationships between soil moisture and watertable depth on a London clay catchment, found that for a particular site, the observed depth of the watertable at any season, varied linearly with the calculated soil moisture deficit in the soil profile beneath a grass cover. Considering McGowan's (1975) and Coleman and Farrar's (1966) findings, the use of a general soil moisture deficit was considered a suitable parameter for inclusion in the stepwise regression procedure. The soil moisture deficit used was calculated on a dally basis after Grindley (1967). In the calculation procedure a zero soil moisture deficit is assumed during late winter when the soil moisture is at field capacity, and subsequently the cumulative effect of rainfall accretion minus evapotranspiration loss is evaluated. Measures of soil moisture deficit, antecedent precipitation index and the sine index combined both theoretical concepts and physically based data. These variables were considered to represent meaningful variations In catchment conditions immediately prior to a water-level recording, and consequently were used in an attempt to derive parameters likely to significantly influence water-level fluctuations in the Narrator Brook Catchment. 9. Reservoir Level Knill (19721), in a comprehensive paper discussing assessment of reservoir feasibility, noted a variety of widespread groundwater pressure effects, brought about in an area by the construction and subsequent impounding of reservoirs.

- 249 In the general case Knlll (1972t) suggested there will be a rise In watertable levels on the reservoir flank and a movement of the groundwater divide towards the reservoir, which will lead to an increase in the proportion of groundwater flow discharging on the opposite side of the flank to the reservoir itself. Kennard and Knill (1969) noticed significant groundwater pressure changes in the layered sequence of limestones at Cow Green Reservoir in Northern England, raising the watertable from a few metres below topwater level to heights of 7m in excess of the reservoirs topwater level. Van Everdingen (1968) noted similar loading features due to the effect of the Saskatchewan Reservoir on local groundwater regimes. Consideration of the effects of such features on water-levels in the Narrator Valley, brought about by water-level variations in Burrator Reservoir, were taken Into account. The change in elevation of the topwater surface of Burrator Reservoir is measured daily by S.W.W. at Head Weir. The mean value for the week prior to well level obser• vations were incorporated into the stepwise regression procedure. This mean value was chosen to be consistent with values for rainfall, and evapotranspiration used in the stepwise regression procedure. 5:5:4 Data Transformation Precipitation, transpiration, evaporation, soil moisture and water quality vary throughout the year in any catchment area. Indices describing them may be non random and non homogeneous (Riggs 1968). Most of the line-fitting statistical procedures assume that the data are normally distributed, with the means of the distribution located along a line and with the variances of the distributions being every• where the same. Leopold (1962) has shown that a logarithmic transformation is more likely to produce a normal frequency distribution of hydrologlcal data. The appropriate transformations for the independent variables in the Narrator Brook are listed in Table 5:11. The variable soil moisture deficit and rainfall totals contained zero values which initially precluded a logarithmic transformation. Traditionally a small increment of the order of 0.001 is added to all variables to facilitate transformation, but Kllmartln and Peterson (1972) have shown that in logarithmic transformation, this particular value becomes -3.0, and in variance computation has a value of 9.0. The

250 desired null effect on a log-transformation was obtained by computation of log-SMD and log-Rainfall as:

P + AP = log (0.0 + 1.0) 0.0 where P => initial value + AP AP = the added increment.

Table 5:11 Transformations Chosen for the Variables in the Narrator Brook Catchment

Variable Transformation

SMD Log + I RF Log + 1 SI No transformation RLEV Log Site 1 Log Site 9 Log APIS Log API30 Log ET Log WL Loff .

5:6 Stepwise Regression Results for the Narrator Brook Catchment 5:6:1 Correlation Matrix The matrix of correlation coefficients produced by the stepwise regression procedure, for each of the observation wells are tabulated in Tables 5:12a - 5:12m. From the 13 individual correlation matrices the significant correlations have been abstracted and presented in Table 5:13. This table effectively summarises the most significant (at 0.01 significance level) variables at each well site which affect the water levels measured in these wells. In general it can be seen from Table 5:13 that evapotranspiration and soil moisture conditions are the most Important factors contributing to the water level observations experienced. It is apparent that the longer term antecedent precipitation index Is a much better Indicator of the general state of the catchment 'wetness* than are site specific soil moisture conditions.

251 Table 5:12a Matrix of Correlation Coefficients Produced by the Stepwise Regression Procedure for Wells In the Narrator Brook Catchment

Well 1 RF APIs API30 Site 1 Site 9 SMD ET SI RF ** API5 0.572 ** ** API30 0.662 0.794 ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS SMD NS NS NS NS NS ** ** ET NS -0.465 -0.512 0.793 NS -0.416 SI NS NS NS NS NS NS NS ** * * ** WL NS -0.470 -0.500 0.569 -0.318 NS 0.626 -0.504

Table 5:12b

Well 2 RF API5 API30 Site 1 Site 9 SMD ET SI RF - ** API5 0.572 ** ** API30 0.662 0.794 ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS * SMD NS NS NS -0.272 NS ** ** ET NS -0.465 -0.512 0.793 NS -0.416 SI NS NS NS NS NS NS NS * WL NS NS NS NS NS NS NS -0.305

NS Not significant leit 0.01 significance level 0.05 significance level

252 Table 5:12c

Well 3 RF API5 API30 Site 1 Site 9 SMD ET SI RF ** API5 0.572 ** ** API30 0.662 0.794 ** ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS SMD NS NS NS NS NS ** ** ** ET NS -0.465 -0.512 0.793 NS -0.416 SI NS NS NS NS NS NS NS ** * * ** * WL NS NS NS 0.523 -0.259 -0.289 0.696 -0.283

NS = Not significant ** =3 0.01 significance level * = 0.05 significance level

Table 5:12d

Well 4 RF API 5 API30 Site 1 Site 9 SMD ET SI RF ** API5 0.572

API30 0.662 0.794 ** ** ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS SMD NS NS NS NS NS ** ** ** ET NS -0.465 -0.512 0.793 NS -0.416 SI NS NS NS NS NS NS NS ** WL NS NS NS NS NS NS 0.404 -0.401

- 253 - Table 5:12e

Well 6 RF API5 API30 Site 1 Site 9 SMD ET SI RF - API5 0.572 ** API30 0.662 0.794 ** ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS A SMD NS NS NS -0.272 NS ** ** ET NS -0.465 -0.512 0.793 NS -0.416 SI NS NS NS NS NS NS NS ** ** ** * * ** ** WL NS -0.456 -0.491 0.663 -0.324 -0.246 0.809 -0.416

Table 5:12f

Well 7 RF API5 API30 Site 1 Site 9 SMD ET SI RF - ** API5 0.572 ** API30 0.662 0.794 ** ** ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS SMD NS NS NS NS NS ** ** ** It* ET NS -0.465 -0.512 0.793 NS -0.416 SI NS NS NS NS NS NS NS ** It* * WL NS NS -0.460 0.388 NS NS 0.321 -0.389

- 254 Table 5:12g

Well 9 RF API 5 API30 Site 1 Site 9 SMD ET SI RF -

API5 0.572 ** API30 0.662 0.794 ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS SMD NS NS NS NS NS ** ** ET NS -0.465 -0.512 0.793 NS NS SI NS NS NS NS NS NS NS ** ** ** ** WL NS -0.461 -0.502 0.563 -0.338 NS 0.696 -0.408

Table 5:12h

Well 10 RF API5 API30 Site 1 Site 9 SMD ET SI RF - ** API5 0.572 - ** API30 0.662 0.794 ** *• ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS

SMD NS NS NS NS NS ** ** ** ET NS -0.465 -0.512 0.793 NS -0.416 SI NS NS NS NS NS NS NS ** ** ** WL NS -0.483 -0.520 0.675 -0.365 -0.262 0.746 -0.403

- 255 - Table 5:121

Well 11 RF APIs API30 Site 1 Site 9 SMD ET SI RF - ** APIs 0.572 ** API30 0.662 0.794 ** ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS SMD NS NS NS NS NS ** ET NS -0.465 -0.512 0.793 NS NS SI NS NS NS NS NS NS NS ** ** * ** ** WL NS -0.479 -0.511 0.612 -0.287 -0.362 0.744 NS

Table 5:12J

Well 12 RF APIs API30 Site 1 Site 9 SMD ET SI RF - ** APIs 0.572 it* API30 0.662 0.794 ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS SMD NS NS NS NS NS ** A* ET NS -0.465 -0.512 0.793 NS -0.416 SI NS NS NS NS NS NS NS ** ** ** * WL NS -0.481 -0.517 0.573 -0.305 NS 0.747 -0.406

256 Table 5:12k

Well 13 RF APIs API30 Site 1 Site 9 SMD ET SI RF -

APIs 0.572 ** API30 0.662 0.794 ** ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS

SMD NS NS NS NS NS ** ** ET NS -0.465 -0.512 0.793 NS -0.416 SI NS NS NS NS NS NS NS ** ** ** WL NS NS NS 0.514 NS NS 0.545 -0.766

Table 5:121

Well 14 RF APIs API30 Site 1 Site 9 SMD ET SI RF - ** API5 0.572 ** API30 0.662 0.794 ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS SMD NS NS NS NS NS ** ** ** ET NS -0.465 -0.512 0.793 NS -0.416 SI NS NS NS NS NS NS NS A* ** ** * ** ** WL NS -0.444 -0.474 0.577 -0.293 NS 0.659 -0.520

- 257 Table 5:12m

Well 15 RF API5 API30 Site 1 Site 9 SMD ET SI RF

API5 0.572 ** ** API30 0.662 0.794 •kit ** Site 1 -0.340 -0.648 -0.636 Site 9 NS NS NS NS SMD NS NS NS NS NS ** ** ** ET NS -0.465 -0.512 0.793 NS -0.416 SI NS NS NS NS NS NS NS ** * *4t ** WL NS -0.427 -0.493 0.574 -0.253 NS 0.689 -0.433

Table 5:13 Summary of Significant Parameter Correlation Coefficients for the Narrator Brook from the Stepwise Regression Procedure

WL+ WL+ WL+ WL+ WL+ WL+ET WL+ Well API5 API30 Site 1 Site 9 SMD SI 1 -0.470 -0.500 0.569 -0.318* NS 0.626 -0.504 2 NS NS NS NS NS NS -0.305* 3 NS NS 0.523 -0.259* -0.289* 0.696 -0.283* 4 NS NS NS NS NS 0.404 -0.401 6 -0.456 -0.491 0.663 -0.324* NS 0.809 -0.416 7 NS -0.460 NS -0.246* - 0.321* -0.389 9 -0.461 -0.502 0.^63 -0.338 NS 0.696 -0.408 10 -0.483 -0.520 Q..675 -0.365 -0.262* 0.746 -0.403 11 -0.479 -0.511 .0,612 -0.287* -0.362 0.744 NS 12 -0.481 -0.517 0.573 -0.305* NS 0.747 -0.406 13 -0.362 NS 0.517 NS NS 0.545 -0.766 14 -0.444 -0.474 0.577 -0.293* NS 0.659 -0.520 15 -0.427 -0.493 Q.574 -0,253* NS 0.699 -0.433

= Highest (+) and (-) correlation coefficients = Lowest (+) and (-) correlation coefficients where WL = water level API5 = five day antecedent precipitation index API30 = thirty day antecedent precipitation index Site 1 = soil moisture content as measured at site 1 SI = seasonal index (Sine Index) Site 9 = soil moisture content as measured at site 9 SMD = soil moisture defeclt NS = no significant correlation observed in the correlation matrix Figures denoted (*) are significant at the 0.05 level, all other figures are significant at the 0.01 level.

258 The seasonal sine Index is a significant (0.01 significance level) parameter at all sites, except Well 11, where the correlation is not statistically significant (Table 5:13). The correlation between the sine index and water level observations is a negative one, demonstrating that as watertable elevation rises the sine index decreases. From the 13 Individual correlation matrices the significant correlations (0,01 significance level), of the variables which are common to all sites have been abstracted. Table 5:14 summarises the parameters whose effects are of a similar magnitude over the entire catchment. Rainfall and antecedent precipitation indices are strongly correlated as are evapotranspiration and soil moisture parameters. It is evident that soil moisture and the antecedent precipitation indices are measuring the same 'control' in the Narrator Brook Catchment. The antecedent precipitation index is a more efficient measure because It is providing a more accurate longer terra statement of average catchment conditions, than the one point in time and space soil moisture observations.

Table 5:14 Summary of Correlation Coefficients of Variables Common to all Wells in the Narrator Valley

Correlated variables Correlation coefficient

Rainfall and APIs 0.572 Rainfall and API30 0.622 APIs and API30 0.794 Site 1 and API5 -0.648 Site 1 and API30 -0.636 ET and APIs -0.465 ET and API30 -0.512 ET and Site 1 0.793 ET and SMD -0.416

all the above correlation coefficient are significant at 0.01 level.

Table 5:15 summarises the correlations between groundwater levels and reservoir level fluctuations in the Narrator Brook Valley, from

259 which significant (at 0.01 significance level) negative correlations are evident. One interpretation of this negative correlation is that drainage of groundwater from the saturated weathered granitic materials contribute significantly to the reservoir storage. This downstream groundwater drainage component from the weathered granitic materials is unlikely to be a process restricted to the Narrator Brook Valley and may be an important feature in similar valleys elsewhere on Dartmoor.

Table 5:15 Negative Correlations between Reservoir Levels and Water Level Observations in the Narratory Brook Valley

Well Correlation Coefficient Significance Level

1 -0.675 0.01 2 -0.585 *• 3 -0.658 *• 4 -0.547 t* 6 -0.571 ti 7 -0.666 •* 9 -0.669 t* 10 -0.651 ti 11 -0.541 12 -0.608 " 13 -0.675 «• 14 -0.701 M 15 -0.662 H

5:6:2 Stepwise Regression Procedure Runs A and B Two runs of the stepwise regression procedure were carried out on data from each of the observation wells. Run A utilised the rainfall periods (as outlined in section 5:5) and the reservoir levels together with soil moisture, water-levels and meteorological parameters. The most significant explanatory variables chosen to account for the variations observed in the watertable levels were reservoir level, evapotranspiration and depending on the well rainfall periods 8 to 15, and Site 1 soil moisture measurements. These variables were considered to be responsible for some 70-80% of the variations in the watertable level. The inclusion of reservoir level in the regression equation in Run A may be explicable in that the reservoir experiences the same meteorological Inputs as the groundwater, and is accordingly influenced by antecedent conditions in much the same way as the water-

260 - table in the Narrator Valley. Response to such influences Is manifested by a greater magnitude of level changes, partly due to direct precipitation to the water surface over O.eOkm^, and because the reservoir is at the downstream end of the hydrogeological system to which all accrued Inputs gravitate. However the correlation of water-level fluctuations with reservoir level fluctuations. Table 5:15, have shown that a negative correlation exists. This may be explained by a significant lag in time between the response of the wells to Input and the reservoir levels response to input. Response to input over the reservoir surface will be more immediate than the response of the wells in the weathered granite aquifer further upstream, and the reservoir level will rise whilst the groundwater levels In the wells are still receding prior to the next recharge event. As the later rainfall jfrlods, in excess of 84 hour antecedent totals, were the most frequently chosen as being important parameters in Run A, it was decided to assign a single rainfall parameter to the independent variables. The value adopted was the total rainfall for the seven days prior to the water-level observation. Table 5:16 illustrates the significant parameters chosen by the stepwise regression procedure for Run B. The rejected parameter column in Table 5:16 lists those variables chosen as next having an effect on water level at each well site, but they are subsequently rejected as not adding significantly to the explained variance. As can be seen from Table 5:16 evapotranspiration appears to be the most important parameter influencing water-level fluctuations in most of the wells and accounting for between 48 and 55% of the variance. Wells 2, 7 and 13 do not fall into this general trend, which is outlined below by reference to the most important parameters chosen at each well site.

- 261 Table 5:16 Significant and Rejected Parameters in the Narrator Brook (Run B) as Selected by the Stepwise Regression Procedure

Well Order of Significant r2 (%) F-tables Rejected Parameters Significance Parameter Level

1 E.T. S.I. 63 0.01 A.P.I.30 2 R.F. 9 N.S. S.I. 3 E.T. 49 0.01 Site 1 4 E.T. 67 0.01 R.F. 6 E.T. Site 1, R.F. 75 0.01 S.I. 7 A.P.I.309 S.I. 33 N.S. Site 1 9 E.T. 48 0.01 S.I. 10 E.T. S.I. 71 0.01 Site 1 11 E.T. 55 0.01 A.P.I.30 12 E.T. 56 0.01 S.I. 13 S.I. E.T. 87 0.01 R.F. 14 E.T. S.I. 69 0.01 A.P.I.30 15 E.T. S.I. 66 0.01 A.P.I.30

Well 1 Evapotranspiration and the sine index are chosen as the most important parameters explaining 63% of the total variation in water- levels of this location. The regression equation chosen, using these two variables is significant at the 0.01 level. This choice of parameters to explain the water-levels observed at Site 1 suggests that groundwater at this site although from a deep source has a significant component of groundwater from a shallower origin than those groundwaters of Wells 2 and 7. Such a component may be derived from more recent precipitation-fed sources either from vertical or lateral flow processes, as compared with the exclusively deeper derived waters in Well 7 and Well 2 which are not affected by evapotranspiration.

Well 2 No parameters were found to be statistically significant for this site. The most important parameter chosen by the stepwise regression procedure was rainfall which accounted for only 9.00% of the variation in water-levels, and consequently the chosen regression equation is not statistically significant. This shows that the range of initial parameters chosen to help explain water-level variations are

- 262 inadequate in this case. As can be seen from the borehole logs in Appendix 1, this well penetrates a large proportion of sands and gravels, particularly around the screened section. Lateral and vertical movement through these is likely to be very rapid and hence total rainfall inputs and their seasonal distribution, as characterised by the sine index, will affect the water-level fluc• tuations more markedly. A parameter characterising grain size distri• bution and/or permeability variations may well have been a more useful inclusion into the stepwise regression procedure at this location. Well 3 and Well 4 At these two sites evapotranspiration was chosen as being the most significant parameter explaining total water-level variations. Both regression equations are significant at the 0.01 significance levels, with an r^ of 49% for Well 3 and 67% for Well 4. It is likely that the incidence and duration of site flooding, particularly at Well 3, will have some bearing on water-level fluctuations observed at this location. The watertable positions at these two locations is never greater than 0.5m below the surface so evapotranspiration would be expected to affect the water-levels. River stage variations may have been a more appropriate explanatory variable for use in the stepwise regression procedure in these bankside locations. However, due to the variable hydrogeological properties of the weathered granite material in the valley, as outlined in Chapter 4 it was felt that river stage measurements from Station 11 and Station Cutt would not have been appropriate for use at this location. Well 6 Evapotranspiration, soil moisture at Site 1 and rainfall account for 75% of the variations of the watertable at this location. The regression equation is significant at the 0.01 level. The r^ value is one of the highest In the observation wells (Table 5:16). This situation may be explained by the shallow depth to the watertable, so evapotranspiration processes would be expected to have a significant effect on the watertable height. Site 6 Is subject to occasional flooding resulting in saturated soil horizons, and the coarseness of the valley materials at this location will result in a more rapid transmission of moisture towards the watertable. The nearness of the

263 reservoir itself may be an influence on the shallow depth to ground• water at this site. Well 7 The thirty-day antecedent precipitation index (API30) and the sine index account for only 33% of the variation of water-level fluc• tuations at this location, and consequently the regression equation selected by the stepwise regression procedure is not statistically a significant one. Antecedent precipitation and the sine index are relatively Important variables, but alternative parameters need to be elucidated to explain water-level variations at Well 7. From the borehole logs in Appendix 1, it can be seen that approximately 6m of sands are penetrated along with 3m of granitic material around the screened section. Movement of water horizontally or vertically through fissured granite bedrock may be responsible for a proportion of the water-level variation In the borehole, and a suitable parameter to characterise gross granitic permeability may be useful for inclusion into the stepwise regression procedure at this site. Soil moisture measurements, as determined at Site I higher up the northern slope. Is a rejected variable In the regression, but the nearness to the surface of sandy material would suggest a more rapidly draining subsoil at Well 7 and antecedent soil moisture conditions would pre-determlne the relative speed of water movement downwards through this thickness. A parameter characterising the relative thicknesses and permeabilities of the different llthologies at a given site may prove to be a useful inclusion into the stepwise regression procedure, in an attempt to both explain water-level fluctuations, and to predict further changes at a specific site. Well 9 Evapotranspiratlon accounts for 48% of the total variation in water-level observed at Well 9, resulting in a regression equation which is significant at 0.01 level (Table 5:16). Well 10 The Influence of evapotransplration and the sine index represent 71% of the explained variation in water levels observed at this location, giving a regression equation which is significant at 0.01 (Table 5:15).

- 264 Well U The watertable position at Well 11 is rarely greater than Im below the ground surface^ and as such evapotransplratloa would be expected to account for some variation in the obseryed water-levels. As can be seen from Table 3:16 evapotranspiration accounts for 55% significant at the 0.01 level. The observation well is thought to penetrate a semi-confined situation where a granitic boulder or layer of more coherent granitic materials, partially confines the underlying sand and gritty lithologies, and where the antecedent precipitation index influences watertable height over a longer time period than would be expected for shallow watertable conditions. Well 12 'Evapotranspiration accounts for 56% explained variation of water-levels at this site, and the equation selected as the best available from the stepwise regression procedure is significant at the 0.01 level. Sites 9, 10 and 12 are all of a similar topographic elevation and have similarities in materials penetrated, although Well 9 has a greater quantity of clay. Well 13 The sine index and evapotranspiration are the most important parameters explaining the total variation In water levels accounted for by the regression of site 13. An r^ of 87% is derived with the regression being significant at 0.01 level. However, it is felt that Well 13*s results must be viewed tentatively as damaged casing may be responsible for a large proportion of the variation in water-levels. Well 14 and Well 15 Evapotranspiration and the sine index are the most important variables having an influence on water levels at Sites 14 and 15. Both regression equations chosen by the stepwise regression procedure, are statistically significant at 0.01 level, and r^ values for Well 14 and 15 are 69% and 66% respectively. Both these wells are similar from topographic and llthological viewpoints, and although being the fourth and fifth deepest wells in the catchment, still show modification due to evapotranspiration processes active in the forested parts of the catchment. This factor again suggests that groundwaters at these sites are of a relatively shallow origin, being derived from more recent precipitation-fed sources.

- 265 5:6:3 General Patterns The data presented suggests that evapotransplratlon, sine Index, rainfall and APl3o, In association with soil moisture conditions typified by those at Site 1, appear to be the main parameters which have a significant effect on the groundwater level fluctuations experienced. On the basis of the statistically significant parameters presented in Table 5:16, the observation wells can be divided into three groups. These are as follows:

Group (1) Wells 1. 10. 13, 14, 15 Group (11) Wells 3, A, 9, 11, 12 Group (ill) Wells 2, 7, 6.

Group (1) These consist of wells whose explained variation in water-level fluctuation can be attributed to evapotransplration and the sine index parameters (Table 5:16). With the exception of Well 10 Figure 12, these wells are situated on the topographically higher positions on the north and south side of the valley. Group (11) These consist of wells whose explained variation in water-level fluctuation are attributed to evapotransplratlon effects only (Table 5:16). As can be seen from Figure 12 the wells in this group are generally situated at lower elevations, with the shallowest wells 3 and A at the lowest points. Group (ill) This group consists of wells which show anomalous results and are quite distinct from the characteristics shown by either Group (1) or Group (11). Table 5:16 illustrates that various combinations of the rainfall, evapotranspiration. Site 1 soil moisture conditions and the thirty day antecedent precipitation index parameters, may account for the explained variations of water-levels experienced at these locations.

266 - Group O^i)

Group (i )

^ Group (il^

I25m

FIGURE 5:12 Well groupings determined from the Stepwise Regression Procedure 5:6:4 Summary From material presented In this Chapter, four independent forms of analysis have been utilised. In an attempt to group the wells in the observation network. Table 5:17 summarises these groups, and it is obvious that all four analyses produce groupings which have common wells.

Table 5:17 Summary of Hydrogeological Groupings of wells in Narrator Brook from four types of analysis

Type of Analysis Designated Group Wells in Group

Hydrograph Amplitude A 3,4,6,11 B 1,2,7,10,12,13 C 9,14,15

Change in a 3,4,6,11 water-level b 1,2,7,10,12,13 mapping c 9.14,15

Standard deviation 1 1,2,7,10,12,13 of water-levels 2 3,4,6,11 and the 3 9,14,15 analysis of variance

Stepwise (i) 1,10,13,14,15 Regression (11) 3,4,9,11,12 (ill) 2,7,6

Grouping based on hydrograph amplitude, change in water-level map, standard deviation of water-levels and the analysis of variance all produce similar hydrogeological well groupings. Geological features, in the form of formation stability during drilling, (Chapter 3), may have some effect on the hydrogeological response of individual wells* Wells 3, 4, 6 and 11 exhibited badly caving formations during drilling procedures, and these shallow wells frequently occur in the same hydrogeological grouping. Wells 9, 14 and 15 lost their screen- bungs during completion, but this effect on the hydrogeological well response may be insignificant because of the large proportion of clay materials around or in the near-vicinity of their well-screens.

268 - Evidence for Che influence of well completion procedures on Che response of Che wells Is not available from Che NarraCor Brook Valley. On the basis of the previous four groupings as Indicated In Table 5:16 the wells In the catchment can be broadly assigned Co three hydrogeological provinces In the Narrator Brook and these are Illustrated In Figure 13. I Regions predominantly of recharge characterised by wells in groups (1), 3, C and c. II Regions mainly of discharge characterised by wells in groups (11), 2, A and a. Ill Areas of anomalous fluctuations suggesting deeper flow patterns as in the case of Wells 7 and 2, or flow patterns influenced by surface water bodies as in the case of Well 6. These conditions are found in group (ill) and to a lesser extent in groups 1, b and B. 5:6:5 Predictive Equations The generalised regression equation is given in Table 5:18 along with a listing of observed water-level (y) predicted water-level (y*) constant (bg) observed dependent variables (b^ b2 b3) and the partial regression coefficients (x^ X2 X3) of the dependent variables. The generalised equation is regarded as an estimating equation for future water-level positions at each of the sites, given rainfall, soil moisture deficit, evapotranspiratlon and soil moisture conditions. As seen from Section 5:6:2 r^ = 48-86%, leaving 14-52% unexplained variance for the water-levels in the catchment as described by the parameters utilised in the stepwise regression procedure* To account for this proportion of unexplained variance, and to test the adequacy of the regression equations an analysis of residuals was conducted. 5:6:6 Analysis of Residuals From the point of view of achieving an appropriate explanatory regression model for the groundwater-level variations apparent in the Narrator Brook, r^ needs to be maximised for each site* This achieves as much explanation for water-level fluctuations as is possible at each site, using the chosen independent variables. However other independent variables, not included in the stepwise regression procedure may be affecting the behaviour of the dependent variable, the water-level.

- 269 - I Recharge

Discharge

\\\'' Anomalous Province

I25m

FIGURE 5.13 HYDROGEOLOGICAL PROVINCES IN THE NARRATOR VALLEY T^ble 5:18 Predictive Equations for the Narrator Brock as selected by the Stepwise Regression Procedure

Partial Regression Ooef f icients of dependent variables

observed predicted constant observed dependent variables y y' bo bl b2 b3 XI X2

1 1.0540 1.0651 1.05232 1.541 1.085 - 0.05189 - - 2 0.7780 0.8893 0.96335 0.7000 - - -0.10576 - - 3 -0.7440 -0.8180 -1.48983 0.9420 - - 0.71323 - - 4 0.1900 0.1994 0.25040 0.9420 - - -0.05635 - - 6 -0.3660 -0.2656 1.05548 0.9420 1.452 0.700 0.33140 -1.09721 -0.05733 7 0.6780 0.6859 1.03027 1.132 0.9900 - -0.22918 -0.08580 - 9 0.3320 0.3821 -0.00394 0.9420 - - 0.40979 - - 10 0.4500 0.4610 0.42017 1.085 0.9900 - 0.06206 -0.02674 - 11 -0.2290 -0.2739 -0.70711 0.9420 - - 0.45986 - - 12 0.2920 0.3412 -0.00130 0.9420 - - 0.36363 - - 13 0.6610 0.6715 0.67551 0.9900 1.085 - -0.05735 0.04860 - 14 0.6970 0.7335 0.69407 1.085 0.9900 - 0.08778 -0.05640 - 15 0.5880 0.6248 0.51964 1.085 0.9900 — 0.17736 -0.08820 —

bo + bj + b2 X2 logioy bo + bl loglO XI + b2 loglO X2 bl bo y ^antilx)glO + ^1 ' + «2 ^

271 - To clarify the situation an analysis of residuals was carried out for each site. A residual is the difference between Y observed and Y estimated by the equation, that is, the amount which the regression equation has not been able to explain. Residuals may also be regarded as the observed errors If the model is correct. For the Narrator Valley the stepwise regression procedure prints out values for Y estimated and the residual for each site. These have been plotted, Y estimated against residual, and are illustrated in Figures 5:lAa-5:14c. Certain assumptions are made about the errors, (Draper and Smith 1980) (1) they are independent (11) have zero mean (ill) have a constant variance (Iv) follow a normal distribution. Thus if the fitted model is correct the residuals should exhibit tendencies that tend to confirm the assumptions made, or at least should not exhibit a denial of the assumptions made. In general, if a plot of residuals of a regression equation against Y estimated, produces a horizontal band either side of the Y estimated axis, within which the points are distributed, then this is taken to indicate that no abnormality occurs and the least squares regression analysis would not appear invalidated, and that the above four assumptions are confirmed (Draper and Smith 1980). From the analysis of residual diagrams for the Narrator Valley Wells 1, 6, 9, 11, 12, 13, lA and 15 exhibit the characteristic horizontal spread of points which suggests that the regression model chosen by the stepwise regression procedure is adequate in all these cases. There are however some deviations from this distribution, exhibited by Wells 2, 3, A, 7 and 10. The marked vertical spread of plotted points in the case of Wells 2 and 10, Figure 5:lAc, indicate that there is a dependence of the residuals on the chosen variable In the regression model at these locations. Well A (Figure 5:lAc) shows this feature also, but to a less well marked degree. As a consequence the model may be regarded as inadequate for these three sites. The spread of residuals as indicated for the plot of Wells 3 and 7 (Figure 5:lAc) suggests that the variance of the data is not

272 0.^ WELL 1 0.2r WELL 6 1 0.0 If ,' I I I ^00• . -110 115 120 -0-1. .-0-2 -0-3 -0-4 • raS -0-6 Of "-0.1 °=-0.2 # - y' estimated -»> -0.2 y' estimated

0.2 WELL 9

0 0.1

00 1 0.1 0.3 0.5- 0-7 °^-0.1 -02'- y'estimated

02 r WELL 11 0.2 r WELL 12

o 01 • o 0.1 h

2 0.0 ^ 0.0 J I a. -0-2-0-3 -0 4 •0-6 0-1 -O^ 0-3.0.4 0-6 -0.1 °=-0.1|-

-0-2L y' estimated -0.21- y' estimated

FIG. 5:14 a DISTRIBUTION OF RESIDUALS PRODUCED FROM THE REGRESSION EQUATIONS FOR INDIVIDUAL WELL SITES IN THE NARRATOR BROOK. 0.1 r WELL 13 0.1 WELL U

0.051- 0.05 o o o 00 I fv;„ I I ^ 0.0 I'l: I.. I I 0.5 0.6 0.7. - 0-9 cr as 0 6 -07 -as os Of cr -0.05 L •0-05 h y' estimated

0 01L y' estimated

010 r WELL 15

o 0.051- 0..4 .0..5' 07 S 0.0 0.6 .•••;0 8 0.9 -0 05

OIOL y' estimated

FIG. 5.14b WELL 2 0.06

0-04 O.^r WELL 3

0 0-02 0.2

1 0.0 0 0 .1.5 aH^9 I .10 20 °^ -0.02 (2 -0.2

0.04 -0.4 y' estimated

•006 L y'estimated

0-04 WELL 4 a 0-02 3 w 0.0 "I0S; 0-1 0.2 .0.3 0-4 •1 -0.02

-0.04

-0-06 y' estimated

WELL 10 0.05r 02 WELL 7

o 0.02 1 •o w 00 0.5 m • 0 9 * 0.1 0-2 0-3 O.A-.0;5 0-6 tr -01 -002

-0.2 L y' estimated 0.05 y' estimated

FIG. 5:i4c constant as assumed in the initial hypothesis, and hence the model is not appropriate for these two locations. Well 2 and 7's regression equations, as outlined in Section 5:6:2, are not statistically significant, which is substantiated by this analysis of residuals. The variance of the data produced by the selected regression equations may be reduced, in most of the well locations, by the inclusion into the stepwise regression procedure of further parameters likely to affect water-level at each individual site. These additional variables have been referred to at individual sites in Section 5:6:2, but are summarised here. Parameters describing grain size distribution, permeability variations, thicknesses of individual materials and variations at a well site, incidence and duration of site flooding, variations in river stage, slope factors and elevation along with contributions from interflow, may further aid in the production of predictive equations to estimate the watertable elevation in the Narrator Valley. In addition, the effect of vegetation cover and type, which will effect evapotranspiration totals, may cause significant variations in groundwater levels in the catchment. 5:6:7 Statistical Significance of Multiple Regression Equations Little attention is paid to the significance of equations derived from stepwise regression procedure in the literature, due possibly to the fact that statistical tables assume hypotheses proposed in advance of the examination of sample data. For the stepwise regression procedure the dependent variable is loosely hypothesised to be related to one or more variables from a set of independent variables, and a search is made for the most important ones. The resulting model is derived from the same data set which is then used to test the level of significance. Normal test statistics are Just not applicable in such situations. However, to determine whether the multiple regression equations produced by the stepwise regression procedure are statistically significant an analysis of variance was carried out. Very few applications of this technique are documented because most analyses of variance are based on data from a designed experiment, and it is this application for which best results are obtained (Riggs 1968). An analysis of variance was used specifically in the Narrator Valley to

- 276 test to see if there was a difference between the sum of squares of the regression, or explained variation; and the residual sum of squares, or the unexplained variation. The total variation of the data sets used in the Narrator Valley, Is the sum of the regression and residual variation for each well. Tables in Appendix 4 tabulate these values for each well site but table 5:19 summarises the information so derived. This table also summarises the r^ values derived for each well from the stepwise regression procedure. It can be seen that Well 2 and Well 7 do not have significant regressions. Further comments regarding these features are outlined in section 5:6:2.

Table 5:19 Significance of Stepwise Regression Coefficients and r^{%) values

Well r2 % F-Ratio (calculated) F-Ratio at 0.01 Significance Level

1 63.10 9.53 3.12 2 9.32 0.57* 3 48.53 5.25 4 67.15 13.13 6 74.83 16.56 7 33.37 2.79* 9 48.46 4.97 10 70.55 13.35 11 55.34 6.9 12 55.80 7.04 •• 13 86.49 35.81 *• 14 68.87 12.32 15 66.28 10.95

* Regression equations are not significant at 0.01 level

5:6:8 Discussion of the Stepwise Regression Procedure Results Although the stepwise regression procedure appears to have produced some useful results, the technique itself may not be an entirely suitable method for analysing groundwater fluctuations in the Narrator Valley. The seasonal trends in input to the catchment, will be similarly received and reflected by the overall change in reservoir levels at the downstream end of the valley. Reservoir levels show a greater magnitude of these responses to inputs and this may mask any correlation with other explanatory variables.

- 277 Seasonal trends will be of a similar magnitude in most situations in the catchment, and it is highly probable that, given a longer data collection period, it would be possible to remove observed trends, and such data could be more usefully correlated. In the first instance water-levels in the catchment were regarded as typifying recharge- responses of a common homogeneous aquifer, which has been shown not to be the case in the Narrator Brook Catchment. As a consequence of a multi-aquifer system composed of perched watertables at differing elevations, predictive equations derived from the stepwise regression procedure may not be valid for all cases in the Narrator Brook Catchment. 5:7 Conclusions In general, it has been observed in the Narrator Valley ground• water network, that changes in water-levels can be attributed to the infiltration of precipitation towards the saturated zone. Water-level fluctuations and rainfall amounts of the week prior to water-level determination are strongly correlated, and as such observed water- level changes are unlikely to be caused by air entrapment in the unsaturated zone. Water-level fluctuations appear to behave in accor• dance to seasonal trends experienced on Dartmoor, resulting in the normal pattern of higher winter levels and lower summer levels. Any lack of coincidence between water-levels and weekly rainfall totals in some areas of the valley suggested that there is a significant small scale variation in the rates which deficits are satisfied, and that some wells have lagged response time dependent upon site conditions. The main features of contrast between the observation wells appear to be the magnitude of the different responses of the water- levels to rainfall input. On this basis an analysis of hydrograph amplitude was carried out, providing a method of dividing the ground• water network up into hydrogeologlcally determined groupings. Three groups of wells were derived exhibiting different fluctuation charac• teristics which could be attributed to topographical location, and site lithology. Well groupings derived from water-level change maps, standard deviation of water-level fluctuation and analysis of variance produced a similar pattern of well groups In the Narrator Valley.

- 278 The use of the stepwise regression procedure in the Narrator Valley produced three distinct hydrogeological provinces delineated by antecedent catchment parameters common to all observation wells, in association with individual water-level fluctuations. Province 1 is regarded mainly as a recharge area characterised by Wells 1, 9, 10, 13, 14, 15, whilst Province II is predominantly a discharge zone typified by the characteristics of Wells 3, 4, 11 and 12. Province III is represented by Wells 2, 6 and 7 having anomalous fluctuation characteristics, relative to other wells in the catchment. All of the well groupings utilised show a persistent feature of a three-fold hydrogeological division of the wells in the weathered granite aquifer. Although the use of the stepwise regression procedure produces a division of the wells in the valley aquifer which corroborates with other methods used, the results must be viewed with some caution. Tests of significance on the selected regression equations, and the analysis of residuals, indicates that the use of the model in the Narrator Valley is not an outright success. It is proposed that further variables, such as parameters describing grain size distri• bution, permeability variations, thicknesses of individual materials and variations at a wellslte, incidence and duration of site flooding, variations In river stage, along with slope position and elevation, may produce more meaningful predictive equations. Hauser (1974) noted that the tentative and suggested nature of the findings obtained by using a stepwise regression procedure must be stressed, since they should not be presented in the trappings of rigorous statistical inference, which are not applicable to the tech• nique. In the attempt to isolate the important controlling Independent variables on water-level fluctuations in weathered granite aquifers, the results from the Narrator Brook Valley highlight the usefulness of the stepwise regression procedure as an exploratory technique. The groundwater contour maps produced for the south-western portion of the Narrator Brook Catchment, as presented in Section j:3:2, may be somewhat misleading. For the construction of any groundwater contour map the basic assumption Is that the aquifer is homogeneous, which Is not the case in the Narrator Brook Valley. The

279 - heterogeneity of the valley materials is illustrated by the distri• bution of clay lenses Interspersed between sands, grits, gravels and cobbles In the borehole logs In Appendix 1. Perched watertables whose distribution varies spatially and temporarily in the weathered granite, result in large variations in the depths to the watertable in the Narrator Brook Valley. As a consequence the groundwater level contour maps produced must be of a tentative nature only. From a consideration of soil moisture data collected in the Narrator Brook, in association with seasonal variations in water- levels and precipitation, the evidence suggests that recharge Is taking place to the granite groundwater system before any soil moisture deficit is made good. The soil cover, like the underlying weathered granitic materials possesses spatial variations dependent on site lithology. Any infiltrating water in the unsaturated zone is transmitted from the surface towards the watertable within a one to four week period in the valley. The variations in spring discharge with antecedent meteorological factors was examined in the case of Spring 12/13 in the catchment. Good correlation was observable between the spring discharge and the precipitation lagged by a two-week period, giving a correlation coefficient of 0.466 significant at 0.01 level. Spring 12/13 is not considered to be representative of all the springs In the catchment, but the derivation of a recession period of fourteen and a half weeks without recharge, indicates the stability and importance of the groundwater component in the Narrator Brook Valley.

280 Chapter 6 Aquifer Properties in the Narrator Brook Catchment

6:1 Introduction Quantitative groundwater investigations are an integral part of a complete evaluation of the occurrence and availability of groundwater resources* The worth of an aquifer as a source of water depends largely upon two inherent characteristics, its ability to store and to transmit water (Lohman 1972). These characteristics are referred to as the storage coefficient (S) and the transmissivlty (T). These two •formation constants' are generally considered as the physical indices of the aquifer characteristics, and are used as the basis for the theoretical prediction of the future yield of groundwater in storage (Alexander 1977). 6:1:1 Definitions The transmissivity (T) of an aquifer is the rate at which water of the prevailing kinematic viscosity is transmitted through a unit width of an aquifer for the full saturated thickness under a unit hydraulic gradient. It has the dimensions of T~^t and

Kb where b is the saturated thickness of an aquifer and K is the hydraulic conductivity. Hydraulic conductivity (K) is the amount of flow per unit cross- sectional area of an aquifer, under the influence of a unit gradient and has the dimensions of L.T.-^. Likely ranges for T and K in some of the aquifers in the U.K. are outlined in section 6:6. The storage coefficient or storativity (S) is defined as the volume of water an aquifer releases from or takes into storage per unit surface area of the aquifer per unit change in head. (S) is a dimensionless coefficient. la confined aquifers (S) has magnitudes of 10-** to 10-^, which, according to Jacob (1940) depends on the elasticity of the aquifer material and compressibility of the water. Consequently considerable pressure changes over large areas are required to produce substantial yields.

281 For unconflned aquifers (S) Is equivalent to Che specific yield (Sy). Specific yield Is defined as the percentage of gross volume of water from the aquifer released under gravity. For sandy materials (Sy) may be In the range 0.1-0.2 (Kruseman and De Ridder 1970). The specific yields of unconflned aquifers are much higher than the storatlvltles of confined aquifers. Freeze and Cherry (1981) suggested that Che usual range of Sy Is 0.01-0.30. These higher values reflect the fact that releases from storage In unconflned aquifers represents an actual dewaterlng of the pores» whereas releases from storage In confined aquifers represent only the secondary effects of water expansion and aquifer compaction caused by changes In the fluid pressure. 6:1:2 Flow Conditions In an Aquifer Induced by Pumping 1. Unsteady state or non-equlllbrlum This occurs the moment pumping of an aquifer starts until a steady state Is reached. Consequently In an Infinite, horizontal, completely confined aquifer of constant thickness pumped at a steady rate, there will always be an unsteady state as long as the changes of water level with time, due to pumping alone, are measurable. 2. Steady state or equilibrium This occurs when there Is equilibrium between the discharge of the pumped well and the recharge of the aquifer by an outside source. In practice It Is rarely achieved, but a steady state Is assumed when changes In drawdown with time have become negligible, or the hydraulic gradient becomes constant. Two types of flow equations exist In the literature; those describing steady state flow, and those describing unsteady state flow. The derivation of equations for unsteady state groundwater flow are outlined in section 6:2 but for purposes of the analysis of groundwater conditions in the Narrator Valley, steady state flow is not considered in great detail. 6:2 Groundwater Flow Equations The flow of fluids through a porous media is governed by the laws of physics. Flow is a function of several variables, and can be described by partial differential equations in which the spatial coordinates, x, y, z and time t, are Independent variables. The basic

- 282 law of flow is expressed by Darcy's equation of linear seepage given by:

V = -k' [1] dx where V = velocity k = permeability ~ = change in hydraulic gradient, dx In the Darcy flow equation (k) is a constant. •

Darcy's equation, when put together with an equation of continuity that describes the conservation of fluid mass^ during flow through a porous medium, results in a partial differential equation of flow and is derived as follows: Simplification of the continuity equation results in

av dV dV — + ^ + _i = 0 [2] dx dy dz

Substitution of Darcy's equation for V^, and yields the equation of flow for steady state through an anisotropic saturated porous medium:

dx ^ dx dy y dy dz ^ d3

The continuity equation, for transient flow in a saturated porous medium, expresses the net rate of fluid mass flow into an element volume of the aquifer as being equal to the time rate of change of fluid mass storage within the element volume. This is given as:

a(pV ) d(pV ) d(pV ) X y z dp dn ,, , + i_ + = n_+p — [4] dx dy dz 5t dt

283 - Changes In (p), water density, and change In (n) porosity, are both produced by a change In hydraulic head (h), and the volume of water produced for a unit decline In head Is (specific storage)* The mass rate of water produced (time rate of change of fluid mass storage) Is pS 5h/5t. Substituting this expression into equation [A], eliminating (p). Inserting Darcy's equation and rearranging the equation we obtain:

^).^(U3^) = S^^ [5] dx dx ay ^ ay dz d3 ^ at

This Is the equation of flow for transient flow through a saturated anisotropic porous medium* If the aquifer medium Is homogeneous, and Isotropic of thickness b, s = b, and T = kb. then equation [5] reduces to:

a^h ^ a^h ^ s ah • + + — - = — — [oj ax^ dy2 dz^ T at

The values for (S) and (T) can be determined by aquifer tests employing a variety of methods of analysis using the non-equlllbrlum theory. 6:2:1 The Development of Methods for Determining Aquifer Characteristics The Thels (1935) non-equlllbrlum theory has been utilised by numerous Investigators, predominantly In the U.S.A. and U.K. and modified to facilitate Its use under a variety of flow conditions as summarised In Table 6:1. The Thels formulae take Into account time and the storage characteristics of an aquifer and Is given as:

u 0 /« e~ S'r = -5- / — du [7] 4TCT U U where S*r = drawdown In an observation well a distance (r) from the pumped well Q = discharge of pumped well T = transmlsslvlty and (u) Is given by

284 - u = [8] 4Tt where t = time since pumping began S = storage coefficient of the aquifer. The integral in equation [7] is usually written as W(u) the well function. W(u) is evaluated from the series

2 3^ W(u) = -0.5772 - ln(u) + u — + ^— + -ii— [9] 2.2! 3.3! n.nl

From equation [8]

r2/t = ilu [10]

therefore log (Sr) = log (Q/AnT) + log (w(u)) and log (r^/t) = log (4T/S) + log (u) Q/47IT and 4T/S are constant for a given test, so the relationship between log (Sr) and log (r^/t) are similar to the relationship Log (W(u)) and log (u). The method of solution for the Theis equation is to plot Sr versus r^/t and W(u) versus (u) on separate but identical scaled logarithmic paper. The plots are then superimposed keeping the axes parallel and then adjusted until some portion of the two curves match. The coordinates of the match point are then taken and are substituted into equations [7] and [8]. This equation was developed primarily for use in confined aquifers with unsteady state flow conditions.

- 285 Table 6:1 The Utilisation of the Non-equilibrium Theory in the Development of Procedures to Determine Aquifer 'Formation Constants'

Flow conditions Locality Other features Author(s) in the aquifer

Unsteady state U.S.A. Graphical method Theis (1935K flow in confined aquifers U.S.A. Straight line plot Chow (1952) graphical methods (S), (T)

Graphical method which takes Into account the storage capacity of a large diameter well

Steady state U.S.A. Graphical method which Hantush and in semi- takes account of Jacob (1955) confined aquifer leakage

Unsteady state U.S.A. Single plot straight- in semi- line solution takes confined into account the Hantush (1961) aquifers effects of Partial Penetration U.S.A. Step-drawdown pumping Walton (1962) test analysis

Unsteady state U.K. Graphical methods Boulton (19s **-) in unconfined allowing determina• aquifers with tion of a delay index delayed yield; as well as the and in semi- U.S.A. standard (T) and (S) Prickett (1965) unconfined values. aquifers U.K. Graphical method which takes aquifer aniso- Bouiton (1970) trophy into account

Cooper and Jacob (1946) modified the Theis equation by simplifying W(u), by assuming (u) to be small. The terms in equation [9] after ln(u) were ignored, and these modifications allowed a single plot of drawdown versus time to be constructed, and a straight line solution of (T) and (S) was achieved. Cooper and Jacob's (1946) approach is applicable in semi-confined aquifers with unsteady state flow conditions. Chow (1952) also developed a straight line solution for the solution of (T) and (S) suitable for unsteady state flow in confined aquifers.

286 The assumptions on which the non-equilibrium equations are based are outlined below: (after Theis, 1935) (1) the aquifer is confined, horizontal, homogeneous, isotropic, of uniform thickness and of infinite areal extent (11) the pumping well is of infinitesimal diameter and fully penetrates the aquifer (ill) flow to the pumped well is constant, radial, horizontal and laminar (Iv) all water comes from storage in the aquifer within the area of Influence of the pumped well, and is released from storage instantaneously with decline in pressure (v) (T) and (S) are constant in time and space* Further modifications of the non-equilibrium theory have centred around overcoming the restrictions dictated by the above five assumptions, which rarely exist under field conditions. Type curve and straight line solutions were modified by Brown (1953), to use recovery drawdown data* This technique implies a constant discharge which is extremely difficult to attain In practice, but Todd (1959) co: nslders that recovery analysis provides an easy check on pumping test results* Hantush (1961) was the first to develop an unsteady flow equation in relation to the effects of partial penetration of the water-bearing horizon by pumping wells, in semi-confined aquifers. This author developed both a type-curve and straight line method for the solution of (T) and (S). The development of well hydraulics and the subsequent determination of aquifer hydraulic properties from the non-equilibrium theory was concerned primarily with confined conditions. The work of Boulton (1963) initiated the development of approximate solutions enabling the calculations of unconfined aquifer hydraulic properties. The solution of (T) and (S) In water table aquifers were attempted by Prickett (1965)* In unconfined aquifers gravity drainage is not immediate, and use of the Theis equation results in (S) which appears to vary with time. The significance of this variation will depend upon the ratio of the depth of the aquifer to the drawdown, and at what stage of pumping the readings were made. Prickett (1965), using the theory developed by Boulton (1963), developed a type-curve

287 solution which Incorporates a 'delay index* which accounts for slow drainage of the aquifer material. In most cases all aquifer units possess a leakage component which is incorporated into type-curve solutions, by the use of a 'leakage factor' (Hantush and Jacob 1955). Papadopulos and Cooper (1967) produced a graphical technique which takes into account Che storage capacity of a well if the aquifer is pumped by a large diameter well. Boulton (1970) presented a graphical solution of (T) and (S) which takes aquifer anisotrophy into account in unconflned aquifers. A numerical model was developed by Rushton (1978) to estimate transmisslvlty and storage from pumped well data la confined aquifers. Evaluation of (T) by this method is usually possible but (S) is difficult due to the problem of identifying the effective radius of the pumped well* This model can cope with aquifer boundaries, leakage, variable rates of discharge and recovery provided enough suitable data is available (Rushton 1978). Modifications of the non-equilibrium theory have been undertaken in order Co determine (S) and (T) values from recovery times of the water table as a response to a small volume of water being added or withdrawn from the well. The development of bailer and slug tests are discussed in section 6:4, and the use of recovery data is further discussed in section 6:3. 6:3 Pump Tests The determination of aquifer constants through pumping tests has become a standard step in the evaluation of groundwater resource potential (Freeze and Cherry 1979). A pumping test provides in situ parameter values which are averaged over a large and more representative aquifer volume. Information on K, through the relation K = T/b and S can be determined from a single test. Leakage factors can also be obtained from a groundwater system providing observations are made In the aquifer and in the less permeable zones in the sequence. Installation of a well network, in order to provide adequate data for pump test analysis, constituting basically of a central pumping well and more than one observation well, is prohibitively expensive. Such a network is probably only Justifiable where abstraction from the aquifer Is considered, and the test pump well can be used as an abstraction well.

288 In the case of the Narrator Valley finances restricted the number of wells installed, and the hydrogeologlcal complexity resulted in only one well at each site penetrating a site-specific set of geological conditions. As a consequence the well network at any location was not ideal for the provision of adequate data for pump test analysis. Despite less than ideal situations in the Narrator Valley a more conventional pump test was conducted to provide a basis for comparison for the aquifer constants determined using different methods in the catchment (section 6:6). 6:3:1 Pump Testing in the Narrator Valley A series of pump tests were carried out on selected observation wells in the Narrator Brook, June 1980. A small suction pump with a portable generator was utilised to pump wells 9, 10, 11 and 12, chosen on the basis of their proximity to one another to facilitate movement of the pump and generator. As these wells, with the exception of well 9, were pumped dry after five minutes of constant pumping at 0.101 1 sec-^, a recovery pump test procedure was adopted. With recovery pump tests the residual drawdown (the difference between the watertable before pumping and the water level at some measured point during recovery) at any Instant, will be the same as if the well had continued to discharge, but a recharge well with the same flow had been introduced into the system. This method implies a constant discharge (Q) which is often difficult to control accurately under field conditions. Prior to the installation of pump tubing at the base of the wells, the water level recording was taken. After pumping at a more or less constant rate for five minutes determined by timing discharge into a gallon bucket, the time was noted and the pump switched off. The water level recovery was timed at Intervals of every i minute for the first five minutes; every 2 minutes for the next hour; and a final measurement taken in the following hour. The recovery data for all wells was plotted as log S* (drawdown) versus log t (time) and the resultant curves (computer plotted) are illustrated in Figures 6:2-6:5. As the data is from recovery the water-level rise was reversed in time before being plotted against (t).

289 - The plotted curves show a similar pattern in that an elongated and flattened 'S'-shape is apparent, to varying degrees, along the four observed curves. Initially it was assumed that the wells showed a delayed yield response, and analysis using the Boulton (1964) type curves was considered. However, due to the short pumping duration it was not thought to be an appropriate technique and as Boulton (1963) pointed out owing to the delayed yield to the water table the very early time drawdown curve follows the Theis curve for an artesian aquifer. As such the Theis curve was used to determine (T) in the Narrator Valley. The short rapid rise in the early part of the recovery curves suggest the draining of the well backfill at each location. This factor needs to be considered In the light of the results from other methods and is discussed in section 6:6. The change between backfill and aquifer material in situ presents a barrier effect. This 'barrier' Is gradually overcome in the flatter part of the curves, figures 6:2-6:5, and where the latter part of the observed curve rises, water is being drawn from the surrounding formation. Hence this latter part of the recovery curve is the only useful part of the pump test data obtained in the Narrator Brook. The recovery curves were analysed using Theis non-equilibrium equation and type curve. The type curve used is illustrated in Figure 6:1. Todd (1959) suggested that the Theis non-equilibrium method is applicable for analysis of the recovery data of a pumped well. 6:3:2 Theis Method A type curve of W(u) versus u to the same scale as the observed plot was produced. Figure 6:1. This type curve was superimposed on the observed curve so that the coordinate axes of the two curves were kept in parallel. A position was found whereby most of the plotted points of the later part of the data curve fall on a segment of the type curve. An arbitrary point along such a matching section was chosen and the corresponding coordinates W(u), and (u) on the type curve; and (S') and (t) on the observed data curve for this 'match point' were determined in each of the four cases in the Narrator Valley. By substitution into equation [7J S' = W(u) as outlined 4iiT in section 6:2 a value of (T) was obtained. The storage (S) cannot be

290 - Figure 6:1 THEIS TYPE CURVE

H(U) 1E2 -

5 -

' lEl - 1 I

6

2

lEO — 1 M r

5

2

1 1 1 1 1 III I 1 1 1 1 III lE-1 - 1 1 1 1 1 III I 1 1 1 1 III

Drawdown {m)(s'} lEl 8 - 8 - 7

8 - Sa3m 5 - tm 20 minutes

ru).95 xyo"^

J I I I J l_L lEO J ' ' I ' 6 8 7 8 8 8 7 8 8 5 8 7 8 8 lE-1 lEO lEl 1E2 Time in minutes (t) Figure 6.3 Well 10 Recovery

Drawdown (m) s' lEl

I SB Wm fa 30minutes VVfli)a 2 6 -5 (i5)= 0-22x10 To OBOin/d

I I ' I I I I I I M ' I I I I I J I I I I I M lEO J I I

lE-2 lE-l lEO lEl iE2 Time In minutes (t) Figure 6:4 Well 11 Recovery

Drawdown (tn)s' lEl

a B 7 8

5 I 4 h

t 8 3Om/nur0£ 3-8 (w)-3 0x10 T. l-320/n'/rf

J L I I I lEO J I I ' ' J i I I I 8 7 8 9 6 6 7 8 B 5 8 7 8 8 lE-1 lEO lEl 1E2 Time minutes [t) Figure 6:5 Well 12 Recovery

Drawdown (m) s lEO

\

s B 6m

t • \00minutes

(^)«8-4xl0" r- 0121 in/d

lE-1 2 3 4 5 5 8 7 8 8 lE-1 lEO 1E2 Tine minutes (t) determined because the effective radius of the pumped well cannot be measured without observation wells. Table 6:2 simmiarises the T values obtained using this method. These results are further discussed in section 6:6 In the light of slug test values achieved In the Narrator Brook Catchment.

Table 6:2 (T) Values using the Theis Method of Analysis

Well T (m2/d)

9 0.880 10 0.180 11 1.320 12 0.121

During the pumping tests Well 9 was the only well which was not pumped dry. It was pumped for a further five minutes after the end of the recovery period and showed little variation in discharge rate during this time. It is considered that generally the very short duration of pumping used in these recovery tests may not have been sufficient to produce observed data plots suitable for accurate analysis by the Theis method. As the cost and complexity of aquifer tests increase, slug tests provide a cheaper and more rapid alternative in that no long pumping periods are required to evaluate (T) and (S). The slug test technique was ideally suited to conditions in the Narrator Brook for the following reasons: (1) Equipment could be transported and operated by one person in some of the more inaccessible sites. (11) As the observation well at each site was the only source to be tested, and there were no other observation boreholes to monitor the drawdown, the test was performed under suitable slug test conditions, and analysed accordingly. Such results are directly comparable with analysis of slug tests data from other weathered granite areas.

- 296 (lii) Continuous pumping from narrow wells was not possible In the Narrator Valley, so the withdrawal of a slug of water from selected wells was a more feasible proposition, 6:4 The Use of Slug Tests In Groundwater Investigations Well response tests or 'slug tests' measure the force free response of a well aquifer system to an abruptly Induced change of the water level in the well (Van Der Kamp 1976)• With a slug test (T) and (S) values are determined from the rate of recovery of the water level In a well after a certain volume or 'slug' of water was removed. Instantaneous lowering of the water level in a well can be achieved by submerging a bailer, letting the water level reach equilibrium, and' then quickly withdrawing the bailer. If the aquifer is highly permeable the water level may recover within a few minutes. Such rapid rises can only be measured accurately with sensitive pressure transducers and fast response strip chart recorders. The portion of the aquifer sampled by this method is much smaller than that for a standard pumping test. As a result the values .of (T) and (S) are site specific, in that they reflect the nature of the aquifer material in the immediate vicinity of the test well only. 6:4:1 The Development of Instantaneous Discharge or Recharge Aquifer Tests Skibltzke (1958) developed a method for determining the transmisslvlty from the recovery of the water level in a well that had been balled. The 'bailer method' is applied to a single observation of the residual drawdown after the time since bailing stopped. (T) is computed by:

S' [111 12.57 T 1 where S' residual drawdown (ft) V volume of water removed in one bailer cycle (U.S gallons) T transmisslvity t time 12.57 constant (for U.S. quantities)

297 Transmissivlty is computed by substituting in equation [11] the observed residual drawdown. The volume of the water (V) is considered to be the average amount removed by the bailer in each cycle, and the summation of the reciprocal of the elapsed time, in days, between the time each bailer of water was removed from the well, and the time of observation of residual drawdown. This is similar to the Theis recovery method for determining (T) as outlined by Jacob (1963). Ferris and Knowles (195A) further developed Skibitzke's (1958) equation for determining (T), but instead of a bailer method it was referred to as a slug test technique. The effects of injecting a slug of water into a well are identical (in the opposite direction) except for their sign, to those of instantaneously bailing out a slug of water. That is: the rate of water level decline following the injection of a slug equals the rate of water level recovery following the withdrawal of a slug of equal volume. The method proposed by Ferris and Knowles (1954) for analysing the test data is based on a solution that assumes a well of infinitesimal diameter and can be expressed as:

H/H = r 2/4Tt [12] o c

where = instantaneous head change in the well H = head in the well at time >0 r = radius of well casing in the interval over which the head c change takes place T = transmissivity t = time since the instantaneous head change. Transmissivity is determined from the slope of an arithmetic plot of H or H/H against 1/t. Later Cooper et_ £l^. (1967) presented a solution for a well of finite diameter, which has the form

H/H^ = F(P,a) [13]

where P = T^/r^' a = r S/r 2 s c r = effective radius of well s

- 298 F(P,a) = a function whose tables and graphs were presented for the five orders of a 10-^ - 10-5 (Cooper etal. 1967) 10-^ - 10-^° (Papadopulos et al. (1973) The Cooper et al, (1967) equation was based on non-steady flow to a pumped, completely penetrating well. Values for (T) and (S) are evaluated from type curves showing the best fit with experimental data (Figure 6:9). Bouwer and Rice (1976) presented theory and equations for slug tests on partially or completely penetrating wells In an unconflned aquifer for a wide range of geometry conditions. The wells for their analysis may be partially or completely perforated, screened, or otherwise open along their periphery. While these solutions were developed for unconfined aquifers using solutions developed for the augerhole and piezometer techniques to measure soil hydraulic conductivity, they may also be used for slug tests in confined aquifers if water enters the aquifer from the upper confining layer through compression or leakage, (Bouwer and Rice 1976). The slug test has not previously been widely used in groundwater studies because prior to the development of pressure transducers, the rapid water level changes associated with the recovery~rate made accurate measurements very difficult. Except for Ferris et al. (1962) standard hydrogeologlcal texts contain little Information on the methods of measurement and analysis of this technique. More recent papers. Cooper et^ al. (1967), Papadopulus et aj.. (1973), Black (1978) have remedied this situation. Cooper et al. (1967) presents a set of type curves for the solution of (T) while Papadopulus et al. (1973) reviewed methods of analysing data and provides additional type curves for data with low storage coefficients. Black (1978) provided valuable Information on use of equipment in the field, and a detailed description of the subsequent analysis of slug test data. Two of the most recently published hydrogeological texts suggest the usefulness of the;slug test technique. Freeze and Cherry (1979) considered pump tests to be widely overused and that slug tests were more appropriate and could provide adequate (T) values. Fetter (1980) suggested that slug tests are appropriate for use in low permeability materials and augered holes in soils horizons. In the light of the above discussion

299 - and the conditions in the Narrator Valley, the slug test technique Is well suited for use In this catchment* 6:4:2 Field Equipment The equipment required for the slug test is minimal when compared to that needed for a standard pumping test. The slug itself, as used in the Narrator Brook, comprised a hollow brass cylinder with a ball valve in its lower end. A full description of this can be found in Chapter 3. The ball valve in the base of the tube allows the slug to be lowered beneath the water level with the minimum of disturbance, as recommended by Ferris and Knowles (1963). The two slugs used in this Investigation produced a vertical displacement of 30 and 50 cm in the wells. Other apparatus used included a battery-operated water level recorder, described in Chapter 3, to measure the position of the static water level before the test; and a pressure-transducer capable of spanning 2 m, together with a cable which can be connected to a water-level monitor unit at the surface. This unit is connected to a pen recorder which monitors the test. The whole unit is powered by a Honda portable generator or a 24 volt battery. 6:4:3 Procedure to Perform a Slug Test 1. Water level is recorded. 2. The pressure transducer, designed to measure water level change according to the variation in head above it, in a manner analogous to the operation of an aneroid barometer, is lowered to approximately 1.7 m below the static water level. This ensures aim span in the water column with sufficient head above in which the 'slug* can be allowed to fill up smoothly. 3. The pressure transducer cable Is marked with adhesive tape at both depths of 0*7 m and 1.7 m below the level of the watertable. 4. The chart span on the recorder is then calibrated by setting zero on the Water-Level Monitor (WLM) unit with the pressure transducer at 0.7 m below the static water level, and full- scale deflection with the pressure transducer at 1.7 m below the water table. The tape positions usually require altering by a trial and error procedure until the chart can

300 FIGURE 6:6 DIAGRAM TO ILLUSTRATE THE SLUG TEST PROCEDURE AT AN OBSERVATION Water WELL Level / / Monitor / Ground Level Water Level Fixed I volume Pressure ^slug' Transducer Pi H

I I I I I I J. _ J L - -< (i)Rest lii) Positioning (iii)Re-equilibration (i\^RemovalMRe-equilibration conditions Transducer and slug of slug

+ T + + + + + + T T + + + + + + + + + + + + + -f- + + + + + + + + + + + + + + . + + .IMPERVIOUS LAYEP + + + + + + + + + DIAGRAM'^TO ILLUSTRATE IDEALISED SLUG-TEST TRACE

/ change of time scales on tracd

withdrawnl quickly I 0

position of static water level(SWL)

re-equilibration slug lowered below static water level be set to span 1 m precisely. The zero on the WLM unit can then be used to set the pen at a convenient point on the chart, usually with one of the 10 cm divisions on the chart paper, for ease of reference. 5. The slug is then lowered into a position just below static water level, taking care that it does not interfere with the transducer already in position. 6. Time is then allowed for re-equillbratlon as illustrated on the sequential diagram Figure 7:6. This time length varies depending on the response of the well under consideration. Those wells with a higher proportion of clay in the aquifer matrix took longer to re-equilibrate than those wells surrounded by coarser materials. Such responses can be monitored on the pen recorder. 7. The chart recorder is then set at the maximum speed (320 mm/sec) and the slug is withdrawn as quickly and smoothly as possible. 8. A trace is recorded as Illustrated in Figure 6:7 and the rate of return to equilibrium is determined as outlined in section 6:4:6. 6:4:4 Observation Well Choice Slug tests were carried out on observation wells 1, 6, 7, 9, 10, 12 and 14, primarily because these wells had a sufficient depth of water from the base of the screen to the watertable level in which the pressure transducer could be safely set, without possible impedence by infilling from the base of the well screens. There was no damage of the casing down the tested lengths, and consequently movement of the slug in close proximity to the transducer without snagging the cable in these small diameter wells was satisfactory* In practice the reproducibility of the test in the wells utilised was found to be good and consistent traces were produced from wells tested several times in one day, and tested again on the following way. An indication as to the representative form of traces produced can be seen in Figures 6:8a and 6:8b. 6:4:5 Derivation of Type Curves Various attempts have been made at analysing the resultant head changes in a well produced by a slug test (Black 1978). The basis of slug test analysis depends on the treatment of well column storage.

303 OGUflf 6:8a SLUG-TEST TRACES NARRATOR BROOK

Site6 2116179 HO • 0-73

change in time scale

C

I Site 6 2216179

HO-0-53 SWL-static water level positibn Site 7 2116179 Run 3

SWL HO - 0-27

SWL

• III I I cms -SWL I I I I I 1 1 1 L FIGURB 6.Bb Site 14 21/6179 Run 1 SLUG-TEST TRACES NARRATOR BROOK

H0 = 0-5m

Sitel 19/6179 Run 2

HO = 0-4m I SWL 0 static water level 0 position I

Site 7 21/6/79 Run 2 SWL-- J HO -O-aBm -SWL

cms I r T 1 1 r T r T—r T—r This may be approached In three different ways: 1. The approach adopted by Ferris (1962) is to adopt a flow equation without water storage in the observation well. This makes the analytical expression an extension of the Thels equation for a pumped well* 2. The method proposed by Hvorslev (1951) which used a flow equation with water storage in the observation well, but without aquifer storativlty. Boast and Klrkham (1971), Bouwer and Rice (1976) adopted this same approach which approximates to watertable conditions. 3. A flow equation for the redistribution of head after a sudden change, taking into account observation well storage and aquifer storativlty, (Cooper et al* 1967) which approximates to confined conditions. The flow equations mentioned in 2 and 3 above are very similar and are the most widely used. Black (1978) suggests a construction of both Bouwer and Rice (1976) and Cooper et al. (1967) type curves in order to ascertain which gives the best fit to the results obtained from the slug test, regardless of any preconception about aquifer conditions* It is very seldom that the results from a properly conducted slug test cannot be matched by the two analyses cited above. If a match cannot be obtained then either the conditions do not satisfy closely enough those involved in the analytical assumptions or some more practical deviations, such as water running slowly down to the water level occurs (Black 1978). From the comparisons of type curves of Cooper £t al., Bouwer and Rice, Ferris et_ £l^., as constructed by Black (1978), the data obtained from the Narrator Valley was, by a visual appraisal observed to best fit the Cooper et al. analysis* On this basis the type curves were constructed for this analysis as outlined by Cooper, Bredehoeft and Papadopulus (1967) and used for the determination of (T) and (S) values in the Narrator Brook Catchment. These type curves utilised range from a = 10"^ to 10"^, as Illustrated in Figure 6:9. Type curves for formations with very low storage coefficients were also plotted as recommended by Papadopulus et_ (1973), but were found to be unnecessary for the present analysis. 6:4:6 Aquifer Parameters Determined by Cooper and Papadopulus Method From the chart record of the slug test the following parameters are derived: t = time in seconds H = instantaneous change-igQ^ead (after withdrawal of slug) o FIGURE 6:9 Type curves for instantaneous change

in wells of finite diameter

a = 101-10-5 Cooper, Bredehoeft Er Papadopulos(1967)

\ c

I

B=Tt/rc2

10 10 10 10 10 100 1000 H = change in water level when t > 0. The data for individual wells are plotted on semllogarithmic paper, on the same scale as the type curves, producing graphs of t vs H/H as o illustrated on Figure 6:10. The curves so derived are matched against the type curves figure 6:9 of Cooper et al. (1967). With the arithmetic axes coincident, the data plot Is translated horizontally to a position where the data best fit the type curves as illustrated in Figure 6:10. Where the data coordinates are found to overlie the value Tt/r^^=1.0 on the type curve, a value for t is ascertained, and (T) is completed by 1.0 r 2 T = [14] where T - transmisslvlty r = radius of casing over which water level fluctuates t = time The coefficient of storage (S) is determined from:

r 2 a = ^ S [15] r ^ c where a = value of the chosen type curve rJ ^ = radius of screen or open hole

xj' = radius of casing

S = storage coefficient. The average water levels, measured in the observation wells over the period the slug tests were conducted in the Narrator Valley (June 1979), were used to derive an estimate of the saturated thickness (b) of the aquifer at a specific point. The saturated thickness had to be assumed in this fashion due to the following factors: (1) no well penetrated solid bedrock (although Well 7 hit a more dense material between 10 and 15 metres) (11) inconclusive geophysical results concerning depth to bedrock in the Valley. The depth to the borehole base (from the surface) minus the depth to the average water level in a particular well was taken to equal the assumed saturated thickness at that point. As K = T/b (m day-^) an

- 308 - FIGURE 6:10 SLUG DATA PLOTS AND ia MATCH POINTS FOR TWO OBSERVATION WELLS, NARRATOR BROOK 8 •7 2)

(m) .5^ Site 10 Run 4 15/6/79 •4 t=17sees •3 T =• 1-30m^day 2 Ho - 0-12m •1 00

10 100 IO1 t(seconds) » 8

•7 Site 9 Run 3 15/6/79 & Ho-0•49m & t-18 sees Weil Response Time M -10-* , 4- T = 1-22n^^/day "/Ho-0-37 •3-

2\

•1

-4 K>0 oc-IO Jvpe Curve 100 , j . 1000 10 t (seconds) estimate of hydraulic conductivity is thus derived for each site tested, using the (T) calculated from Cooper et_ method and the assumed saturated thickness.

Table 6:3 Showing Assumed Saturated Thicknesses of the Aquifer in the Narrator Brook Valley

Well Assumed Saturated Thickness June 1979

1 7.14 m 6 3.69 7 7.96 9 11.26 10 5.94 12 6.06 14 5.99

6:4:7 Bouwer and Rice Analysis The Bouwer and Rice (1976) technique, which is applicable to completely or partially penetrating wells in unconfined aquifers was used to estimate (K) and (T) values in the Narrator Brook Catchment. The geometry of the wells and the associated symbols used are illustrated in Figure 6:12. Values of (Re) effective radius of the well, expressed as In Re/rw were determined by an electrical resistance network analogue by Bouwer and Rice (1976). They deduced the following empirical equation relating Re/rw to the well geometry:

1.1 A+B In [(D-H)/rw] In Re/rw [16] ln(H/rw) L/rw

In this equation A and B are dimensionless coefficients that are functions of L/rw as shown in Figure 6:11. If ln[(D-H)/rw] is greater than 6, a value of 6 is chosen for the term ln[(0-H)/rw]. (This applies when the saturated aquifer thickness is greater than H, ie H > b). If D = H, the case for a fully penetrating well the following equation is used:

310 - In Re/rw = (—~ + -^^] " [17] ln(H/rw) L/rw

1 n which C Is a dlmenslonless parameter also a function of L/rw as shown In Figure 6:11. In the case of the Narrator Brook analysis when D > H equation [16] was used and when D is taken to equal equation [17] was used, to determine in Re/rw in each case. (D) is synonymous with (b), the saturated aquifer thickness. 6:4:8 Determination of K (Bouwer and Rice) The flow into the well at a particular value of y can be calculated by modifying the Theim equation to

Q = 271 KL ^ [18] In (Re/rw)

where Q = flow K = hydraulic conductivity L = screen length y = vertical distance between water level in the well and equilibrium water table in the aquifer Re = effective radius over which y is dissipated rw = horizontal distance from well centre to original aquifer. The rate of rise, dy/dt of the water level in a well after removing a slug of water can be related to the inflow Q by the equation

dy/dt = - Q/n r 2 [19] c

where nr ^ = cross-sectional area of well where water level is c rising. Combining [18] and [19] yields

311 FiGUREb-.n Curves relating coefficients A.B and C to L/^^

14

12

10

8 I AandC B I 6 + 3

/ 4 2

2 + 1

I I I I I I I 111111 I I I I I I I |i1111 I I I I I I I 111 III I I I I I I I i 1 5 10 50 100 5001000 5000 •-/rw FIGURE 6; 12 GEOMETRY OF A PARTIALLY PENETRATING PARTIALLY PERFORATED WELL IN AN UNCONFINED AQUIFER

Ground Surface h-2rcH

Water table

Developed zone bsD H around well screen

Impermeable SYMBOLS

y change in head boD saturated thickness (m) H depth from WT. to base of well(m) L length of screen (m) diameter of well casing (m) radial distance between undisturbed aquifer and centre of well (m)

-3i3 - i dt = — dt [20] y r 2 In (Re/rw) c which can be Integrated to

In y = - + constant [21] r 2 In (Re/rw) c

Applying the equation between limits at t = 0 and solving for K yields:

r ^ in (Re/rw) , y K = i In^ [22] 2L t yt

Values for (t), y^ (water level after withdrawal of slug) and yt (water level at some specified time (t) after withdrawal of slug) can be obtained directly from the strip charts of the slug tests conducted in the Narrator Valley. These values are substituted in [22] above along with In Re/rw derived from equation [17] or equation [16], and values for K are so derived. (T) values by this method are obtained by multiplying K by the assumed saturated thickness of the aquifer at a specific site, see Table 6:4. 6:5 Slug Test Parameters in the Narrator Valley Van Der Kamp (1976) suggested that a well aquifer system responds in a manner analogous to the classical mechanical system of a mass on a spring in a viscous medium. The water in a well corresponds to the mass, and the aquifer corresponds to the spring* Depending on the mass of water in the well and the hydraulic characteristics of the aquifer, especially its conductivity, the response of the water level in a well may be as an overdamped oscillator, a critically damped oscillator, or an underdamped oscillator. In an underdamped case the water level oscillates about the equilibrium level. In an overdamped case the water level returns to equilibrium level In an approximately exponential manner. The critical damping case is the transition between these two types of response. The water level oscillations for the Narrator Valley follow the case for the overdamped conditions, and as such are readily analysed by the methods outlined by Cooper et^ al. (1967).

314 The value of (a) is chosen from the type curve plots as being the curve with which the observed data points coincide. The matching of the observed plots depends upon the shapes of uniquely curved type curves, which differ only slightly when (a) differs by an order of magnitude. Hence a determination of (S) by this method has questionable reliability (Cooper et al, 1967). However, an estimate of (S) within an order of magnitude can be made from a consideration of the geologic conditions, and hence some (a) values.can be eliminated, and the appropriate type curve chosen (Papadopulus et al. 1973). The determination of (T) is not so sensitive to the matching of the curves. 6:5:1 Results Transmlssivity, hydraulic conductivity and storage values derived from the slug tests are presented in Table 6:4. Two columns of (T) and (K) values are depicted as being derived from the Bouwer and Rice method. The first column corresponds to the situation where H = b, the second column to where H > b. The variation between these two columns for the Bouwer and Rice method can be atttrlbuted to Inherent errors in the method of estimating hydraulic conductivity when H = b. If H = b then the well is deemed to be fully penetrating, which Is not the case in the Narrator Valley. Since no definitive depths to the granite basement, and hence saturated thicknesses are known, these two different assumptions were adopted in determining transmlssivity and hydraulic values. It is likely that the condition H > b holds for the Narrator Brook Catchment. Values of transmlsslvity and storage were derived using estimates of the saturated thickness being 10%, 40% and 60% larger than the average saturated thickness adopted for each site. These values are illustrated in Table 6:5 for the Bouwer and Rice analysis where H > b. It can be seen from this table that as the saturated thickness Increases, both hydraulic conductivity and transmissivity increase in a similar proportion. The variation between the Cooper et al. (1967) and the Bouwer and Rice (1976) analysis as summarised in table 6:4 is more difficult to explain. In the first instance it may be attributable to the fact that the analysis of the slug test as outlined by Cooper et^ al. (1967), was developed for use in a confined aquifer, and as such some of the assumptions made in the unconfined case, are not valid when

315 using this method. However, Bouwer and Rice (1976) pointed out that both methods are compatible, and that hydraulic conductivity or transmissivity calculated with equation [22] should be of the same order as hydraulic conductivity calculated with the type curve fitting procedure of Cooper £t al, (1967). In the light of the independent streamflow and baseflow components and their analysis, as outlined in Chapter 4, It is considered that the Bouwer and Rice values in Table 6:4 are likely to be more representative of aquifer constants of the weathered granite aquifer, in the Narrator Brook Catchment. Mean (T) and (K) values henceforth alluded to in the Narrator Valley will refer to the average of parameters derived using the Bouwer and Rice (1976) technique. An average value for (T) in the catchment is 5.42 m^ d~^ with an average catchment hydraulic conductivity of 1.106 m d"^. These average values exclude Well 7 results as these are exceptionally high. 6:5:2 Spatial Patterns Averaged values for (T) and (S) derived from the slug test technique are outlined in Table 6:4. The highest (T) and (K) were recorded at Site 7, 210.47 d"^ amd 26.43 m d"! respectively. Site 7 is situated in a low topographic position and In an area with a low hydraulic gradient as illustrated in figure 5:6 Chapter 5. From all the parameters measured during this present investigation in the catchment. Site 7 features are markedly different, suggesting different aquifer properties at this location. The high (K) and (T) values may be attributed to the interception of fissure flow in this location. Away from site 7, (T) values fall to a mean of 8.73 m^ d~^ at Site 9, and a mean value of 6.134 m^ d"^ at Site 12 (Table 6:6) (T) values become lower with increasing topographic height up to Sheepstor, and increasing clay content as illustrated by the borehole logs at these topographically higher sites. This apparent influence on (T) values is also shown on the north side of the valley where Site 6 (220.24) and site 1 (239.51 m) have (T) values 3.31 m^ d"^ and 1.72 m d-* respectively. Higher (T) values would have been expected on the hlllslope locations such as Site 1, because of the higher sand and grit content in the profile. However, the drilling effects, due

316 Table 6:4 Results of Slug Test Analysis Narrator Brook Catchment June 1979

Cooper»Bredehoeft, Bouwer and Rice (1976) Papadopulus (1967)

Aquifer Constants H => b H > b

Sites S T(m2 d-M K(m d-^) T(m2 d-1) K(m d-1 )T(in2 d-1 ) K(m d-^)

9 10-3 0.79 0.07 9.61 0.85 5.49 0.49 10-' 0.76 0.07 9.34 0.83 5.34 0.47 10-^ 1.00 0.09 9.27 0.82 5.31 0.47 10-^ 0.88 0.08 15.35 1.36 8.78 0.78 10-^ 1.22 0.11 12.03 1.07 6.88 0.61

14 10-3 0.48 0.08 1.00 0.17 0.60 0.09 10-3 0.17 0.03 0.84 0.14 0.50 0.08

12 10-^ 1.38 0.23 7.70 1.270 0.60 , 0.09

10 10-3 0.82 0.14 not suitable for this analysis 10-5 1.30 0.22

1 10-^ 0.34 0.05 2.17 0.30 1.276 0.18

7 not suitable for this analysis 266.03 83.42 155.28 19.5 265.58 33.36 155.02 19.5

6 10-^ 0.11 0.03 7.17 * 1.94 4.386 1.19 not suitable for this analysis 1.06 0.29 0.652 0.18

Mean 10-3 0.77 0.098 46.70 5.83 27.236 3.40

Mean excluding Well 7 6.868 0.82 3.979 1.39

General catchment ranges T = 5.42 -- 36.96 m2 d-1 K = 1.11 - 4.61 m d-^

- 317 Table 6:5 Changes in (K) and (T) with saturated thickness of the aquifer 10%, 40% and 60% larger than the estimated average value of (b) at each site

Bouwer + Rice (1976) method H>b K and T values when saturated thickness of aquifer are at certain percentages larger than estimated (b)

Sites 10% 40% 60% K(m/d) T(m2/d) K(m/d) T(m2/d) K(m/d) T(m2/d) K(m/d T(m2/d)

Site 9 b=l1.26m 0.17 1.91 0.17 2.60 0.16 2.60 0.16 2.95

Site 14 b= 5.99m 0.78 0.78 0.78 9.72 0.80 12.56 0.80 14.48

Site 12 b= 6.06m 0.75 4.57 0.76 5.06 0.78 6.58 0.78 7.58

Site 1 b= 7.14m 0.18 1.28 0.18 1.42 0.18 1.85 0.19 2.12

Site 7 b= 7.96m 19.50 155.28 19.66 172.08 20.09 222.82 20.22 257.47

Site 6 b= 3.69m 1.19 4.39 1.20 4.86 1.23 6.33 1.24 7.33

to the smearing of clay over the drilled surfaces may well account for the lower values experienced.

Table 6:6 Aquifer hydraulic properties of selected sites in the Narrator Brook

Topographic Sites T(m2/d) T(m2/d) K(m/d) K(m/d) Height Range Mean Range Mean

248.09 m 14 0.50 - 1.0 0.73 0.08 - 0.17 0.12 242.44 m 12 4.56 - 7.70 6.13 0.75 - 1.27 1.01 239.51 m 1 1.28 - 2.174 1.73 0.18 - 0.30 0.24 237.62 m 9 5.31 - 15.35 8.73 0.47 - 1.36 0.77 228.33 m 7 155.02 -266.03 210.47 19.47 -33.42 26.43 220.24 m 6 0.65 - 7.17 3.32 0.18 - 1.94 0.892

The occurrence of an Increasing clay content in the lithologles around the screened sections of the observation wells, will Influence the (S)

- 318 and (T) values. Clay lithologies will impede flow and hence lower (T) and (K) values in a given locality. 6:5:3 Slug test traces Some comment should be made at this stage concerning the traces derived from the slug tests conducted. Data curves from the slug tests were fitted with the Cooper et al. (1967) type curves as described in Section 6:4:5. Those which were not suitable for analysis by this method showed anomalous recovery curves as is the case of Well 7 (Figure 6:8b) and Well 6. In the case of Well 7 response to the withdrawal of the slug was almost immediate, (within 15 seconds), and consequently the H/Ho vs. t curve was very short and unsuitable for the Cooper and Papadopulus method of analysis. Most of the traces unsuitable for the Cooper and Papadopulus analysis could be Interpreted via the Bouwer and Rice technique as outlined in Section 6:4:7. The following Table 7:7 illustrates the range of re-equllibrium times for the individual wells:

Table 6:7 Re-equlllbrium times for slug tested wells in the catchment

Well Re-equilibrium time (seconds)

1 400, 400 6 380, 140 7 22, 27, 12 9 160, 75, 180, 200, 165 10 62, 50, 60 12 180, 150 14 1000, 1500

Well 7, and Well 10 show the smallest values of <70 seconds, and the largest. Well 14, has values >1000 seconds. Well 6 showed some Irregular fluctuations about equilibrium level thought to be the effect of the reservoir oscillations nearby. As a result of this a majority of Well 6 traces were unsuitable for any analysis.

319 - The sharp rise shown on the traces immediately before the main change due to the withdrawal of the slug, may indicate the uprising and subsequent falling of a column of water stored above the slug whilst immersed below the static water level. By a trial and error method, such a peak on the trace may be subdued by decreasing the head of water above the submersed slug. In cases where such a peak cannot be so eliminated, it may represent immediate storage effects of the formation stabiliser. Although as previously pointed out, the condition H>b may hold for the Narrator Valley, no partial penetration corrections for the wells have been made. Partial penetration adjustments are usually only applicable to data from pumped wells and observation wells in an aquifer with pumping periods extending from a few hours to several days (Hantush 1962). The duration of the slug test falls so far short of these constraints as to make such adjustments invalid. 6:5:4 Applicability of results The duration of a slug test Is very short hence the estimated (T) determined from the test will be representative only of the water• bearing material close to the well, (Ferris et al. 1962). Vertical permeabilities of most stratified aquifers are only small fractions of their horizontal permeability (Cooper et al. 1976), therefore the induced flow within the small radius of the cone that develops during the short period of observation is likely to be essentially two- dimensional. Consequently the determined transmissivity approximates the transmissivity of that part of the aquifer In which the well is screened. If the aquifer values refer strictly to the conditions Immediately around the well, then care must be taken in interpreting the results. Conditions in the near vicinity of the well have invariably been altered to some extent by the drilling procedures. Black (1978) noted that where holes were drilled using a mud-flush rotary method, slug test results gave K values lower by more than a factor of ten, than the average value for the aquifer as a whole. This was taken to infer significant mud invasion giving rise to a substantial 'skin effect*. The compressed air flush used in the down- the-hole hammer technique utilised in the Narrator Valley, is not considered to have such a substantial effect, although some smearing

320 of drilled surfaces in the hole as clay and water mixes were blown out is likely. 6:5:5 Observation well effects: In slug tests, flow to or from the well, and therefore the rate of change of water level and its induced pressure difference, is governed by the performance characteristics of the observation well. In groundwater investigations the use of observation wells range from long-term measurement of natural groundwater fluctuations to short- term monitoring of aquifer tests. The fact that observation wells themselves Introduce errors was first recognised by Hvorslev (1951) and recently expanded by Black and Kipp (1977). Stallman (1965) using an electrical analogue model, briefly considered possible response delays of water levels within observation wells during pump tests. For long term observations fluctuations can be approximated by a cyclic change about a mean water level. Hvorslev (1951) showed that with an increase In well response time the amplitude of any cyclical fluctuation is reduced and the timing altered. Observation well effects during aquifer tests are similar in that the magnitude of the effect Increases with increasing well response time. The slug test can provide a measurement of well response time using the concepts of Hvorslev (1951) as used by Black (1978), Black and Kipp (1977). When

lo = the well response time can be obtained from the graph H/HO vs. t as illustrated in Figure 6:10. The following table 6:8 illustrates clearly determinable well response times for the Narrator Valley.

321 - Table 6:8 Well response times

Well Response time recorded Average Response time (seconds) (seconds)

9 44, 60, 60, 60 56 10 50, 40 45 12 43 43 1 180 180

These response times are very small when compared to values of 440, 740, 880 and 2830 minutes as reported by Black (1979) for the Carnmenellis granite, Cornwall. The suggestion is that in the Narrator Brook the observation well response times are so rapid that in the case of Wells 6, 7 and 14, these are not measurable. Black and Kipp (1977) noted that well response effects are inversely proportional to the aquifer storativity so that they are unlikely in unconfined aquifers. 6:6 Comparison of Results from Differing Methods in the Evaluation of Aquifer Constants in the Narrator Valley The pump test carried out in the Narrator Valley on selected wells provided (T) values. Table 6:2, which have little in common with values derived from the slug tests. Table 6:4. Well 9 is an exception to this generalised case as its recovery test value falls within the range suggested by slug test estimates. During the pump test Well 9 was the only well which was not pumped dry suggesting a larger source area than Wells 10, 11 or 12. It is considered that the short duration of the pumping time possible in the observation wells, before recovery occurred, was not sufficient to provide suitable data from which to derive reliable hydrogeological parameters. This factor highlights the valuable nature of slug testing in the Narrator Valley aquifer. However, the main limitation on slug tests is that they are heavily dependent on a high quality well-intake section. If the well screen is clogged measured values may be highly inaccurate. Freeze and Cherry (1979)

322 proposed that if the well has been developed by surging or backwashing prior to testing, then the measured values may reflect the increases in conductivities in the artificially induced gravel pack around the screen. In contrast Black (1978) noted that K values can be lower by a factor of ten, where mud invasion has taken place over the drilled surfaces in a well. On this basis it is recognised that there are some severe limitations in the use of slug tests in the Narrator Brook aquifer. Subsequent interpretation of results so derived must be viewed with caution. Direct evidence on the storage effects from the formation stabiliser are available from the pumping tests. The short rapid rises in the early part of the recovery curves may suggest the draining of the well backfill at each location. If the storage effect of the formation stabiliser is large in proportion to the slug of water withdrawn, then aquifer constants are more likely to be representative of the hydraulic properties of the formation stabiliser and not immediately attributable to the aquifer characteristics. In the published literature available on slug testing no reference is made to the volumes of slugs used in each case, or the appropriateness of a known volume to specific site situations. In the narrow wells in the Narrator Brook Catchment a larger slug volume may produce well responses more characteristic of the aquifer material. If the volume of the slug is large enough it may incorporate storage effects from the formation stabiliser, and hence produce a more characteristic well response of the aquifer materials. In the Narrator Valley two slug sizes were utilised at each site slug tested, one of 150 ml and another of 50 ml capacity. The well response of each site for these two volumes showed very little difference, probably because the capacities were not of a large enough magnitude. Due to physical constraints of the observation well sizes in the Narrator Valley, the use of larger volume slugs was impractical, and time considerations curtailed further developments required to evaluate this approach. It is felt however that the variations of aquifer constant determinations in relationship to slug volume utilised, merits further Investigations in weathered granite regions.

323 - Problems arise in the interpretation of pump test data due to the non-uniqueness of time-drawdown responses. Very similar time-drawdown plots can be obtained in unconflned, semi-confined and leaky situations. The fact that a chosen theoretical curve can be matched to the pumping test data does not prove that the aquifer fits the assumptions on which the type curve was based. This is also a feature of slug test analysis when using Papadopulus et al. (1933) technique for estimating storage. Transmissivtty values determined by this technique are not so sensitive to the matching of curves as are storage values, and hence may be regarded as the more reliable determination. Through the use of slug tests it is possible to ascribe to a specific well local values of (T) and (S) which will, according to Black and Klpp (1977), probably differ from the values for the aquifer as a whole. Papadopulus et al. (1973) suggest that a large number of such point transmlsslvity and storage values are often of greater use than a single value of transmlssivity obtained from a long-term pumping test. Because of the hydrogeologlcal complexity of the water• bearing zones in the Narrator Brook valley, results from a long duration pumping test would only be applicable to a small area contributing groundwater at the location of a pumping well. More numerous slug tests carried out at a variety of locations In the valley, representing the spectrum of hydrogeologlcal materials present, are likely to provide a more realistic range of results in the Narrator Brook. Bearing in mind the associated 'skin effects' Induced in wells by the drilling technique employed, it is felt that slug tests are more likely to provide representative values for transmlssivity and storage in weathered granite aquifers, and the Narrator Brook Catchment in particular, than standard pumping tests. 6:6:1 Discussion From the analysis of the slug tests In the Narrator Brook, it can be seen that the results obtained, by using the Bouwer and Rice watertable analysts, provide aquifer constants compatible with conditions implied from groundwater recession analysis in Chapter 4. These suggested that parts of the weathered granite aquifer had a large storage capacity with rapid through flow times. In the first instance the data plots did not appear to indicate a water-table

324 - response, and so the confined type-curves of Cooper et al. (1967), were utilised. The resulting aquifer constants from this analysis in particular (K) and (T) values, were lower than expected, consequently the Bouwer and Rice analysis was deemed to be more appropriate for use in the catchment. A moderately permeable aquifer may have a hydraulic conductivity value of 1 m day^ (Bouwer and Rice 1976). The Otter sandstone, a fades of the Bunter sandstone in East Devon has a hydraulic conductivity value (determined by long-term pumping tests and slug tests) of 1 m day-^ (Jones 1979, pers. comm.). By comparison with this, and Table 6:9 Illustrating hydraulic conductivity values of representative aquifer materials, (U.S. Dept. of Interior 1977) the ranges of hydraulic conductivity values for the Narrator Brook are in the moderate permeability category. (T), (b) and K values for some of the major aquifers in the U.K. are found in Table 6:10. Included in this table are ranges for aquifer constants determined in valley-sand and gravel aquifers during water resource potential investigations. These aquifers have since been developed and contribute to public water supplies, suggesting that their ranges of (K) (T) and (b) values are indicative of their suitability for such purposes. In the light of these values the downstream section of the Narrator Brook valley aquifer under investigation, looks promising from a potential water resource viewpoint.

•feble 6:9 Cbnparlson of permeability and representative aquifer materials, U.S. Dept. of Interior (1977)

MEnSKS^/teCEK^/D/ff (m/day) irf* 10^ 102 10^ j icn 10-3 icr^ 10-5

RELATIVE PER^EABIL^Y Very hi^ High Moderate Low Very Low

Clean gravel - Clean sand and - ELne sand - Silt, clay and mixtures - ^fassive sand and gravel of sand, silt and clay

Vesicular and scorloceous - Clean sandstone - laminated sandstone - ffasslve basalt and cavernous and fractured shale, nudstone Igneous limestone and dolanlte igenous and and metamrphlc metaiorphlc! rocks rocks

325 Table 6:10 (T) and (K) values from water resource investigations in the U.K. in unconfined strata

Formation T(m2 d-^) K(T/b) (m d-^)j b(m)

1 1 8.6 X 10-^-4.8 14-39.5 Corallian aquifer 0.45-190 limestone & sandstone Vale of Pickering Reeves et al. (1978) Bunter sandstone 95.5-433.14 — 90-125 Nottinghamshire Alexander (1977)

Fine sand aquifer 150-530 17.93-35.44 4.6-26.7 S.W. Netherlands De Rldder and Wit (1965)

Hampshire Chalk 43-5050 0.43-168.33 15.100 Headworth (1972)

Yorkshire Chalk 1000-2200 — maximum 420 Foster + Milton (1976)

Yazor Brook 425-2670 100-465 4.12-7.35 Gravel aquifer Hereford W.W.A. (1975)

Trent Alluvial 90-970 16.36-230.95 4.2-5.5 Gravel aquifer Broadhead and Mackey (1972)

6:6:2 Comparisons with slug test results in other areas With two notable exceptions, (Black 1978, 1979; Glendining 1980) the results of slug tests in specific areas have not been widely documented in the literature. Tests have been carried out on parts of the east Devon Bunter aquifer (Tubb and Jones pers. comm. 1979) and in isolated cases in site investigations for waste disposal (Redland Purle pers. comm. April 1980). Those published investigations have fortuitously been conducted in granite regions of Cornwall and

- 326 - Scotland. Results so obtained are presented here for a comparison with those in the Narrator Brook Catchment, Table 6:11. The high (K) values in the Carnmenellis granite may well be related to discrete areas of fissure flow contribution at varying depths in the coherent granite mass. In general the Sco^sh granite has a much higher porosity, often greater than one per cent, compared with the Cornish granites (Glendining 1980). In the Caithness area where these slug tests were carried out, the granite was weathered to depths greater than 40 m. The K values for the Scottish granite in the weathered sections of the boreholes are of a similar magnitude to those in the Narrator Valley, both being associated with a weathered granite matrix.

Table 6:11 Slug test results from two other granite areas in the U.K. 1) Carnmenellis granite

Tested length Average values for whole open borehole b (m) S T(m^ d-) K(m d-^) X 105

286 5 X 10-** 0.029 10.1 274 5 X 10-5 0.029 10.6 287 5 X 10-5 0.022 7.7 253 5 X 10-5 0.009 3.6

The above boreholes were 300 m in total depth,

- 327 - 2) Caithness (Altnabreac) granite

b (m) S T (estimated Kxb) K(m d-^) (m2 d-^)

31.84 0.35 28.33 0.89 31.02 0.28 44.66 1.44 37.02 0.11 5.55 0.15 39.70 0.13 0.79 0.02 40.0 0.05 1.2 0.03 40.0 0.59 17.2 0.43 40,0 - 10.0 0.25 40.0 0.07 8.0 0.20 40.41 0.13 11.71 0.29 40.0 - 13.6 0.34 40.0 - 10.4 0.26 38.29 - 8.80 0.23 299.15 0.35 200.4 0.67

3) Narrator Brook, Dartmoor granite

Site Tested length Average values for whole open borehole b (m) S T(m^ d-^) K(m d-1)

9 11.88 10-^ 6.35 0.56 14 6.80 10-3 0.55 0.09 12 6.75 10-^ 4.56 0.75 10 6.65 10-^ 1.06 0.17 1 7.26 lo-"* 1.28 0.17 6 4.85 10-3 4.38 0.68

6:7 Specific Yield Determinations In the Narrator Valley Specific yield (S.Y.) as defined in Section 6:1, is a function of the size and number of Interconnected voids in a material. Values

328 - range from 0.02-0,30 for unconfined aquifers to 0.00001-0.001 for confined conditions (Walton 1970). Because specific yield represents the void space that will yield water to wells and is effective in furnishing water supplies, it is also a measure of effective porosity (Johnson 1976). The less uniform, finer grained and more dense the material the smaller the (SY). The sum of (SY) and specific retention (SR) (that portion of water retained in the rock against gravity) equals the porosity (0) of a material. As the texture of the material becomes coarser, and by implication, the importance of larger interstices increases, the (SR) and total (0) decrease while specific yield increases.

It is possible by matching rainfall events over short winter periods (where evapotranspiration is insignificant compared to rainfall) to fluctuation of water levels, to deduce the specific yields of observation wells. (SY) may be obtained from:

where ERF = effective rainfall AWL = change in water level over time period considered. Inspection of the groundwater data collected in the Narrator Valley showed the period 31-12-79 to 7-1-80 to represent a distinct recharge event. This was evident from the increase in precipitation and the rapid change in the elevation of the water table from the previous week's value. The water levels from this period was used to estimate (SY) in the above manner. Total effective rainfall (total rainfall - potential evapotranspiration) = 263 mm and soil moisture deficit (SMD) = zero. Recharge was assumed to be the change in water level at each site and these values with the associated storage values are summarised in Table 6:12.

329 Table 6:12 Storage values for selected observation boreholes

In the Narrator Brook

Well Change In Storage/specific yield water level (m)

9 0.79 0.31 15 1.23 0.21

14 0.81 0.32 12 0.67 0.39

10 0.50 0.47 7 0.96 0.27

2 1.26 0.21 1 0.89 0.29

Mean 0.30

Boreholes with the water level closest to the surface give higher values of (SY). This is not unexpected since the porosity of the weathered materials nearer the surface arc likely to be the highest in the catchment. Wells 6, 11, 3 and 4, the shallowest boreholes, have been omitted from the (SY) calculations, since these sites were in fact waterlogged during this very wet period of late December 1979. Site 13 has also been omitted as damaged casing gives erroneous values.

The aquifer in the Narrator Valley contains silts and clays which greatly change the amount of infiltrating water by intercepting or concentrating the flow. These changes, which are Important in calculation of (SY) cannot be reliably estimated in the Narrator Brook, so this method can at best only be a gross estimation of (SY). Specific yields of 15-30% in an unconfined unconsolidated aquifer^-^^'^ quite reasonable, (Kruseman and De Ridder 1976) and as can be seen from Table 6:12 values for the Narrator Valley are of comparable magnitude with a mean of 0.30.

330 Some higher values are evident amongst the observation wells in the Narrator Brook aquifer. These values can be explained with reference to the llthologies at each site as shown by the borehole logs in the appendices. Wells 9, 14, 15, despite having their screened section in a clay show a surprisingly high (SY)." This may be related to varying thicknesses of sandy materials above the clay band and to the nature of the well back-fIdling. Well 15 shows the lower value in Table 6:12, and is probably related to another clay band 3 m above the screened section.

It has already been commented that in general water level fluctuations can yield valuable Information about aquifer characteristics and dominant hydrogeologlcal conditions, and this is evident in the case of the Narrator Valley. With clay-based constituents in an aquifer a low hydraulic conductivity is usually expected, so It is surprising that the water levels in the clay-based wells respond quickly to rainfall. This characteristic would lead one to suppose that the clays penetrated are only of a limited lateral extent, a feature of a clay lens, and hence the rapid response of such wells is due to contributions from the more porous materials present in the valley aquifer.

Such observations are applicable to Wells 10 and 12, but with Well 7 the (SY) Is lower by comparison which may be attributed to llthologlcal site conditions. Site 7 only possesses 2 m of clay in the borehole log, but the likelihood of 3.5 m of granite at its base may account for the slightly lower value, as the denser and more compact the material the lower the (SY). Wells 1 and 2 give slightly lower than average values for (SY) which Is surprising in view of their llthologies. Both these sites exhibit a larger proportion of sands, gravels and granite blocks down their profiles, however the variation in material size may partly account for this variation in (SY) in these two cases.

Another source of variation in SY/storage values in the Narrator Valley may be due to the nature of the well backfilling at each site. The back-fill, although of previously drilled material, is likely to have a higher effective porosity value (depending on grain size distribution) and hence higher specific yields. Storativity values, when compared to storage values determined from the slug tests

331 - (10-^ to 10-^) are higher. These storatlvity values are estlmaed from a single measurement in time, while those derived from slug tests are obtained via the analysis of the aquifer response over a specified time period and may be more representative. Cooper et al. (1967) however comment that the determination of (S) by the slug test method has questionable reliability.

If water levels in the observation wells In the Narrator Valley represent rapid infiltration through the well back-fill at each site, then it would be expected that any water level fluctuations would be of the same magnitude of response in all wells* As can be seen from inspection of the well hydrographs in Chapter 3 and rainfall totals this is not the case. The groundwater observation well-network can be divided up into three hydrogeological provinces, based on water level variation within such groups, further indicating that water levels experienced at individual sites are the response of the local or regional watertable at that site and not totally a response to more permeable backfill characteristics of the well. However the higher storativity values may be in part attributed to back-filled sections of the boreholes in the Narrator Valley. 6:7:1 Storage in the Narrator Brook Aquifer

Permanent storage in an unconfined aquifer is defined as the specific yield multiplied by the saturated thickness. Under confined conditions the volume of water released is controlled by the appropriate coefficient of storage. Bouwer (1978) refers to the amount of water that an unconfined aquifer can store per unit rise in watertable per unit area, as fillable porosity.

In the Narrator Valley no definitive value has been determined for the depth and disposition of the weathered granite alluvial deposits, so any storage volumes are tentative as the saturated thickness is an estimated value. Assuming an average depth to the watertable over the catchment of 3 m; and a thickness of weathered and unconsolidated material in the order of 15 m; then the saturated thickness Is 12 m. With a mean storage value from the slug tests as 10"^ then storage over the catchment area found by:- outcrop area x saturated thickness x specific yield giving:

- 332 4.75 X 10^ m^ X 12 m X IQ-^ = blO-x. 10^

of water held in storage in the weathered and unconsolidated materials of the Narrator Brook aquifer. From water balance considerations as outlined in Chapter 4, it was estimated that 557 x 10^ m^ year"^ is recharged to groundwater reserves. The estimated volume of water held in storage is -^.-,.ic.r that recharged annually to the weathered granite aquifer. 6:8 Conclusions The slug test has proved to be a useful means of determining point (T), and to a lesser extent (S) and (K) values in the Narrator Brook. Due consideration must be given to the modification on aquifer constants, brought about by drilling and completion procedures, when attempting to interpret derived (T) and (S) values. Because of the physical constraints of the boreholes small diameters, it remains the only suitable technique for the estimation of aquifer constants in this area. Comparison of Table 6:4, the slug test results for the Narrator Valley, with Table 6:10, values derived from water resource investigations in the U.K., show the Narrator Valley to have (K) results in l.U to 4.61 m day-^ range. This is of an intermediate standing when compared to other aquifers in Table 6:10, and indicates that the water resource potential of the weathered granites^ in the Narrator Valley may be significant.

The location of the well screen at each site in the catchment may be in an unsuitable position with respect to the aquifer materials surrounding it. As such, aquifer constants derived may be uncharacteristic of the low clay content water-bearing zones in the weathered granite aquifer, and may incorporate formation stabiliser values. Subsequent infilling of the observation well due to influxes of finer material through the well screen may also modify (T) and (S) values obtained.

Since the aquifer is hydrogeologically complex, derived (T), (S) and (K) values in Table 6:4 form only a tentative guide as to the water-bearing and yielding properties of the Narrator aquifer. The general range for the Narrator Brook Valley aquifer of (T) and (S) and (K) values are as follows:

333 S T K

10-5 _ 5^42 - 36.95 m^ d'^ 1.106 - 4.61 m day'^

An estimate of permanent storage in the unconfined aquifer of the Narrator Valley obtained 570 x 10^ m^.

For the purposes of this groundwater investigation the valley aquifer was regarded as being unconfined. Data plots for the slug test analysis appeared, in the first case, not to indicate a watertable response since they best-fitted the confined type curves of Cooper et_al. (1967). However, this cannot be taken to imply conditions other than unconfined, for Black (1978) suggested fitting the data to best-fit curves irrespective of preconceived ideas as to the nature of the aquifer. Results from stream discharge analysis. Chapter 4, would suggest that in parts of the catchment, the aquifer is composed of weathered granite with a high storage capacity and the ability to transmit water rapidly. In the light of this evidence, the results from the confined method of analysis were deemed to be unrepresentative of gross catchment aquifer contants.

Response times of the observation wells used during the slug test technique seem to imply unconfined conditions at these locations. However, anomalous behaviour of part of the aquifer system was observed in December 1979/January 1980, at Site 11, where, as a response to particularly heavy rainfall conditions, water was observed to have been shooting out of the breather hole at the well head, under pressure. Prior to this situation, the watertable in this well and In Well 3, was observed to have been above ground level on several occasions. This implies, certainly in the case of Well 11, semi- confined conditions which are manifest under extreme recharge events.

The variation in aquifer constants with topographic location would seem to support the idea that the observation wells In the Narrator Valley penetrate discrete water-bearing horizons. These horizons have differing lateral extents within the weathered granite aquifer, and hence varying contributory areas. From the experience of pumping tests in the Narrator Brook Catchment, Well 9 should appear to have a larger contributory area, than any of the other wells tested by this means. Well 9 was not pumped dry and has, besides Well 7 in

Table 6:6, the highest mean (T) value, of 8.73 m^/d, measured.

334 The comparison of aquifer constants derived in the Narrator Valley, with those from aquifers developed for water supply purposes. Table 6:1 is encouraging in the first instance. However, the Narrator Valley appears to consist of a number of perched zones, whose supply is very small. Groundwater supplies from the main groundwater zone in the Narrator Valley, assumed to be from depths greater than 15 m, may be a more promising resource. It is felt that data from the shallow observation well depths, used in this present investigation, is not adequate to describe and attempt to quantify deeper sources in the Narrator Brook aquifer.

Based on the foregoing discussion, it is suggested that for the most part the aquifer in the weathered granite is unconfined. Semi- confined or confined portions are likely to be the result of the lateral extent of more impermeable materials, confining more porous sand and gravelly deposits in the valley.

- 335 Chapter 7

Chemistry of Groundwaters in the Narrator Brook Catchment

7:1 Introduction

This chapter presents the generalised groundwater chemistry and associated surface water chemistry of the Narrator Brook Catchment as deduced from a weekly sampling programme carried out during the research period. This project provided an opportunity to sample groundwaters at various depths in the catchment. Previously, only stream and springwater chemical analyses had been undertaken in the Narrator Brook Catchment along with measurements of the variations of solutes from interflow horizons (Ternan and Williams 1979). .All the previous chemical analyses of springwaters are from locations upstream of the established well network used in this study.

Measurement of groundwater chemistry forms one component of a general model of groundwater flow systems and in conjunction with other techniques utilised In the weathered granite aquifer (Chapters 4, 5 and 6), will facilitate assessment of water resource potential of the Narrator Catchment. The aims may be stated as follows:-

(1) to present and explain groundwater and surface water

chemistry variations, and to use these to assist inter• pretation of the groundwater flow system in the Narrator Brook Catchment (11) to determine the rate of water movement towards the

watertable in the Narrator Catchment and determine recharge- response rates (ill) to identify the most useful parameters in describing ground• water movement in the weathered granite aquifer of the Narrator Catchment which may be of wider application to other areas 7:1:1 Ionic Content of Natural Waters The principal ions in natural waters are calcium (Ca"*"^) magnesium (Mg+^)^ sodium (Na"*"), potassium (K+), bicarbonate (HCO3), carbonate (CO3 ), sulphate (SOi^), chloride (CI ), and nitrate (NO3). Iron (Fe), aluminium (Al) and silica (Si02) may also be present in

336 significant amounts, and these constituents may also occur in colloidal form. Seasonal variation can be observed in the ionic composition of groundwaters which can be attributed to varying inputs from the atmosphere, seasonal variation in vegetation nutrient demand, evapotranspiration processes increasing concentrations of soluble material in the soil zone, and in the winter season 'flushing out' such acumulatlons, seasonal variation due to weathering e.g. resulting from higher production of carbon dioxide in the warmer months.

The significant factors determining the chemical nature of groundwater in an area are the chemical composition of precipitation, the geochemical composition of the soil and regolith through which the precipitation Infiltrates, and the chemical processes operational in this layer, along with the chemical composition of the aquifer itself. 7:1:2 Origin and Source of Groundwater Constituents in Granite Areas In previous measurments of solutes in springwaters, streamwaters and interflow waters in the Narrator Brook Catchment, upstream of the well network, the following ranges of chemical elements were found (Williams pers. comm. 1978). pH 4.2 - 5.3 Na+ 4.5 - 7.0 mg l'^ K^- 0.8 - 2.0 mg 1-^ Ca+2 1.35-2.0 mg 1-1 0.54 - 0.95 mg 1"^ S^Oj 1.3 - 6.0 mg 1-1 Cl~ 6.8 -10.0 mg 1~1 Electrical Conductivity 75 - 116 \iS/cm Bicarbonate in streamwaters was frequently less than 2 mg 1"^, and in other sources often less than 0.1 mg 1"^ and was therefore considered to be insignfleant, (Williams pers. comm. 1978).

Silica, chloride, potassium and sodium were measured syste• matically in the Narrator Brook Catchment from the well network as described in Chapter 3. The origin and source of these four constit• uents in groundwaters in granite areas are reviewed in relation to conditions in the Narrator Brook Valley. 7:1:3 Silica

Except for oxygen, silicon (Si) is the most abundant element in

aerial distribution and bulk composition in the earth's crust. The

337 dominant dissolved silicon species in natural waters is Hi|S10if (monomeric silicic acid) in the pH range of 6-9, (Krauskopf, 1956). The silicic acid is essentially nonionised and silicate ions are thought to be present in appreciable amounts only above a pH of 9* Silicon concentrations in natural water are generally expressed as Si02 (Silica), and quartz solubility is only 6 mg 1"^ SiOj at 25**C (Morey et al. 1962). There is evidence to suggest that a non• crystalline form of SIO2 (amorphous Si02) rather than quartz controls the solubility of SlOj in water. The solubility of this amorphous silica, which increases with temperature, is approximately 115-140 mg 1*^ at 25*C (Krauskopf, 1956).

Based on the solubility of amorphous silica and the abundance of quartz in the hydrogeological system, silica might be expected to occur in major concentrations in natural waters. This is not the case. Silica concentrations range from less than 0.1 mg 1~^ in precipitation to 4000 mg 1"^ in mineral springs, (Davis 1964). In most groundwaters it is usually only the fourth or fifth most abundant dissolved consituent. Davis (1964) compiled thousands of groundwater analyses from many areas in the United States, and found that dissolved Si02 was typically in the range from 10-30 mg 1~^ with an average of 17 mg 1"^. Groundwater is therefore almost invariably greatly undersaturated with respect to amorphous silica. Davis (1964) reported average values for streamwater at 14 mg 1~^ and noted that silica concentrations in natural waters are marked by their lack of variation in comparison with other major constituents.

Quartz and amorphous silica generally do not exert an Important influence on the level of silica in groundwaters, as most of the silica In natural water Is probably derived from the decomposition of other silicate minerals such as feldspars, rather than from solution of quartz, (Feller and Klmmins, 1979). Alumino-silicate minerals such as feldspars and micas produce solid and dissolved silica phases as a by-product of their dissolution. Some of these are illustrated in Table 7:1 after Freeze and Cherry (1979). The lack of mobility of silica in the hydrosphere can be attributed then to the slow rate of solution of certain natural silicates and to the relatively low solubility of silica compounds in water.

- 338 Table 7:1: Reactions for Incongruent Dissolution of Some Alumlnosilicate Minerals after Freeze and Cherry (1979)

Gibbsite- Al90:^.3H?0 + 2 Si(OH)^ = AloS±70^(0]i)u + 5H2O Kaolinite

Na-montmorillonite Nan * rt^Alo o^Siq .^-r01 n (0H)o + -j + HJO = -Kaolinite j Al7Sl90c;(0H)u + -j Na+ + 4/3 Si(OH)i,

Illite-Kaolinite KQ.6Mg0.95Al9.3QSi3.50l0(0H)9 + + || HjO ^ Al9Si90s(0H)u + f K"*"

Blotite-Kaolinlte KMg:^AlSirtO] n(0H)9 + 7 + i H2O = iAl9Si?05(0H)u + K+ + 3 Mg2+ + 2Si(0H)i^

Albite-Kaollnite NaAlSlgOa + H+ + "I H2O = iAl?Si90s(0H)u + Na+ + 2Si(0H)4

Microcline- KAlSi^Ofl + H^ + -j H2O = Kaolinite , iAl9Si?0c;(0H)u + K^ + 2Si(0H)u

Silica although not an essential plant nutrient is contained in varying amounts in plants, depending on species, (Siever 1969). Silica accumulations in plants are redistributed when the plant decays and opaline particles released can be detected in the soil (Acquaye and Tinsley 1965), Such silica sources are factors contributing to higher silica values determined from litter waters in the Narrator Catchment (Ternan and Williams 1979). Silicon values measured in the Narrator Valley during this present Investigation are discussed in Section 7:3:6. These are converted to silica values which are widely used in water-resource literature when referring to the silicon content of natural waters. 7:1:4 Chloride

Although a major dissolved constituent of most natural waters, chloride is a minor constituent of the earths crust. A large fraction of the total chloride in common igneous rocks occur in biotite and hornblende which, according to Lahermo, (1970) may contain several tenths of a percent of chloride. Apatite may account for much of the minerogenic chloride found in waters from igneous areas, and liquid

339 - inclusions within an igneous mass may also contribute. Apatite is one of the accessory minerals of the Dartmoor granite as described In Chapter 2. However, the chloride content from this source is so slight that chloride released from the weathering of mineral matter is insignificant.

Quantitatively the most important source of chloride in near surface waters appears to be chloride transported via precipitation as outlined in Section 7:1:8. All chloride salts are highly soluble, so chloride is rarely removed from precipitation except under the influence of freezing or evaporation. Chloride is also relatively free from effects of ion exchange, adsorption and biological activity, so once taken into water at the earth's surface it is difficult to remove through natural processes.

Chloride concentrations found in natural waters vary between 0.5 mg 1"^ in snow, (Feth e£al. 1964), to between 5-24 mg 1"^ in granitic groundwaters In the US (White ££al. 1963) to 150,000 mg 1"^ in brines, (Davis and DeWiest 1966). Shallow groundwaters in regions of heavy precipitation generally contain less than 30 mg 1~^. 7:1:5 Sodium

Sodium Is the most abundant member of the alkali-metal group. Other naturally occurring members of this group are lithium, potassium, rubidium and caesium. In Igneous rocks sodium is slightly more abundant than potassium, although sodium is not found as an essential constituent of many rock forming minerals. Its primary source is from the release of soluble products during the weathering of plagloclase feldspars. Some quantities are also brought into an area by precipitation. Sodium in precipitation in the Narrator Catchment ranges from 0.37-11.00 mg 1~^ (Williams pers. comm. 1982).

The common feldspars are orthoclase (KAlSi30g) and plagloclase which has a composition ranging from the pure Na-form albite (NaAlSi303) to the pure Ca-form anorthite (CaAl2Si208). Both these minerals are present in the Dartmoor granite (Chapter 2) and their relative speed of weathering will determine proportions of potassium, sodium and calcium in groundwaters. Potassium feldspars are very resistant to chemical attack but species containing sodium and calcium are more susceptible and yield sodium, calcium and silica to solution.

340 - Clay minerals from the kaolinite, montmorillonite or illite groups result as a by product of this weathering.

Feth et al. (1964) in studies of water in granitic rocks of the

Sierra Nevada, California found In general that calcium and sodium ions were most important in stream and spring water and these reflected the abundance of the ions in the rock type and the rate at which minerals weathered. Sodium salts are soluble and will not precipitate unless concentrations of several thousand mg 1-^ are reached. Sodium tends to remain in solution rather persistently once it has been liberated from silicate mineral structures. The saturation point for sodium chloride is 264,000 mg 1-^ at 20*C or about 105,000 mg 1-^ sodium (Davis & DeWiest 1966). Because of the high sodium concentrations that can be reached before any precipitate is formed, the sodium content of natural waters has a wide range.

Sodium content varies from less than 1 mg in precipitation and dilute stream waters in areas of high rainfall, to greater than

100,000 mg 1"! in brines in contact with evaporites (Hem 1970).

Areas of igneous and metamorphic rocks that are also in regions of moderate to high rainfall generally have waters with 1-20 mg 1-^ sodium (Garrels and Christ, 1965). Sodium values measured in j^no c^-waters and surface waters in the Narrator Valley are in the range 4.5-7.0 mg 1-^ (Williams pers. comm. 1978). Lahermo (1970) in work In Igneous areas in Finland found mean concentrations of 3.3 mg l"^sodium in these groundwaters, while White et al. (1963) in areas with lower rainfall totals, found ranges of 5-197 mg l-^in American and South African igneous rocks. The sodium of dilute waters where total solids concentrations are below 1000 mg 1-^ is generally in the form of sodium ion, while in more concentrated solutions a variety of complex ions and ion-pairs is possible (Garrels and Christ, 1965).

The only common mechanism for removal of large amounts of sodium ions from natural water is through ion-exchange which operates if sodium ions are In great abundance. The effects of ion-exchange reactions is only to replace one kind of solute ion with another and therefore these reactions do not directly control the solubility of ions. The process of ion-exchange involves cations and is a revers• ible reaction. The reversal of the exchange direction may arise from small changes in the composition of the aqueous solution. A common reaction is the replacement of calcium and magnesium in groundwater by

341 sodium held in exchangeable media, such as clays, which have high cation exchange capacities. 7:1:6 Potassium

The common sources of potassium are the products of weathering of orthoclase (KAISI3O8), microcline (KAISIJOQ), blotite (KMg,Fe)3(AlSi3)Oio(OH,F)2), and leuclte (KAlSi206) in igneous and metamorphic rocks* The potassium feldspars are very resistant to attack by water but are eventually altered to silica and clay ions by the same processes as other feldspars only more slowly. Although the abundance of potassium in the earths crust is about the same as sodium, potassium is commonly less than one tenth the concentration of sodium in natural water. In many cases this variation is due to high atmospheric contributions of sodium. The rather narrow range of concentration of potassium in natural waters suggests that a significant chemical control mechanism may be involved. The resistance to solution exhibited by K-feldspar in relation to the Na- feldspars and the strong tendency for potassium to be reincorporated into solid weathering products, particularly clays, may well account for the lower potassium concentrations in groundwater. Vegetation concentrates potassium as it is an essential plant nutrient, but most of the potassium is returned as organic material decomposes in the litter layer and relatively little is removed by erosion.

As the potassium ion is larger than the sodium ion it would be expected to be adsorbed less strongly than sodium in ion-exchange reactions. In fact potassium is Incorporated In spaces between crystal layers. In illites for example, where they are not removable by further ion-exhange reactions, (Kurtz and Melsted 1972). The solubilities of potassium salts are all high and generally similar in magnitude to the solubilities of sodium salts. At 20®C

255,000 mg 1"^ KCL or 133,000 mg 1"^ potassium will be held in solution, (Davis & DeWiest 1966). Owing to this extreme solubility potassixim will not usually be removed from water except by sorption, ion exchange or precipitation during evaporation. All natural waters contain measureable amounts of potassium, some precipitation may contain as little as 0.1 mg 1~^. Most potable groundwaters contain less than 10 mg 1~^ and commonly ranges between 1-5 mg 1"^. Lahermo

342 - (1970) records a mean value of 1.53 mg 1~^ potassium in groundwaters from igneous rocks in Finnish Lapland. Potassium values measured in the Narrator Brook Catchment from springwaters, streamwater and interflow horizons are in the range of 0.8-2.0 mg 1~^ (Williams pers.comm 1978). 7:1:7 Atmospheric Contributions

Groundwater and surface waters derive part of their chemical composition from the atmosphere through rain, snow and dry fallout sources. Salt spray, fine clay particles from wind blown soils, and particulate matter in atmospheric pollution give rise to aerosols around which condensation may occur. Such aerosols form an integral part of the precipitation inputs. The principal Ions in rainfall are sodium, calcium, magnesium, potassium, chloride, sulphate and nitrate. Very little bicarbonate is present in rainfall (Rodda et al. 1976). Chloride and sulphate are derived from atmospheric sources, whilst most other Ions, (sodium, magnesium, calcium, potassium and nitrate) are derived from both atmospheric and mineral sources.

The total ionic concentration of rainfall is generally between 10-20 mg 1"^ although near coastlines values are higher, (Freeze and Cherry 1979). The mean ionic concentration for precipitation in the Narrator Brook is 12.77 mg 1"^ (Williams pers. comm. 1982) Table 7:2.

Ionic concentrations in rainfall are increased by evaporation, which can account for signflcant increases in the concentrations of all the main ions, except for calcium and bicarbonate, in actively circulating groundwater. The fact that rainfall is not a 'pure* solvent can be important in general studies of catchment solute dynamics, where dissolved con^tuents of stream water can be used as tracers for water of a particular origin, or from a particular subsurface flow route. Cryer (1976) found that concentrations of calcium, magnesium, sodium, potassium and chloride in bulk rainfall were found to be highly variable from week to week, and also displayed a distinct seasonal variation, with higher values In the winter months. Such variations have been utilised in this present study to elucidate patterns of groundwater movement in the weathered granite aquifer of the Narrator Brook Catchment.

343 Table 7:2 Summary Statistics for Bulk Precipitation Chemistry 1977-78, Narrator Brook Catchment Solute Mean concentration mg 1~^ Na 4.13

K 0.95 Ca 0.57

Mg 0.51 SiO. 0.02 CI 6.59 Total 12.77 7:1:8 Chloride in Rainwater In the UK the influence of the sea is marked by a decrease inland of chloride values, 15-30 mg 1"^ on the north-west coast to 3 mg 1~^ In rural areas in the centre of the UK (Rodda et al. 1976). Table 7:3 illustrates the chemical variation of precipitation with geographical location. Winter values are 2-3 times higher than summer values due to an increase in spray content of the air and the salt content at cloud level caused by winter storms. The chloride concentration of precipitation may depend on the amount of snow or rain falling. Gorham (1958) noted that light showery rains are much more concen• trated with chloride than is heavy rainfall.

Table 7:3 Annual Mean Values of Various Constituents (mg 1"^) in Rainfall over the UK for 1959-1964, from Stevenson (1968)

Solute Aberdeen Eskdale Muir Leeds Rothamsted Newton Stornoway Abbot

Calcium 1.1 0.3 2.7 1.4 1.5 1.1 Magnesium 0.6 0.2 1.0 0.3 0.6 1.9

Sodium 4.6 1.8 2.3 1.8 5.5 17.0 Potassium 0.4 0.2 0.5 0.3 0.7 0.9

Chloride 5.7 2.8 5.3 3.1 6.3 28.0 Sulphur 2.1 1.2 3.9 2.1 2.9 2.4

Nitrogen 0.3 0.1 0.6 0.4 0.7 0.2 in Nitrate

Nitrogen 0.3 0.2 1.1 0.5 1.4 0.2 in Ammonia

- 344 The concentration of salt In air near the ground (dry fallout) is known from the Western European chemical network, (Eriksson 1960). There appears to be a strong correlation between chloride in river runoff and chloride brought down by wet fallout. The ratio of chloride in wet fallout to chloride in river runoff have been computed for Sweden and Fin land. These are remarkedly constant varying around 0.30, (Eriksson 1960). It seems from this data that wet fallout chloride constitutes one third of the total concentrations delivered to the ground surface. Other components resulting in the concen• trations of chloride found in groundwaters and surface waters are from dry fallout (including evaporation effects), and weathering products. Cryer (1976) noted that dry depostion of atmospherically derived elements accounts for 2-4 times as much as inputs through wet fallout. As the Narrator Catchment is partially forested enhanced chloride levels may be input through dry fallout caused by impaction processes.

7:1:9 Potassium and Sodium in Rainwater

A great deal of the total available potassium and sodium may circulate through vegetation of a catchment area, (Tamm 1953). Gorham (1955) used data on the composition of rainwater before and after percolation through forest canopies, to illustrate potassium and sodium variations. An 18-fold enrichment of potassium concentrations compared to a 3-fold increase in sodium concentrations from precipi• tation collected at ground level, after percolation via the forest canopy was observed (Gorham 1955). This increase, Gorham (1955), attributed to foliar leaching as the precipitation passes over the vegetation surfaces, but such increases in concentrations will also result from evaporation effects. 7:1:10 pH of Rainwater

Granant (1972), Involved in studies of pH and major elements in rainfall from the European Atmospheric Chemical Network, determined that 90% of precipitation samples had pH in the range of 4 to 6. Sulphuric acid was suggested as the dominant factor in acidification of rainfall along with nitric acid created from nitrogen oxides. Mean values for precipitation in the Narrator Catchment during 1977-78 were pH 4.46 (Williams pers. comm. 1982), while the average value recorded

345 for rain falling through the forest canopy during this investigation (1978-80) was 3.86 (section 7:3:3). 7:1:11 Soil Zone Contributions

The soil zone has a unique capability to alter water chemistry as most recharge to groundwater Is through rainwaters percolating through this biologically active zone. Rainfall being an extremely dilute, slightly to moderately acidic oxidising solution, can quickly cause chemical alterations in soils or any geological materials through which it infiltrates. The most important change occurs as a result of active leaching and the transport of dissolved species, resulting from interactions of carbon dioxide and oxygen rich water with mineral constituents and organic matter in soil and weathered horizons.

The soil has a capability to generate relatively large amounts of acid and to consume much of the available dissolved oxygen in infil• trating water, Geochemically the most important acid is carbonic acid, (H2CO3) derived from the reaction of atmospheric and soil carbon dioxide (CO2) with water. The CO2 is generated by the decay of organic matter and by respiration of plant roots. Anaerobic reactions like the reduction of sulphate and nitrate also produce €02*

CO2 partial pressure of the soil atmosphere are typically in the range of 10"^ to 10'^ bars (Freeze & Cherry 1979). Partial pressures of CO2 are spatially and temporally variable due to changes in temper• ature, moisture conditions, microbial conditions, availability of organic material and the effects of soil structure on gas diffusion. When CO2 at partial pressure (10~^ to 10"^ bars) reacts with water, the pH declines markedly. For example with CO2 at partial pressure of 10"^ bar water in the temperature range 0-25**C will have an equili• brium pH in the range 4.3-4.5. In the Narrator Brook Catchment the buffer effect of CO2 in the soil and weathered zones may well account for the low pH (4.2-5.2) experienced in the groundwaters. Mineral water reactions consume H2CO3 in the soil zone, while oxidation of organic matter together with root respiration replenish CO2 to the soil air. As new water from a recharge event enters the soil the CO2 combines with water and more carbonic acid is formed. Biochemical and hydrochemical processes in the soil body are capable of providing a continuing supply of acidity to promote mineral water reactions. In addition to inorganic acids there are organic acids produced in the

- 346 soil zone by biochemical processes, Humlc and fulvlc acids play a major role in the development of soil and weathered profiles and in the transport of dissolved constituents downwards towards the watertable. 7:1:12 Geological Contributions

Quartz» and the aluminosilicate minerals such as feldspars and micas present in igneous rocks, were originally formed at temperatures and pressures far above those occurring at or near the earths surface. As a consequence these minerals in the soil or subsurface regions and In the groundwater zone, are Chermodynamically unstable and tend to dissolve when in contact with water. The dissolution processes cause the groundwater to acquire dissolved constituents and the rock to alter mineralogically.

The dissolution of feldspars, micas and other silicate minerals is strongly influenced by the chemically-aggressive nature of the Infiltrating waters caused by dissolved CO2. When the H2CO3 waters low in dissolved matter encounter silicate minerals high in cations, then ions such as calcium and sodium along with silica are leached out. An aluminosilicate residue is produced with an increased Al:Si ratio, which includes oxides of ferric iron.

Many minerals that effect the chemical evolution of groundwater dissolve incongruently, that is, one or more of the dissolution products occur as minerals or amorphous solid substances. By contrast, congruent dissolution of minerals produce dissolved species. Most aluminium silicate minerals dissolve incongruently.

The residue produced by the dissolution of silicate minerals in granites is a clay mineral such as kaolinite, or illite. The cations released to the groundwater are normally sodium, potassium, magnesium and calcium. Magnesium can be built into the clay mineral crystal structure and thus its removal by solution processes is retarded. Potassium, which competes strongly in ion exchange processes on clay mineral surfaces can also become fixed by inclusion between layers in layer-structured clay minerals. In the case of Igneous rocks, Eriksson (1960) suggested that it can be assumed quantitatively that all sodium and calcium goes into solution and that practically all calcium and potassium in river water can be accounted for by weathering processes. However, Gorham (1961)

- 3A7 suggested that of the concentrations of chloride found in river water, only a very small amount comes from the weathering of Igneous rocks in a catchment, since such rocks are deficient in carbonates, sulphate and chlorides. These constituents, found in waters from Igneous catchments are mainly derived from precipitation, (Rodda et aj^. 1976). In the case of Dartmoor both chloride and significant amounts of potassium are related to maritime sources of precipitation. 7:1:13 Contributions to River Water

The quality of water at a point in a river reflects the conditions prevailing In the catchment area above that point. Geological condltionia exert an influence on water quality as does the import of ions via precipitation in a region. The chemical nature of rivers is subject to seasonal and shorter term variations, related to flow rates. These in turn are related to varying contributions of different sources making up the total flow at any one time. There appears to be a broad Inverse relation between the volume of river flow and ionic concentration of water. During dry periods the ionic concentration of river water Is higher because more concentrated groundwaters may make a significant contribution. During higher rainfall periods river flow has a large proportion of direct runoff which generally has lower ionic concentration. However, Cleaves et_ al. (1970) found that potassium, magnesium and sulphate concentrations Increased during flood flows. The source of these constitutents was from the leaf litter and upper soil horizons, which upon being 'flushed* in heavy rainfall events released significant amounts of potassium and magnesium to streamwaters• 7:1:14 Summary of Sources of Ions

The sources of the four elements measured in the Narrator Valley (chloride, potassium, sodium and silica) may be suimnarised briefly: 1. Atmospheric Contributions (precipitation, dry fallout) This source provides sodium and potassium, but predominantly chloride In the Narrator Catchment, The concentrations of these ions vary greatly, particularly in the case of chloride which increases In the winter months due to the frequency of Incoming depressions from maritime areas.

348 2. Biologic Contributions (vegetation and litter decay) This source provides potassium and silica contributions which vary seasonally. Some chloride is provided by atmo• spheric dry fallout onto vegetation surfaces.

3. Geological Contributions (source rock geochemistry, weathering processes in surface zones) This source provides mainly sodium, potassium and silica to groundwaters. The amount of chloride produced is minimal* Each of the above three sources are characterised by variations which will impart different geochemical responses in waters in the Narrator Brook Catchment. These changes in time are utilised to determine the hydrogeological response of a weathered granite aquifer.

7:1:15 Seasonal variations in groundwater chemistry Tardy (1971), during investigations with geochemical data of waters from European and African crystalline massifs, presented evidence to support the hypothesis that chemical compositions of water are sensitive to both seasonal and climatic variations. In an attempt to relate climatic variations in groundwater chemistry to those occurring as a result of the geochemical dynamics of weathering, it was found that temperatures and annual rainfall variations intervene in the course of seasonal rhythms (Tardy 1971). Such interventions modify the geochemical dynamics of weathering and as a consequence the geochemistry of the groundwaters alter. It is inferred from this global study that silicate mineral decomposition Is likely to increase during warmer and wetter periods of the year, and cation concentration in groundwaters increases as a result. In crystalline areas the chemical composition of water are sensitive to seasonal variations and they reflect the evolution of weathering at the mo ment of sampling (Tardy 1971).

7:2 Presentation of Hydrochemical Data

One of the major problems of hydrochemical investigations is the magnitude of the quantities of data generated, and the subsequent problem of presentation. Several methods of data analysis have been devised to simplify interpretation and presentation. Various simple procedures such as averaging, determining frequency distributions and making simple or multiple correlations are widely used in water . analysis interpretation (Hem 1970).

349 Histogram displays have long been used in the literature, but trilinear diagrams (Piper 1944) and Stiff diagrams (Stiff 1951) have become more commonly used because of their greater versatility. Trilinear diagrams have two main drawbacks in the light of this present investigation: they ignore many parameters suitable for hydrochemical studies, eg SIO2; and they may be ambiguous as they use the relative percentage of different ionic concentrations rather than absolute concentrations. Stiff diagrams can be distinctive for showing groundwater chemical differences or similarities, in milliequivalents per litre of cations and anions. These diagrams rely on analysis of all the major ions being undertaken in groundwaters, but again omit silica concentrations, and so were not considered for present purposes. Ashley and Lloyd (1978) pointed out that in view of the limitations of existing methods and the increasing number of chemical parameters now being measured in groundwaters there is need for more wide ranging statistical analysis of data. These authors suggested multivariate tests, including factor and cluster analysis especially for chemical interpretation on a regional scale.

Lahermo (1970, 1971) in the Interpretation of the geochemistry of groundwaters in Finland, made extensive use of graphs of periodic solute variations, cumulative curves, bargraphs of regional chemical concentrations, and average ratios of dissolved constituents. Exhaustive use^as made of correlations between each of the dissolved elements, pH and electrical conductivity to explain the underlying causes of variability in water chemistry. Although Lahermo's (1970) investigation is on a regional scale, no resort is made to multivariate analysis in his Interpretation.

Keller (1970) used average values and standard deviations to investigate variations In streamwater chemistry in five catchment in the Prealps of Switzerland. Variances and standard deviation of chemical data are also used successfully by this author, to isolate chemical components showing considerable variation in time and location, which were then incorporated into a stepwise regression. Foster (1979) used mean, standard deviation and Chi-square statistics to isolate the degree of variability in solute characteristics

350 - exhibited by a small catchment in east Devon, and to explain such differences in terms of biogeochemical control. This analysis (Foster 1979) indicated the potential geochemlcal, biological and atmospheric controls on solute levels under differing land use and drainage conditions using these simple statistical techniques. 7:3 Chemistry of Groundwater and Surface Waters in the Narrator Valley

The main sources and general concentrations of chloride, silica, sodium and potassium in igneous areas have been presented in order to place the groundwater and associated surface water chemistry of the Narrator Valley into its expected natural geochemical setting. 7:3:1 Seasonal Trends in Water Chemistry: Temporal Patterns

Maximum and minimum values for temperature, pH, electrical conductivity, chloride, silica, sodium and potassium in groundwaters, streamwater, precipitation and stemflow waters in the Narrator Valley are summarised in Table 7:4. Mean values for each site over the research period are listed in Table 7:5. As there was little weekly change in the measured parameters longer term trends were sought and the average monthly values are presented in Figures 7:1 to 7:7, The standard deviation of the parameters listed above, determined from the weekly observations, were computed and their distributions discussed in section 7:A.

Table 7:4 Maximum and minimum values of temperature, pH, electrical conductivity, chloride, silica, sodium, and potassium in groundwaters, surface waters, precipitation and stemflow in the Narrator Brook

Catchment

Source E.C. St02 Na K •^c \iS/an •

Precipitation 3.4-5.2 30-205 2.6-43 <0.1 2.2-33.0 0.5-5.9

Stemflow - 3.5-5.5 142-341 11.0-81.0 <0.1 12.0-46.0 1.8-7.5

Stream flow 5.6-12.3 4.6-5.5 52-64 9.9-U.O 4.2-6.8 4.5-6.0 0.56-0.87

Spring flow 8.9-9.9 4.0-4.9 58-100 9.0-18.0 4.2-10.0 4.a-«.8 0.5-1.4

W?n vaters 4.2-13.5 4.2-5.9 52-172 9.0-41.0 2.5-10.7 3.0-22.0 0.5-2.8

* ionic concentrations are in 1- 1 N.B. Electrical condictlvity Mnhos/cm = nS/cm

351 Table 7:5 Mean values of temperate, pH, electrical conductivity, chloride, silica, sodium and potassium in the Narrator Valley

Location Temperature pH Electrical Cl• SIO2 Na+ K+ •*c Conductivity jiS/cm mg :

Well 1 9.7 4.9 73.64 io.4 5.3 5.2 0.9 Well 2 9.3 5.2 110.6 15.4 5.7 7.2 2.6 Well 3 9.2 4.7 81.6 13.6 8.5 6.9 1.9 Well 4 9.6 5.4 122.9 14.3 8.4 12.0 1.4 Well 6 9.5 5.4 99.4 13.1 7.2 6.1 1.8 Well 7 8.5 4.9 125.3 23.5 3.2 7.6 0.9 Well 9 9.2 4.8 77.5 13.8 6.8 6.8 1.8 Well 10 9.3 4.8 70.9 12.6 9.5 6.4 1.5 Well 11 8.9 4.4 86.9 15.6 7.0 7.7 1.2 Sp.ll 9.5 4.6 93.2 15.5 8.7 7.6 1.2 Well 12 9.2 4.5 63.5 11.8 6.8 6.0 1.0 Sp.12/13 9.4 4.3 63.1 11.7 5.4 5.6 0.8 Well 13 9.3 4.8 83.4 13.9 7.1 6.5 1.5 Well 14 8.8 4.8 107.5 14.5 7.4 7.1 1.8 Well 15 9.0 5.2 98.12 15.8 5.4 6.9 2.0 N. Brook 9.2 5.1 57.4 11.2 5.5 5.4 0.7 RF7 - 4.0 76.91 13.7 Trace 6.5 1.2 RF9 - 3.8 130.2 18.2 Trace 14.4 1.9 RFU - 3.8 110.3 17.6 Trace 11.5 1.4 Stem Flow (SF) — 4.4 213.1 50.0 <0.01 30.2 3.9

352 - 5R 12/13

SP. 11

WELL 5

WELL U

WELL 13

WELL 12

WELL 11 uj 10

WELL 10

WELL 9

WELL 7

WELL 2

WELL 1

1979 1980

FIGURE 71a. MEAN MONTHLY VALUES: TEMPERATURE NARRATOR BROOK 1979-1980

35 3 0 0 r- M E A N A I R TEMP

N .BROOK

WELL 6

WELL 3 WELL L

19 79 98 0

FIGURE 7.1b. MEAN MONTHLY VALUES: TEMPERATURE NARRATOR BROOK 1979 - 1980 4.0,- R F. 7 = 3-5 RF. 11 3.0_ 4 0 RF. 9

30

5 0 N. BROOK

L • 5 SR 12/13

SP.11

WELL 15

WELL U

WELL 13 5 4 WELL 12 Q. 5

WELL 11 4 5 ^WELL 10

—• WELL 9

WELL 7

WE LL 4

50 WELL 6

WELL 3

50 WELL 2

5.0 WELL 1

4 0 F M M N J F 1979 19 80

FIGURE 7 • 2 . MEAN MONTHLY VALUES: pH NARRATOR BROOK 1979 -1980 120

110

100

90

WELL 15 80 9d

80 75 WELL 13 65=

55 WELL 12 E 9 0' WELL 1 m 80 o li £ 6 5 WELL 10 75' 70 WELL 9 > 60 SP.12/13 ;i i3o' WELL 7 o Q 120 2: 1 15 o * o 1.10 _i o 100

CE I- 90 o WELL 6 80 UJ

70

80

WELL 3 70. 115' 110

100> WELL 2

70 WELL 1

60 J 1 F M M J F 1979 19 80

FIGURE 7.3 a. MEAN MONTHLY VALUES: ELECTRICAL CONDUCTIVITY NARRATOR BROOK 1979-1980 210 200 1 9 0 1 80 1 70 1 60 1 5 0 1 L 0 R.F. 9 130 1 2 0 1 1 0 1 0 0 90 80 70 E 60 R. F. 11 u 5 0 « L 0 o 3 0 E 6 0 a*

5 0. N. BR 00 K

U 0 o 1 30 a 2 1 2 0 o o 1 1 0

1 0 0

cr 90 SP. 11

80 WELL U lU 70

1 AO

I 3 0 WELL L 120

II 0

1 00

90

8 0 J 1 F M M J F 1979 198 0 FIGURE 7.3 b. MEAN MONTHLY VALUES: ELECTRICAL CONDUCTIVITY NARRATOR BROOK 1979-1980 R.F.7

R.F.11 R.F. 9

• N. BROOK

SP. 12/13 SP. 11

WELL 15

WELL n

WELL 3

WELL 2 • WELL 11

WELL 10

WELL 9

WELL 7

WELL WELL 6

WELL 3

WELL 2

WELL 1

F 1979 19 ao

FIGURE 1 • U . MEAN MONTHLY CHLORIDE (CL') VAL U ES NARRATOR BROOK 1979- 1980 6-4

4.2 N. BROOK

6-4

S P. 12/13

SP. 11

WELL 15

WELL 14

WELL 13

WELL 12 7.4 E 6- 4 7- 4^ WELL 11 O 6- 4 WELL 10

8.5 7- 4 42 WELL 9 3 2 WELL 7 8.5

6-4 WELL 4 WELL 6

8- 5

WELL 3

WELL 2

4.2 WELL 1

2- 1 J I L F M M J F 1 979 1 9 80

FIGURE 7-5. MEAN MONTHLY SILICA ( S i O2 ) VALUES NARRATOR BROOK 19 79-1980

359 10. Or-

WELL U WELL 15

WELL 13 WELL 12

WELL 11

WELL 10

WE LL 7

WE LL 9

WELL 3

WELL 2

WELL 1

1979 1 9 80

FIGURE 7-6 a. MEAN MONTHLY VALUES: SODIUM (Na) NARRATOR BROOK 1979-1980 - 360 - 26-0

2U- 0

22 0

200

1 80

1 6- 0

R.R 9 1 ^0

1 20

100

8-0

- 60

E R.R 11 ^-0

Z 6 0

50

N. BR 00 K AO

60

5-0 S R 12/13 80

7-0 SP.11

6 0

5.0 WE L L 6

A-O

F M M N J F 19 7 9 1980

FIGURE 7.6b. MEAN MONTHLY VALUES: SODIUM (Na) NARRATOR BROOK 1979-1980 30 r

R.F. 9 R.R 7 R.F. 11 N.BROOK

SP. 12/13

SP. 11

WELL 15

WELL U

WELL 3

WELL 12

WELL 11 o> 1 E WELL 10

^ 1

2 WELL 9

1 WELL 7

2 WELL 6

1 W E LL i

1

2 WELL 3

1

WELL 2

1 OH WELL 1

19 7 9 19 80

FIGURE 77. MEAN MONTHLY VALUES POTASSIUM (K) NARRATOR BROOK 1979 1 980 7:3:2 Groundwater Temperatures Groundwater temperature fluctuations are an effective means whereby the possible source and likely depth of the water can be determined (Freeze and Cherry 1979). Variability In groundwater temperatures may be attributed to: (1) air and soli temperatures which prevail when precipitation falls. These may Impart a seasonal rhythm to groundwater temperature fluctuations at shallow depths. (11) the residence time In the ground and the degree of mixing of different groundwater sources In well, spring or stream locations. Ambient air temperatures may affect deeper groundwaters as they move towards the surface. The range of water temperatures In the Narrator Brook Catchment experienced during the research period are illustrated In Figures 7:1a, 7:1b. The shallow wells 3, 4 and 6 show a seasonal temperature range similar to that experienced In the Narrator Brook Itself, which reflects the pattern of the air temperatures (Figure 7:1b). Wells 9, 10, 11, 12 and 13 show a similar but more subdued annual pattern. Wells 1, 2; 7, 14 and 15 exhibit very small annual temperature ranges, and the two springs maintain a more or less constant temperature throughout the year. Air temperatures at the time of sampling may slightly modify water temperatures but these are not considered to have a significant effect on the measured groundwater temperatures. In the case of the wells, temperatures were all determined at depth, or in the case of the springs at the point of emergence. Inferences about the depths from which the groundwaters originate can be drawn from the standard deviation of water temperature fluctuations (Pitty e£al. 1979). A wide temperature range and large standard deviations suggests near-surface flow, where waters with a narrow annual temperature range, and small standard deviations are characteristic of a deeper groundwater source. Figure 7:8 Illustrates the distribution of standard deviations of temperatures over the catchment. Spring 11 shows the smallest standard deviation. Both springs disrupt the standard deviation pattern shown by the wells, which suggests that the spring waters come from much deeper levels within the catchment.

363 O 0-69

®0-76

KEY vyetls and Standard Deviation of Temperatures ®1.31

® 0-99

125m

FIGURE 7:8 DISTRIBUTION OF STANDARD DEVIATION OF TEMPERATURE FOR WELLS AND TWO SPRINGS IN THE LOWER NARRATOR VALLEY 1978-1980

364 With reference to Table 7:6 it can be seen that groundwaters with standard deviations greater than unity are interpreted as being from shallow sources which are markedly influenced by seasonal temperature fluctuations. Wells with groundwaters from deeper sources, 1, 2, 7, 14, 15, show lower standard deviations as do the waters from Che springs with standard deviations of (0.54) Sp-U, and (0.71) Sp. 12/13. Ternan and Williams (1979), working on springs higher upstream in the catchment found standard deviations as low as 0.12, suggesting that some sources outside the immediate area under consideration here, have much deeper origins within the weathered granite valley.

Table 7:6 A standard deviation of water temperatures and soil temperatures in the Narrator Valley with well depth and site elevation

Well Depth Topographic Mean Standard Height (m) Temperature Deviation of Temperature

1 19.25 239.51 9.7 0.69 2 18.75 230.31 9.3 0.76 3 2.29 223.09 9.2 2.98 4 1.30 223.13 9.6 3.12 6 5.63 220.24 9.5 2.81 7 13.50 228.33 8.5 0.82 9 14.72 237.62 9.2 1.05 10 9.67 234.62 9.3 1.31 11 4.80 230.86 8.9 2.14 12 9.39 242.44 9.2 1.43 13 10.33 248.11 9.3 1.22 14 12.76 248.09 8.8 0.93 15 14.45 240.89 9.0 0.99 N.B. 220.0 9.2 2.36 Sp.U - 230.0 9.5 0.54 Sp.12/13 245.0 9.4 0.71 Soil temperatures - Yellow Mead 0.30 304.0 8.69 3.15 0.60 304.0 8.64 2.69

The relationship between well depth (and hence depth of groundwater source) and the standard deviations of temperature fluctuations in the Narrator Valley are illustrated in Figure 7:9. In the weathered granite horizons of the Narrator Valley, the deeper the

365 FIGURE 7:9

WELL DEPTH AGAINST STANDARD DEVIATION OF WATER TEMPERATURE NARRATOR BROOK CATCHMENT 1979-1980

2 < LU 5

4 -I i

^ 3 A

2 O

< 2 H

2 1 ^ 2

14 16 18 20

WELL DEPTH METRES groundwater source, Che smaller the annual temperature range. Soil temperatures at depths of 30cm and 60cm from Yellowmead In the Narrator Catchment, have standard deviations of 3.15 and 2.69 respectively (Williams pers. comm. 1982). These have been plotted in Figure 7:9 and follow the general trend. A regression line was fitted and a correlation coefficient of 0.92 calculated (significant at 0.005 Level). Using the derived .regression equation and the standard deviation of spring temperatures an estimate of the probable depth of flow was derived: Standard Deviation Depth Spring 11 0.54 17.70 Spring 12/13 0.71 16.50 These locations are indicated in envelopes on the regression plot of Figure 7:9. Five main groupings are evident from Figure 7:9. Group A consists of the two springs and the two deepest wells (Well 1 and 2) in the catchment all possessing the lowest standard deviation of temperature. Group B consists of Wells 7, 9, 14 and 15 ranging from depths of 13.50 - 14.45m. Group C Includes Wells 10, 12 and 13 at depths between 9.3-10.3m, and group D includes Wells 6 and 11. Group E consists of the shallowest wells 3 and 4 along with soil temperatures from the higher reaches of the catchment. This group shows the highest standard deviation. Waters derived from a homogeneous aquifer-unit from comparible depths would be expected to have a similar temperature range, a feature which is not characteristic of the Narrator Valley wells. There is a wide variation in the depths of the wells, and hence groundwater sampled and even in situations where wells of similar depths are measured there are marked differences in standard deviation of temperatures (Table 7:6). This suggests that wells may penetrate localised discrete groundwater bodies at different levels within the valley aquifer, a hypothesis which is supported by analysis of water level fluctuations (Chapter 5) and by chemical considerations, section 7:4. Based on the standard deviation of the groundwater temperatures and well depths in the Narrator Brook Catchment (Table 7:6) the sources of flow can be assigned as follows: the shallowest source at approximately less than 6m below ground level penetrated by Wells 3,

367 4y 6 and 11; the deepest source at depths between 16-20m, penetrated by Wells 1, 2, and possibly by the two springs. Intermediate source depths can be found from 9-15m as intersected by Wells 10, 12, 13, 7, 9, 14, 15. These intermediate sources may well be the result of a combination of perched water bodies above clay layers on the valley sides. 7:3:3 pH pH of groundwaters varied between 4.2 and 5.9 (Table 7:4), Rainfall and stemflow waters were even more acidic with values between pH 3.4-5.5 (Figure 7:2). pH values measured in groundwaters showed little variation throughout the research period. The distribution of the standard deviation of pH in the catchment is illustrated in Figure 7:10. The lowest standard deviation is in the location of Well 11 and Spring 11, and the higher zones are around Well 6 and 4. The greater variations of pH at Wells 6 and 4 can be attributed to the influx of surface waters at various periods during the year. 7:3:4 Electrical conductivity Conductivity data for the waters of the Narrator Brook Catchment are presented in Tables 7:4, 7:5 and Figure 7:11, High conductivity values occurred at several sites, particularly Wells 3 and 4, and may be attributable to evaporation of standing water in the vicinity of these wells, together with solutes released from decomposing litter. Large variability in electrical conductivity at Well 6 may also be related to the ingress of drainage waters, as the site becomes flooded during wet weather. In stemflow measurements in the catchment electrical conductivity varied from 142-341 nS/cm constituting the highest values recorded in the catchment (Table 7:4). Well 1 possesses the lowest standard deviation of electrical conductivity, indicating a more constant source of recharge waters. Wells 13, 14 and 15 have the highest standard deviation indicating the variability and possible retention of recharge waters at these locations. Wells 9, 10, 11 and 12, together with Sp.U, and Sp. 12/13 all have standard deviations of electrical conductivity measurements of a similar magnitude. Well 7

368 - Din-

FIGURE 7:10

DISTRIBUTION OF STANDARD DEVIATION OF PH NARRATOR VALLEY 1979-1980

FIGURE 7:11

35 _

DISTRIBUTION OF STANDARD DEVIATION OF ELECTRICAL CONDUCTIVITY NARRATOR BROOK 1979-1980 is noticeable having a standard deviation of conductivity greater than those of Wells 1, 3, 9, 10, 11, 12 and less than those of Wells 2, 4, 6, 13, 14 and 15. 7:3:5 Chloride Chloride concentration In rainfall is higher in the winter months reflecting the greater frequency of maritime storms over Dartmoor (Figure 7:4). Chloride in groundwater shows a slight seasonal trend in accordance with rainfall values although peaks in chloride concentrations in groundwater are retarded by one to two months (Figure 7:4). Chloride in Well 7 however are almost coincident with the peaks and troughs of the rainfall chloride values. Well 7 has the highest chloride value in the groundwater in the catchment. Stemflow waters had the highest chloride value recorded, emphasising the Importance of dry fallout and vegetational influences on chloride inputs. The lowest values were recorded at Well 1, Sp.12/13 and the Narrator Brook. The mapped distribution of the standard deviation of chloride values in groundwater in the Narrator Valley Is illustrated in Figure 7:12. Wells 13, 14 and 15 show intermediate standard deviation values for chloride while the majority of the other sources have standard deviations of less than 3.2. Well 7 is a notable exception with the highest standard deviation for chloride of 9.73. Wells 3 and 4 also show high values Indicative of the wide variation in input concentrations of both groundwater and surface water at these locations. Wells 1 and 11 show the lowest standard deviation of chloride in the observation wells in the catchment. 7:3:6 Silica Figure 7:5 illustrates that there is little seasonal trend of silica in groundwaters and surface waters of the Narrator Valley. The annual range is from 2.5 - 10.7 mg 1~^ silica in groundwaters from the catchment. Wells 3, 4, 10 and Spring 11 have the highest mean values (Table 7:5) of 8.5, 8.4, 9.5 and 8.7 mg 1-^ silica respectively. Well 7 has persistently low silica concentrations with a mean of 3.2 mg 1"^. Lower silica concentrations could be associated with the greatest frequency of flow in the aquifer, so that contact time with weathered granite is minimised and hence the solute concentration of groundwater is reduced. Water from greater depths. Spring 11, has a

370 - FIGURE 7:12

©>6 /

Aaf

DISTRIBUTION OF STANDARD DEVIATION OF CHLORIDE NARRATOR VALLEY 1979-1980

FIGURE 7 : 13

o 6o

DISTRIBUTION OF STANDARD DEVIATION OF SILICA NARRATOR BROOK 1979-1980 longer residence time, and therefore a longer contact time with silicate minerals. Williams et^ al. (1981) found that waters from below the fragipan in weathered regollth in the Narrator Valley, was found to have a higher silica content (mean 5.88 mg/1) than shallower interflow waters. This was interpreted as water having a deep source, or a shallower source passing through accumulations of weathered materials which retard the flow velocities and increase residence times and hence solute concentrations. The low standard deviations of silica in groundwaters reflect the constancy of this element in the Narrator Valley and the distribution of the standard deviations are illustrated in Figure 7:13. Well 11 appears to be the centre of an area of low standard deviation in silica reflecting the constancy of the source of groundwater at this location. 7:3:7 Sodium Mean monthly sodium concentrations in the Narrator Brook are illustrated in Figures 7:6a and 7:6b. Sodium is higher in rainfall in the winter months as Indeed are electrical conductivity and chloride concentrations, reflecting the maritime influences on south westerly precipitation events over Dartmoor during the winter. Sodium in groundwaters is higher during June to November when feldspar weathering is likely to be more rapid during the warmer months, and hence more sodium is released. An exception to this general pattern is found in Well 7 whose sodium concentration increases during November to May, similar to rainfall trends. As can be seen from Tables 7:4 and 7:5 the higher sodium totals in precipitation are not maintained in groundwaters suggesting that adsorption and exchange processes are active in the unsaturated zone in the catchment. Wells 4 and 7 have the highest sodium concentrations with the lowest values being recorded at Well 1 and Spring 12/13. The distribution of the standard deviation of sodium in the groundwaters in the catchment is mapped in Figure 7:14. Wells in the clay areas, (Wells 9, 13, 14, 15 and Spring 11) have high standard deviations as do the riverside sites Wells 3, 4 and Well 7. Spring

372 12/13 and the Narrator Brook Itself have the lowest standard deviation of sodium, suggesting a more constant source at these two locations. 7:3:8 Potassium Potassium values in wells and springs in the catchment show a lower seasonal variation than those in rainfall (Figure 7:7). Higher potassium values are apparent in rainfall during the. summer months, a pattern which is reflected in groundwater values. Given that potassium is an essential plant nutrient maximum demands will be made on potassium during the growing season of plants. This uptake may be modified by potassium released during the summer months from feldspar weathering. The higher summer values of potassium may also be attributed to foliar leaching effects as precipitation infiltrates the forest canopy. A peak in potassium values in groundwater would be expected in the autumn/early winter due to contributions from decaying vegetation. These peaks are not evident in groundwaters in the Narrator Brook Catchment (Figure 7:7). The distribution of standard deviation of potassium In groundwaters in the Narrator Catchment are Illustrated In Figure 7:15. The higher standard deviation for potassium are shown by the deeper wells on the south side of the stream. Wells 9, 13, 14, 15 and by the springs. The springs, by virtue of their physical location (hollows in the ground) collect large amounts of decaying vegetation which may account for the higher standard deviations for potassium experienced. The deep Well 1 on the north side of the valley has a low standard deviation value for its elevation, which is at variance with conditions on the south side of the valley. Shallower wells in the catchment show intermediate to low standard deviations for potassium as illustrated in Figure 7:15. The region in the neighbourhood of Well 12 and 11 (Figure 7:15) show the lowest standard deviation for potassium levels of groundwaters in the catchment. 7:3:9 Summary of Temporal Patterns in Groundwaters from the Narrator Brook Catchment Figures 7:16a,b summarise mean values of temperature, pH, electrical conductivity, chloride, silica, potassium and sodium in the wells and springs monitored in the Narrator Valley. The similarities between the groundwaters at each site are evident, and Well 7 displays

- 373 - FIGURE 7: 14

II

(J-78 <5) , o**^

DISTRIBUTION OF STANDARD DEVIATION OF SODIUM NARRATOR VALLEY 1979-1980

FIGURE 7:15

11

DISTRIBUTION OF STANDARD DEVIATION OF POTASSIUM NARRATOR VALLEY 1979-1980 a marked difference from the other groundwaters sampled. These anomalous features are further discussed in Section 7:5:3. From Tables 7:6-7:12 and more particularly from Figures 7:10-7:15 the area for the lowest standard deviations (apart from electrical conductivity) is found predominantly in the vicinity of Spring 11 and Well 11. The standard deviations of all the measured parameters increase in the downstream direction towards the reservoir. Areas with anomalously high standard deviations of electrical conductivity, chloride and sodium are found in the vicinity of Wells 4, 6, 7 and 14. The lowest standard deviations for temperature, pH, electrical conductivity, chloride, silica, sodium and potassium in the Narrator Valley are summarised in Table 7:7. These values suggest that Spring 12/13, Well 11 and Well 1 have the most constant groundwater concentrations, and hence sources, followed by Spring 11, Wells 2, 7 and 12 respectively.

Table 7:7 Standard deviations of the most constant groundwater constituents in the Narrator Brook Catchment

Site 1 2 7 10 11 Sp.ll 12 Sp.12/13

Standard Deviation Temperature 0.69 0.76 0.82 - - 0.54 - 0.71 pH 0.21 - 0.21 - 0.17 0.18 0.21 Electrical 2.9 - 3.0 3.-0 3.3 conductivity - - - Chloride 1.5 2.46 - - 1.91 2.52 - 2.36 Silica - 0.68 0.68 - 0.45 - 0.67 0.64 Sodium - - - - 0.76 - 0.78 0.64 Potassium 0.3 — — — 0.26 — 0.18 —

7:3:10 Hydrogeologlcal Groupings Based on Water Chemistry The parameters measured at each location in the Narrator Catchment suggest that overall there is a similarity between all the groundwaters sampled. From the monthly chemical variation-plots (Figures 7:1 to 7:7) in association with the mapped distribution of the standard deviations (Figures 7:10 to 7:15) the wells may be assigned to six hydrogeologlcal groupings which show similar patterns:

375 - 130

120. WELL 1 WELL 2 WELL 3 110- •25 5. 100H 10- 20

90- •15 L £^

80- X -5 HO 3 3 1 K f 0 70- X -4 5 3 I 0. O (A \- 1 bO-*

130 WELL 4 WELL 6 WELL 7

120.

110 J r25

100 10 •20

90 15

80-^ •10

70 7- I 60

130'

120-

110- r25 WELL 9 WELL 10 WELL 11 100- 10 - 20

90- f M5 X 80- 8 3 -5 10 -L. 1 Q. 3 3 *5 1 eV e 3 70- s It! 7- •f 0^ Q 0 X (A 2 0- c 0) \^ 60- 3 FIGURE 7:16a MEAN VALUES OF TEMPERATURE, PH, ELECTRICAL CONDUCTIVITY, CHLORIDE. SILICA, SODIUM AND POTASSIUM. NARRATOR VALLEY 1979-1980 SP12/13 SPlI WELL 12 25

> 100- .0 10- 20 I-

90i lu 9- 15 » oc =) in •5 10 tf) 80H 2 8- '5 •i a. 1 •i UJ S •e0 •9

3 CO ' ^ 7- J n 1 70 H Q ^ 8 1 o o 3 a Vi 1 "2 i-b

130-1

120 WELL 13 WELL 14 WELL 15

r25 110 H

100- olO- 20

15 90 H o.

•5 BO• -5 KlO s c TCH Si 7 u 0 [El 60

FIGURE 7:16b MEAN VALUES OF TEMPERATURE, PH, ELECTRICAL CONDUCTIVITY, CHLORIDE. SILICA, SODIUM AND POTASSIUM. NARRATOR VALLEY 1979-1980 Groups (i) (il) (iii) (iv) (v) (vi)

Sites 3 10 1 9 Sp.ll 7 4 12 2 lA Sp.12/13 6 13 15 11

Well Groups (i), (ii), (111) and (Iv) comprise the same groups as those groups chosen by the change In water level maps; analysis of variance; and hydrograph amplitude as described in Chapter 5. The springs are not Included in former hydrogeological groupings, and the Isolation of Well 7 In a group on its own highlights the differences in its geochemlcal responses. The anomalous position of Well 7 is further discussed in Section 7:5:3. 7:3:11 Comparison with groundwaters of other granitic areas Table 7:8 summarises the chemical components in the spring and observation well waters in Che Narrator Brook Catchment, and by means of a comparison, tabulates characteristic solute contents of groundwater and sprlngwaters, where available, from other granite terrains in the UK. As can be seen from this table the pH of the groundwaters in the Narrator Valley are the most acidic reflecting the influence of peat runoff on overall groundwater chemistry, similar features are observeable in peat based springs in the Caithness area (Kay pers. comm. 1980). Chloride values from all the wells and springs in granite areas of the U.K. are similar reflecting precipitation inputs and variations in local geological conditions. One exception to this are the high levels experienced in the Isles of Scilly which may be attributable to the influence of the sea on chloride content of rainfall. Silica values are within the range quoted by Davis (1964) of between 10-30 mg/1, although some waters from the Scottish granites and the Narrator Valley are lower.

- 378 - Table 7:8 Comparison of Groundwater Chemistry from Granite Areas In the U.K.

SITE Narrator Brook Altnabreac Isles of Cornish Catchment Scotland Scllly Mines

Parameter Wells Springs Wells Springs Boreholes Fissures In Granite and Springs pH 4.2-5.9 4.0-4.9 5.3-7.4 4.2-7.7 5-5.8 5-7.2 Chloride 9.0-41.0 9.0-18.0 20-27 14.9-31 107-124 27-11,500 Silica 2.5-10.7 4.2-10.0 4-41 <0.05-17.3 9.1-32 18.6-47.1 Sodium 3.0-22.0 4.0-8.8 13-32 10.7-19.5 36-4,440 Potassium 0.5-2.8 0.5-1.4 0.86-6.9 0.4-3.7 2.6-180

Author Blnnle and Kay,(pers. This study Glendlnlng (1980) Partners comm. (1971) 1982)

7:4 Trends In Groundwater Chemistry; Spatial Patterns In order to determine whether the groundwater In the Narrator Brook Catchment comes from a chemically homogeneous aquifer a number of simple statistical tests were carried out on the chemical data. 7:4:1 The Use of Simple Statistical Techniques to determine Spatial Variations Utilising correlation coefficients as a means of identifying similarity or disparity in the chemical data, a cross correlation of element concentrations between the wells, springs and river water was carried out. By this means it was Intended to identify marked differences between groundwaters at each location, and to form a more rigorous basis for hydr^eological groupings on the variations of groundwater chemistry. The correlation matrix so produced are Illustrated in Table 7:9 in which all correlation coefficients significant at 0.01 significance level are tabulated.

Sites which show two, or less than two significant correlations between the sites, have the least degree of similarity in terms of their chemical characteristics. These are marked with an asterisk (*) in Table 7:9.

- 379 The correlation matrix. Table 7:9, reveals six sites which have little relationship with groundwaters from other locations In the valley. These are Well 7, Sp.ll, Sp.12/13, Well 14, Well 15 and the Narrator Brook. Such anomalies may be accounted for by the fact that different groundwater zones are penetrated by the different wells. The chemical characteristics of groundwaters In the Narrator Valley Is a result of large scale mixing between surface waters. Interflow and groundwater contributions and solute concentration varies widely In space and time. Variations In groundwater chemistry between sites on the south side of the valley (Wells 9, 10, 11, 12, 13, 14, 15) bear little relationship to each other In time, and differ from those on the north side whose chemical variations are in closer accordance with each other (Wells 1, 2, 3, 6). Well 7 Is noticeable In that the chemical variations at this site bear little resemblance to the majority of the other groundwater sources sampled in the catchment (Section 7:5:3).

The likely arrangement of Isolated groundwater bodies on the south side of the valley are Illustrated dlagrammatlcally in Figure 7:17. The suggestion here is, in relation particularly to Wells 13, 14 and 15, that these wells do not penetrate the main groundwater body but intersect Isolated pockets of groundwater perched above the main saturated zone, on clay-lenses. Such an arrangement is supported by the electrical conductivity data in the Narrator Valley. On the south side of the stream high electrical conductivity values occur at higher potentlometrlc surface locations, at Wells 13, 14 and 15, and low electrical conductivities at mid-slope locations, at Wells 9, 10, 11 and 12.

On the north side of the valley lower electrical conductivity values are found at higher potentiometric surfaces (Well 1); higher electrical conductivity at lower potentlometrlc surface (Well 2); with another region of high electrical conductivity at wells 7, 3 and 4. If, after Back and Hanshaw (1970), low electrical conductivity can be attributed to higher recharge, then a rather different recharge regime is apparent for the north and south slopes of the Narrator Valley. The variations observed are attributed to differences in llthology and the penetration of distinct groundwater bodies in the weathered granite profile.

- 380 Table 7:9 Cbrrelation Matrix showing the interrelationships between site geochemistry in the Narrator Valley waters

Sites 1 2 3 6 1 9 10 11 ^.11 12 Sp.12/13 13 14 15 N.B.

T.pH. EC. T.pH. EC. T.pHaEC. pH. pH.EC. T.pH.EC. T.pH. EC. pH.EC T.pH. T.pH. T.pH.EC. pH.EC T.pH. EC. 1 Na. a.Na. Na. * Na. Na. Na. a. a.Na. * Na. * Na.K. K.

2 T.pH. T.pH.EC. T.pH.EC. T.pH.EC. T.iAI. pH.EC T.pH.EC. T.pH T.Si. T.pH.EC. T.pH. T.pH. Na.K. * Na. Na. Si.Na. Sl.Na. Na. Na. K. Si. * *

T.iil.EC. FH.S1. T.pH.EC. T.pHaEC. T.pH.EC. T.pH T.fH.EC. T.pH T.iil.EC. T.pH. T.pH. EC. T.irfl.EC. 3 Na.K. * Na. Na. Si.Na. Na.K. Na. Na.K. K. Na.Ka K.

T.pH. pH.EC. T.pH.EC. T.pH.BC. T.ifl. T.pH.EC. T.pH. T.FH.EC. T.pH. T.pH. T.pH. 6 SI. Na. Na. Na. * Na.K. Na.K. a.Na. a. a.Na. *

T.pHa T.i*I. T.pH. pH.K.Na. T.pH. T.pH. T.pH. T.pH. T.pH. pH. 7 4c Si.K. Si.K. * * * 4c 4c 4c

pflaECa TaidI.EC. pH.Na. T.pH. EC. T.pH.EC. T.pHaEC. T.ffl. T.pH.ECa T.pH. 9 * Si.Na. 4c Na.K. Si. Na.K. Naa aa

Tl.pH. pH.Na. T.pH.BC. pH.a. T. T.pH. T.pH.EC. pH.a. 10 SLNa. K. Na.K. Na. Sl.Na. K. Na.

are significant at 0.01 level F^. T.pH. T.pH. T.pH. T.pHa T.pH. T.pH. 11 Si. a. Na. SI. Sia * 4c * 1 1 1

pHaNa. T.Na. a.Si. pH.EC. pH.Na. T.pH. Sp.ll * two or less than tvo * Na. K. Na. significant correlations

UCLV i T.pH.a. T.a. T.iila T.i*l. T.pH.EC. 12 K. Si. Na. Na.

Na. T.pH. Sp.12/13 * * a.

* pHaEC. Na. 13 Na. *

I^aNaa T.pH. 14 * *

T.f«.Si. 15

N.B. FIGURE 7.17 Idealized Cross-Valley Section Illustrating Likely Disposition

of Perched Groundwater Bodies Within the Weathered

Granite Matrix in the Narrator Brook Catchment

I

CO

I

Perched Waierlabfes on Clay lenses

Assumed Position of main Groundwater Table in Weathered (3- ^ Granite Aquifer cj C3

fractured Surface of Underlying Granite Bedrock 7:4:2 The Degree of Intercorrelatlon of Solutes in Groundwaters from the Narrator Valley The correlation matrix shows there is a general correlation of solutes in the groundwater in the Narrator Valley. This suggests that processes responsible for the groundwater chemistry are active throughout the weathered granite aquifer to a similar degree, and imply a common source for all groundwater in the weathered granite in this part of the catchment. 7:4:3 Temperature Variation with Depth Down the profile temperatures were taken at the end of each month from March 1979 to October 1979, in the Narrator Valley observation wells. The maximum temperature range observed down the profile in any one of the observation wells did not exceed 2°C during the temperature recordings, and for the most part did not vary above 1**C. Water temperature variation with depth generally reflected the annual air temperature variation, with warmer temperatures nearer the surface during the summer months, and colder temperatures during the winter period. This was a noticeable feature in the shallow Wells 3, 4, 6 and 11. Well 7 characteristically exhibited temperatures which were markedly different from those experienced In the other wells having consistently cooler waters at its base varying beween 7,9*0 - 8.6°C, reflecting deeper and cooler groundwater resources at this site. Other basal well temperatures in the Narrator Valley varied from 8.5 - 9.8°C during the 8 month monitoring period. Wells 1, 2, 7, 9, 10, 12, and 14 because of their depths, exhibited water temperature variations down the profile which merited closer analysis. Well 15, because of the depth of infilling (Chapter 3) did not possess a deep enough profile in which temperature variation could be monitored. As a consequence temperature data was not accessible from this location. The standard deviation of temperatures, measured at a specific depth below the watertable in each well, is illustrated in Figures 7:18a-c. These figures Illustrate the variability in temperature measurements down the well-profiles which can be related to the Inflow of groundwater at each site. With the exception of Well 9, the standard deviation of the water temperature was greatest in the top

- 383 - few metres. This is to be expected since this is the zone in the wells in which water level fluctuations take place, and ambient air temperatures can rapidly modify groundwater temperatures. Well 7 and Well 9 show the most marked variations of temperature with depth. Figure 7:18b. Well 9 shows the largest variation in temperature towards the base of the well. This is interpreted as the result of the mixing of more than one groundwater feeder source at this location, possibly as a by-product of spill-over from adjacent perched groundwater bodies as outlined in Section 7:5. In addition the larger hydraulic conductivity at this site (Chapter 6) will facilitate more rapid groundwater movement. Well 7 shows the most variable standard deviations of temperature with depth down the well profile, suggesting the mixing of inputs from various levels within the weathered granite matrix. Well 10, Figure 7:18c, exhibits standard deviations of temperature which decrease with depth. Such smaller temperature variations down the profile may be related to the constancy of the groundwater source at this location, and result in the development of temperature stratification. Wells 12 and 14, as particularly Well 7 illustrate the development of mixing In the well column resulting in variations in the standard deviations of water temperature with depth. Such a feature is not as strongly marked in Wells 1 and 2 (Figure 7:18a), suggesting that groundwater at these locations has a deeper and more uniform source, as suggested in Section 7:3:2. The combinations of vertical and lateral flow, rapid and delayed flow will Impart a characteristic temperature to groundwaters contributing to the wells at each location. An infinite number of combinations of differing groundwater sources, producing temporal and spatial variations in groundwater temperatures in the heterogeneous weathered granite aquifer, will result in a gradual temperature stratification, or temperature zonation in the profiles of the wells. If rapid percolation of precipitation water down the well annulus in the valley occurs it is thought that any temperature variation produced in a given well would indicate this. It might be expected that much higher temperatures would be found at the base of the well during the summer months due to the rapid percolation of warmer

- 384 12 WELL 1

13

14 H

15

16 i

17

18

19

005 0-1 0-2 a3 0-4 0-5

STANDARD DEVIATION OF TEMPERATURE

WELL 2

E 9

UJ 10

O £ 11

^ 1?

§ ^3

14

15-

16

—I 01 02 0-3

STANDARD DEVIATION OF TEMPERATURE

FIGURE 7 :18a RELATIONSHIP BETWEEN DEPTH AND TEMPERATURE IN THE WELL PROFILE

IS5 ' FIGURE 7:18b

WELL 7

7

8

lU

O 11

12 J I ag 13

X o

—I— —I— —I— 0-1 0-2 03 0-4 0-5 0-6 0-7

STANDARD DEVIATION TEMPERATURE

WELL 9

3-1

4-

5-

6-

o S 8 g 9H

O 11 X g 12

I I 0-2 10 20 30 STANDARD DEVIATION TEMPERATURE

3^^ - 3

4 WELL 10 5 H

6

7

8-

9-

1 1 1 1 1 1 1 T 0-4 0-8 1-0 1'4 1.6

STANDARD DEVIATION OF TEMPERATURE

2 - WELL 12

3 -

4

5

6

7

8 i

9

04 1-4 20 STANDARD DEVIATION OF TEMPERATURE

6 WELL 14

7H 111

8

9

10

11 -

12 -

T 1 1 o7 0-4 ' 0'.6 * 0-8 STANDARD DEVIATION OF TEMPERATURE FIGURE 7:l8c precipitation waters downwards, and conversely, much colder temper• atures in the winter months. The temperature variations with depth observed in the Narrator Valley do not present sufficiently strong evidence to corroborate or refute this option, although some of the observed temperature varia• tions may be explained in this manner. On the whole the narrow temperature range observed In the wells over a short monitoring period is not sufficient to enable estimates to be made of the recharge source and residence time of groundwater in the weathered granite aquifer of the Narrator Valley. 7:4:4 Summary of Spatial Patterns Considerable spatial variation In the chemistry of the ground• waters in the Narrator Valley has been emphasised by the correlation matrix produced from a cross-correlation of element concentrations between wells, springs and river water. Variation in groundwater chemistry between sites on the south and north sides of the valley bear little relationship to each other. Chemical variations determined in Well 7 again bear witness to the anomalous conditions experienced at this site which is evident from llthological conditions (Chapter 2), temperature variations (Chapter 5) and hydraulic properties (Chapter 6). On the south side of the valley the lack of significant corre• lations between the chemical characteristics of the springs and Wells 13, 14 and 15, support the hypothesis that these wells penetrate water bearing zones, which are different from those of the springs. On the north side of the valley significant correlations between Wells 1 and 2 indicate that these wells penetrate a similar deeper groundwater body. Temperature variations down the well profile are related to groundwater inflows at each location. Wide variations in temperatures particularly on the south side of the valley are due to the mixing of groundwaters from a number of different sources. Temperature strati• fication in the well profile is related to zones where the groundwater has a constant origin and less groundwater mixing takes place.

388 - 7:5 Rate of Water Movement Towards the Watertable In the Narrator Brook Catchment As chloride Is a conservative anion and has a low background concentration in Igneous rocks, any variation in chloride content of groundwater in such rocks Is likely to be due to variations in rainfall content and Its inputs into the groundwater system. The chloride ion does not form strong complexes with cations unless the concentrations in solution are high (Hem, 1970). In fact chloride has a very subdued role in most hydrogeologlcal systems and as such, was chosen for use in the Narrator Brook Catchment as a form of natural tracer. 7:5:1 The use of Chloride as a tracer The mean monthly chloride values recorded over the research period are illustrated in Figure 7:4. This figure indicates that with the exception of Well 7, chloride peaks in precipitation appear in groundwater some 1 to 2 months later. Using the chloride concen• trations of rainfall as a natural tracer, the rate of movement of water from the surface to the watertable in the catchment was determined using a simple statistical approach. Chloride concen• trations in rainfall at a specified date were correlated with chloride concentrations in groundwater in successive time periods after the rainfall event, until a significant correlation coefficient was achieved at an individual well for a particular 'labelled* rainfall input. For this analysis Wells 1, 2, 7, 9, 10, 11, 12, 13, 14, and 15 were used, and correlated with the chloride concentrations recorded in precipitation at the nearest raingauge sites (RF7, RF9, RFU). Wells 3, 4 and 6 were excluded because of possible contamination by flood waters. Analyses were based on the period February 1979 - July 1979. Chloride values were similar at all these rainfall sites (Figure 7:4), as were sodium, potassium and electrical conductivity. A tendency for these components to increase slightly with site elevation was observed. Lag times derived from this analysis ranged from 5 to 8 weeks (Table 7:10, Figure 7:19). Lag times using sodium and potassium data

389 were not calculated on account of the varied sources of these Ions, the effects of cation exchange processes, and the adsorption of potassium onto clay.

Table 7:10 Lag Times in the Narrator Valley

Chloride Well Rainfall Site Weeks Correlation Coefficient

1 R.F.7 8 0.57* 2 R.F.7 8 0.89** 7 R.F.7 8 0.48^^ 7 S.Fa 7 0.81* 9 9 5 0.88** 10 11 7 0.93** 11 11 8 0.95** 12 11 6 0.85** 13 11 7 0.95** 14 9 6 0.75* 15 9 5 0.74*

N.S. Not significant * Significant at 0.05 level ** Significant at 0.01 level

7:5:2 Variation in Groundwater Lag Times Figure 7:19 illustrates the distribution of the lag periods over the groundwater network. On the south side of the Narrator Brook stream lag times are from 5 weeks at higher slope positions to 8 weeks at the slope basea On the north side of the valley lag time are all in the region of 8 weeks. This variation in lag times may be attributed to the site location. Spatial variation in these lag times reflect the downslope movement of infiltrated water from the main recharge areas near the groundwater divides, and the velocity of groundwater flow. Wells in mid-slope or valley bottom positions have lag times in the order of 8 weeks In contrast to those on the upper slopes of only 5 weeks. With continuing rainfall a saturated area progresses back up-slope (Kirkby 1978). This saturated area is easily visualised In the form of a saturated * wedge*, which will be of limited extent.

390 The development of a saturated 'wedge' in the Narrator Brook Valley produces a pressure head which increases downgradient and is a contributing factor in the rapid response of the watertable in the catchment. This may partly explain the discrepancies between the water table response of 1-2 weeks in periods of extreme precipitation (Chapter 5) and those of chloride tracer work giving lag periods of 5- 8 weeks. The saturated 'wedge' may act as a plunger displacing 'older' water (antecedent rainfall contributions) downgradient towards wells at lower elevations. Figure 7:20. This displacement movement has been termed by Hewlett and Hibbert (1967) as translatory flow and is not necessarily restricted to saturated flow conditions. As a consequence of translatory flow the chloride tracer in the groundwater will be noted at an early stage in the higher-elevation wells, and much later at downslope locations. By the time a labelled chloride input is determined in lower wells the next labelled input is being recorded at higher well sites. The simple model outlined above does not take into account sub• surface conditions. It assumes that aquifer properties are Isotropic which in the Narrator Valley is not the case. The occurrence of clay lenses have been invoked to explain spatial and temporal differences in water chemistry between sites, particularly on the south side of the valley. Textural variations in the weathered granite aquifer are likely to vary in the vertical and horizontal planes. This gives rise to anisotropic conditions where the vertical hydraulic conductivity (K^) and the horizontal conductivity (K^) vary. Textural variations will produce zones with different hydraulic properties within the weathered granite aquifer of the Narrator Valley. These will effect the relative velocities of lateral and vertical water movement, and the speed with which a labelled chloride input reaches a given well. Discrete layers of low permeability material may be visualised, as a stack of clay lenses. These clay lenses are off-set in the lateral plane with intervening zones of higher permeability materials between lenses, (Figure 7:17). Perched water-bodies on these clay lenses may not contribute directly to the main groundwater zone, except under very wet conditions.

391 Barrator Reservoir

^12 We// Old Asp 11 Spring S/fes

^ Log rime "Confours'in Weeks (

Scale: 5*5cm = 250m

F/GURE 7:19 Distribution of Lag Times in the Narrator Brook Catchment based on Chloride values RECHARGE

FIGURE 7:20 WATER DISPLACEMENT MECHANISMS IN THE NARRATOR VALLEY

Chloride 8 weeks /weeks 6 weeks lag times 3rd Rainfall Input v

Pressure heads 2nd Rainfall input transmitted down gradient from saturated

wedges' let Rainfall input

MAIN/GROUNDWATEw/m/mR /ZONE v////mF/////////. In rainfall periods of long duration rapid vertical and lateral movement towards the clay leases may cause the perched watertable to rise and overspill as the lateral areal extent of the lens Is attained. This excess water Is transferred to the outer edges of an underlying clay lens, and mixing of the solutes In such waters takes place. If a saturated 'wedge* Is in the process of migrating back- upslope, then at certain elevation some of the clay lenses will be incorporated. This further mixing of solute bearing waters will have a marked influence on the concentrations of groundwaters at various heights in the weathered aquifer. The statistically derived lag times therefore reflect the operation of many complex hydrogeological processes within the weathered granite aquifer* A number of chemical processes operative within the regolith may also effect the calculated lag times leading to an overestimation or underestimation of the true travel time of the groundwater. Hem (1970) suggested that because the chloride ion is physically large compared to many other major Ions in groundwater, it could be expected to be held back in the Interstitial, or pore waters in clays, while the water itself is transmitted. Kurtz and Melsted (1973) suggested that negative or positive adsorption of the chloride ion may also be encountered to a limited extent. Feth et^ al. (1964) on work in granitic areas of the Sierra Nevada, suggested that there is considerable evidence that chloride is removed from solution in some places by adsorption on clay. If this is so, then where conditions of clay content of soil and pH are suitable for anion sorption, chloride cannot confidently be used as a geochemlcal tracer (Feth et al. 1964). Feth^^. (1964) suggested that sorption of chloride is only effective in clay soils where the pH is less than about 6.8. Although the pH of the Narrator brook groundwaters is less than this, (pH 4.2- 5.9) particle size analysis of soil and weathered materials to depths of 3m in the catchment. Chapter 2, has shown that clay and silt fractions are less than 2%. Such low values will reduce the effectiveness of any sorption of chloride which could take place in the catchment.

- 394 - The observations of Feth et al, (1964) were based on the depletion of chloride from snow pack waters to groundwater concentrations of chloride, which were much lower. In the case of the Narrator Valley there is very little difference between mean chloride concentrations in groundwaters and those in rainfall (Table 7:4). The chloride ion then, is conservative, and remains in the groundwater where, once in solution it is not removed. On this basis it appears to be an ideal natural tracer for use in weathered granite terrains with low clay contents. Where groundwater bodies are isolated on clay lenses, and spill-over processes are operational causing groundwater mixing and recharge, the chloride reflects these mixing processes and gives the travel time of the groundwater involved. Chloride used as a tracer would be most effective in an 'ideal' isotropic aquifer, or in a region where there is the opportunity to sample every contributing flowpath. The necessity of monitoring every part of the flow system is unrealistic in terms of regional groundwater recharge potential, and emphasises the problems entailed In assessing the groundwater resources in a weathered granite aquifer. From the methods discussed so far In this chapter it can be seen that generally groundwater at each location in the Narrator Valley has similar chemical properties, whose concentrations vary spatially and temporally. These solute concentrations, particularly the chloride are dependant upon the hydrogeological complexity and elevation of the site in the catchment. The groundwater flow sampled at Well 7 proved to be consistently different from other sources and merits further consideration here. 7:3:3 Anomalous Groundwater Conditions Chloride concentrations in groundwater from Well 7 were consistently higher (mean 23.6 mg 1"^) than any other well or spring site. In contrast to the other sites in the catchment the chloride concentrations correlated significantly (at 0.05 significance level) with chloride concentrations in stemflow, which suggest that the anomalous hydrogeochemlcal character of Well 7 may be related to inputs of water derived from stemflow sources. Such stemflow sources cannot however be the sole origin of the chloride in the groundwater as the chloride lag time at Well 7 is in the order of 7-8 weeks (Table 7:17). Additionally the response of the watertable to rainfall events

395 - has a delay factor of up to a month, and In this respect is similar to the other wells In the catchment. Lag times of all the wells are from 1-2 weeks after intense rainfall events. Well 7 Is also anomalous in terms of its silica concentrations, Figure 7:5, Tables 7:5 and 7:13, being consistently lower than the groundwaters at the other sites. This lends support to the hypothesis that Well 7 receives some component of recharge from waters which have a very short residence time within the weathered granite* Rainfall and stemflow waters contain only trace concentrations of silica. A rapid input of these waters at Well 7 would therefore result in the dilution of groundwaters with longer residence times and higher silica concentrations. It might be expected that during periods of no, or little rainfall, silica concentrations in the groundwaters would increase, but this has not been observed during the research period. An additional explanation of the low silica levels at Well 7 may be due to the abstraction of silica during the formation of Ca- montmorillonlte. Jacks (1973) noted that the formation of Ca- montmorillonite from kaollnite in Swedish granite and gneisses, required large amounts of silica and relatively small amounts of calcium. The pre-existence of hydrothermal kaolinite at depth in the catchment In altered granite cannot be ruled out but is not a matter for dispute here. The groundwater temperatures at Well 7 are lower than the rest of the groundwaters sampled with a mean value of 8.5**C (Table 7:5). The temperature range is from 7.8**C - 9.0**C with the highest temperatures recorded in the winter months October-December. With a maximum mean air temperature in July of 21.5**C (Figure 7:1b) the implication Is that there exists a 3 month temperature lag at Site 7. This would suggest a much deeper contributing source at this locality. If rapid stemflow and rainfall contributions are responsible for the anomalous character of Well 7 then it would be expected that groundwater temperature variation would reflect such rapid recharge pulses. The quicker the watertable Is reached the less will be the change in the temperature of the percolating water and the greater the heating or cooling effect of this recharge input on the groundwater already there will be. Such rapid variation has not been observed at Site 7. Variation of temperature down the well profile (Section 7:4:3) however, illustrates a range in the standard deviation of groundwater

396 temperatures over the research period which are Interpreted as mixing of groundwater sources In the well column at this site* The anomalous groundwater chemistry at Site 7 may be explained by mixing mechanisms which are illustrated dlagrammatlcally In Figure 7:21. From the well logs (In Appendix 1) It appears that Well 7 penetrates over 3m of solid granite at Its base, having previously passed through the overlying unconsolidated sands, grits and clays* It Is assumed for purposes here than this granite is fractured and is continuous with Sheepstor granite bearing contlgous fractures and fissures. The hydraulic conductivity and transmlssivity values determined at this site (Chapter 6) are the highest in the catchment and fracture systems may enhance these hydraulic properties. Precipitation recharge to the bare granite areas of Sheepstor, Figure 7:21, feeds directly into this fractured granite layer at depth, contributing waters with high chloride and low silica concentrations. Chloride additions from fluid Inclusions bisected by fracture surfaces may also contribute on a very small scale to chloride levels in the groundwaters In the fissured and fractured granite, and leakage of more concentrated waters from the base of the unconsolidated material above cannot be ignored. In periods of heavy rainfall a pressure head can be transmitted through the fracture system and as a consequence of this, similar response times of the watertable to input are apparent at Well 7 and in the other wells in the valley. Both the chloride lag time and the groundwater temperature lag time determined at Well 7 supports this hypothesised movement of recharge water through the fractured granite system at this location. 7:6 Geochemlcal interpretation The geochemical interpretation of chemical analyses of water from igneous areas usually Involves two main approaches in the literature. The first involves plotting the data on stability diagrams to determine what may be the stable alteration products, a method utilised by many authors for example Bricker (1968), Feth et al. (1964) and Jacks (1973). Residence time of the groundwater in the hydrogeologlcal system can be deduced from this approach (Jacks 1973) and the extent to which minerals and their alteration products Influence the composition of natural waters can be determined (Bricker 1968).

397 Precipitation

SHEEPSTOR

4-

-f FIGURE 7 21 Proposed Groundwater Pathway Pop/d Movement of to Well? in the Narrator Brook Precip/fofion Downwards Catchment Through Fissure Systems 14 /o Feed Groundwafer F/ow in the Loca/ity Rapid Infiltration down well Weathered Granite of Well 7 7 annulus Infill and Clay Lenses - -

035

+ +

Infill Fissured and Fractured Granife The second interpretive approach involves modelling of the water chemistry by calculating reaction sequences. This is done by reacting primary minerals to product clay minerals and solutes (Cleaves et al. 1970) or by reconstituting the primary minerals through combining the clay minerals present with the dissolved products observed in groundwater (Carrels and MacRenzle 1967). A complete analysis for dissolved constituents in the groundwaters of the Narrator Valley was not undertaken and as the specific chemical nature of the clays in this part of the catchment are unknown, calculation of reaction sequences was not considered an appropriate technique. Consequently the stability diagram method was adopted. 7:6:1 Equilibrium Relations If the assumption is made that silicate minerals are in equilibrium with the waters in the pores, the interrelations of the minerals can be shown as functions of the activities of the ions dissolved in the water (Carrels and Christ 1965). Although thermodynamic data for silicates are sparse it is possible to develop qualitative stability diagrams that are useful to provide a graphic summary of the mineral sequences that might be expected if equilibrium conditions were attained. By means of an example: the Microcllne dissolution reaction of Table 7:1 expressed in mass-action form becomes

[K+1 [Si (0H)u]2 *^mlcro-ICaol " where K . __ - is the equilibrium constant and the bracketed mlcro-Kaol ^ quantities are activities. This equation can be expressed in logarithmic form as

which indicates that the equilibrium condition for the mlcrocllne- kaolinite reaction can be expressed in terms of pH, and activities of K+ and SKOH)^. The other reactions listed in Table 7:1 can also be expressed in this manner in terms of Na"**, or SKOH)^ and (or pH), assuming that aluminium is conserved in the chemical reactions.

- 399 These equilibrium relations are the basis for the construction of stability diagrams. The lines that separate the mineral phases In such diagrams represent equilibrium relations such as K^j^^^^ Kaol' Since minerals in real hydrogeologic systems do not have ideal chemical compositions these stability lines, based on thermodynamic data for relatively pure mineral phases, probably do not accurately represent field conditions. These types of diagrams have been found by many workers (Feth et^al. 1964, Garrels and MacKenzie 1967, and Jacks 1973) to serve a useful purpose In the Interpretation of chemical data from hydrogeological systems, and as such have been utilised for the Narrator Brook Catchment. The stability diagram used here is of the type that was originally developed in an attempt to investigate silicate mineral relations in the zone of weathering. In rocks at shallow depths permeated by groundwaters, and in sediments undergoing shallow diagenesis. For these situations, diagrams constructed at 25*'C and 1 atmosphere total pressure, after Garrels and Christ (1965), are appropriate and have been used for the Narrator Valley data. In very dilute aqueous solutions the molality of concentrations can be used to determine equilibrium and solubility (Fetter 1980). As granitic waters have low concentrations of dissolved constituents, and the values for free energy of formation of the constituents are not known with a high degree of certainty, the assumption that the activity of any solid is unity and that the molality of soluble constituents equals activity seem justified (Feth et al. 1964) and have been adopted here. Figure 7:22 clearly demonstrates the overall similarities in the chemical characteristics of groundwater from the wells, springs and streamwater. The waters from the Narrator Valley plot in the kaollnite field of the stability diagrams which accords with observations made by numerous Investigators (Feth et al. 1964, Garrels and MacKenzie 1967, and Tardy 1971) in groundwaters from igneous terrains. Water quality analyses from interflow, stream and springwaters higher upstream in the Narrator Brook Catchment, (Williams e_t al. 1981) also plot in this field, although some interflow waters tend towards gibbsite equilibrium.

400 F^uartz I saturation amorphous Silica ^ saturation K-MICA K- FELDSPAR (K)" Log _(H)_

^0- GIBBISITE 30- KAOLINITE

MONT MORILUDNITE tol- 05 00 14/12-11 ^sp.11 -05 sp12/i3j -1.0 -15 -2.0 70 -6.0 -50 -4.0 -3.0 -2.0 LoglSilOH)^]

FIG. 7:22 STABILITY FIELDS AMONG SOME SILICATE MINERALS IN THE GIBBISITE - KAOLINITE - MICA- FELDSPAR SYSTEM AT 25oC AND 1 ATMOSPHERE PRESSURE . FROM NATURAL WATERS IN THE NARRATOR BROOK CATCHMENT.

4-0/- According to Freeze and Cherry (1979) a small percentage of groundwater samples from igneous areas plot in the montmorillonlte fields, and hardly any occur in the gibbsite, mica or feldspar fields, or exceed the solubility limit of amorphous silica. Those plotted from the Narrator Valley have dissolved silica contents around the solubility of quartz* The occurrence of most igneous-derived groundwaters in the kaolinite stability field suggests that alteration of feldspars and micas to kaolinite is a widespread process in groundwater systems in igneous areas. For groundwaters to evolve toward equilibrium with respect to the primary silicates such as feldspars it is necessary for the concentrations of SiCOH)^ and cations (K*** Na"*") to progressively increase as dissolution proceeds. If the reaction products in the pore water in weathered granite are continually being flushed out by groundwater movement appreciably more rapid, relative to chemical reaction rates, then equilibrium with respect to primary silicate minerals will never be attained. In the weathered granite aquifer of the Narrator Brook It seems unlikely that the water chemistry in weathered granite aquifers would evolve to the Na-feldspar field because long periods of time and sluggish flow conditions would be required before this equilibrium is attained. Hydrogeochemical investigations in the Archaen granites and gneisses of Sweden by Jacks (1973) further suggested that the composition of groundwater from igneous rocks is governed by mineral- water equilibria. As such equilibria are established slowly it is possible to use these as a rough relative dating technique, differentiating between more recently infiltrated water and substantially older water which has time to equilibrate towards Na- montmorillonite and feldspar phases. As can be seen from the water fluctuation analysis (Chapter 5) and by the chloride tracer approach in this chapter, groundwater movement through the weathered aquifer can be rapid, in 1-8 weeks, thereby excluding the potential to attain Na-feldspar equilibria. 7:6:2 Reaction Rates and Molecular Diffusion Chemical changes affecting groundwater depend on the rate at which a particular body of water moves through the various porous media of the groundwater zone. Because rates of chemical reactions

402 are slow, the bulk mass of groundwater often remains undersaturated with respect to minerals that occur in the porous media. Rates can be slow because ions are not easily released from the crystal structures, or the flux of water and reaction products between the bulk mass of the flowing water and crystal surfaces is slow. Very little is specifically known about rates or mechanisms of reactions in natural waters (Hem 1970). The length of time water is in contact with the rock materials controls the amount of solutions in two ways: 1. If a steady stream of turbulent water is passing over mineral surfaces the water may be moving so rapidly that the elments leached out of the rock do not enter into solution rapidly enough for the water to reach saturation, therefore solute concentration in the body remains low. 2. In small capillary spaces, or in very slow moving water bodies in the saturated zone, the movement of water may be so slow that solute transfer away from the rock-water interface may be impeded. The solute transfer rate may be controlled by molecular diffusion of the reaction products through the fluid in the smaller pores into larger pores, where they are transported in the active hydraullcally controlled flow regime* In a groundwater body differences between active water circulation zones, and the differences in hydraulic conditions within the porous media of each zone, all affect the solution process and transfer rates, through time. In hydrogeologlcally complex areas like the Narrator Brook Catchment differences in hydraulic conditions in the weathered granite aquifer are infinite, and the multiple effects on solution and transfer rates are immense. 7:6:3 Groundwater Quality The potability of a water supply is probably the most important criterion of its quality and generally requires the most stringent quality controls. To establish water quality there must be physical and bacteriological analysis as well as chemical analysis* Such analyses have not been undertaken on the groundwaters of the Narrator Valley, and the chemical analysis carried out can only be used as a preliminary guide indicating the suitability of the Narrator Valley groundwaters for drinking water supplies.

- 403 Cook and Miles (1973) provide a reference table specifying limits for various chemical elements found in groundwaters. Part of this table is utilised here in relation to the ions analysed in the Narrator Brook Catchment.

Table 7:11 Drinking Water Quality Guide

Element Typical concentration Narrator Drinking water limits in natural waters Valley E.E.C. guide level (mg l-h (mg 1-1)

ci- 10-2000 9-41 25 Si02 1-30 2.5-11.0 not specified Na+ 10-1000 3-22.0 20 K+ 0.2-30 0.5-30 10 but maximum admissible concentration 12

As can be seen from Table 7:11 the solute concentrations from groundwaters and surface water in the Narrator Brook Catchment are well within E.E.C. specified limits for good potable drinking water supplies. The Narrator Brook Catchment is one of three catchments which contribute waters to Burrator Reservoir, which provides good- quality drinking water to Plymouth and the surrounding areas. 7:7 Conclusions The methods utilised, to portray the nature and pattern of hydrogeochemical conditions in the Narrator Brook Catchment, have provided evidence of considerable spatial variation in the groundwater chemistry. Variations In groundwater chemistry between sites on the south and north sides of the valley bear little relationship to each other. This pattern is related to the elevation and hydrogeologlcal complexity of individual sites, and to whether the wells penetrated perched groundwater bodies or the main groundwater zone. Large and small scale textural variations in the weathered granite matrix are considered to produce different contributory-flow volumes to the watertable at a given site dependant upon the relative permeabilities of these horizons. In the Narrator Valley contributions to

- 404 groundwater in the vertical direction are thought to be small and hence insignificant compared to horizontal groundwater movement. An exception to this condition is provided by the potential for more rapid Infiltration down the annulus of the observation wells, particularly if these Intercept a dominant contributory flowpath. The variable nature of the physical properties, and the thickness of the valley infill, and the lateral extent of clay lenses all contribute to anisotrophy in weathered granite aquifers. Recharge to the aquifer is controlled by topographic and llthologlcal features. Movement Is predominantly downslope towards the Narrator Brook. Responses to recharge events are observed from watertable fluctuations to be from 1 to 4 weeks, while the chloride tracer Investigations suggest lag times in the ranges of 5-8 weeks. This difference in lag times Is attributed partially to an increase in hydraulic pressure head in the downslope direction, thereby inducing a more rapid watertable response; and to the action of translatory flow first pushing out 'older* water downgradlent and hence retarding the downslope movement of more recent rainfall with its characteristic chloride content. The presence of clay lenses cause the ponding of isolated water bodies above the main saturated zone, and the subsequent spill-over effect in extreme recharge events, will modify the chloride concentration in such mixed waters. The presence of clay zones in the valley infill may retard the passage of chloride ions, although no attempt was made to determine the significance of this effect in the catchment. The chloride ion is assumed to have a subdued role in hydrogeological systems, and seems in the first Instance an Ideal natural tracer for determining rates of recharge in weathered granite areas. In weathered granite zones with clay lenses, and other isolated horizons of low permeability with marked anisotrophy (associated with differing large and small scale textural variations), recharge Is not predictable, and the usefulness of a tracer is much reduced. Consideration of chemical equilibria in groundwaters from the Narrator Valley clearly demonstrate the similarities in the overall chemical characteristics of waters sampled. Groundwaters and surface waters from the catchment plot in the kaolinite stability field, which is in agreement with observations made in waters from upstream

405 locations of the Narrator Brook Catchment (Williams et^ al. 1981). Similar features in groundwaters from igneous terrains in American (Feth^t^al. 196A) and European localities (Tardy 1981) suggest that the alteration of feldspars and mica to kaolinite is a widespread process in groundwater systems in igneous areas. In the Narrator Brook aquifer, groundwater movement through the system is rapid, and as a consequence the potential for such waters to attain equilibrium in the Na-feldspar fields is restricted. Using the limited chemical data available from the Narrator Valley, the potability of groundwaters for domestic supplies was assessed. The solute concentrations in groundwaters and surface waters within the valley aquifer are well within E.E.C. specified limits for drinking water, a conclusion which was previously assumed since the catchment contributes to storage in Burrator Reservoir which in turn supplies good quality domestic water to Plymouth and Its environs.

- 406 Chapter 8 The Potential of the Narrator Brook Aquifer: Summary and Conclusions

8:1 Introduction This chapter attempts to summarise the main features determined from investigations in the Narrator Brook Valley, and to draw some general conclusions concerning the usefulness of isolated weathered crystalline aquifers, in maintaining water supplies to a region. An overall view of a weathered granite aquifer is presented, as typified by conditions in the Narrator Brook Catchment. The alms of this investigations were outlined in Chapter 1 and are stated here as the basis for the ensuing discussion of the hydro- geology of the Narrator Valley aquifer. The alms were as follows: (1) to assess groundwater behaviour in a weathered granite aquifer by the analysis of watertable fluctuations; (11) to delineate recharge and discharge areas; determine the time taken to transmit recharge to the watertable; and determine the direction of groundwater movement and the extent of hydraulic continuity between surface and subsurface systems; (ill) to determine the hydrogeological characteristics of decom• posed materials including the hydraiUic conductivity, porosity, transmissivity and storage values of weathered granite materials; (Iv) to determine the nature and thickness of the saturated horizons and depths to solid bedrock; (v) to assess the degree of success with which watertable obser• vations in a portion of the aquifer may be used to predict groundwater conditions in upstream sections of the valley aquifer. The characteristic features of the Narrator Brook aquifer are summarised and discussed in section 8:1, and these features determine well groupings as outlined in section 8:2. The ability to predict future groundwater behaviour in the valley is discussed in section 8:3, and the potential of the Narrator Brook Catchment as an Isolated weathered granite aquifer is outlined in section 8:4. The general form and diverse hydraulic properties of weathered granite aquifers.

- 407 as typified by the Narrator Brook are presented in section 8:5. Section 8:6 presents proposals and recommendations for future investi• gations concerned with the assessment of weathered granite aquifers, based on the findings from the Narrator Brook investigation.

8:1:1 Characteristic Features of the Narrator Brook Weathered Granite Aquifer Geology, surface and groundwater discharge regimes, water-level fluctuations, aquifer properties and hydrogeochemlcal aspects of the Narrator Brook Catchment have been presented fully in Chapters 1, 2, 4, 5, 6 and 7. These features have been discussed almost exclusively in isolation in each of the relevant chapters, in an attempt to define clearly their characteristic properties. In reality groundwater and surface water regimes and their associated properties are inter• related, and need to be viewed in this context. Bearing this factor in mind this section summarises and discusses the characteristic features of the Narrator Brook Valley aquifer. 8:1:2 Lithologlcal Controls Apart from investigations carried out on Burrator and Sheepstor Dams (Sandeman 1901), very little information prior to the drilling of the observation wells used in this study, is available on the nature and depths of incoherent materials overlying solid granite in the Narrator Valley. Past mining activities, discussed in Chapter 2 have yielded no information on the depth and sub-aerial extent of the weathered granite in the Narrator Valley. Surface geophysical invest• igations using both the hammer seismograph and resistivity techniques were undertaken in the Narrator Brook Catchment, in an attempt to determine definitive thicknesses of weathered granite in the catchment. These gave inconclusive and spurious results. In the case of the hammer seismograph the thicknesses and variation in physical properties of the unconsolidated materials in the valley absorbed or dispersed the shock waves produced by a hammer impact on a metal plate, making interpretation extremely difficult. It is considered that the use of a shot down an augered hole might be more appropriate, but limitations of the equipment and lack of an explosives licence precluded further investigations.

- 408 - The results of resistivity Investigations into depths to solid granite in the Narrator Brook Catchment using the Wenner configuration were also inconclusive. Burgess^ (pers. comm. 1978) found the resis• tivity technique using a Wenner configuration, to be successful in determining the thickness of alluvial material or strongly weathered granite above a granitic basement in the Isles of Scilly, to depths of up to 10m. Work in the Scilly Isles (Burgess, pers. comm. 1978) suggested that penetration depths greater than 10m may be possible in weathered granite terrains, providing more powerful equipment and longer spreads are used. In the case of the Narrator Valley, it is evident from obser• vation well drilling that depths of weathered granite materials are in excess of 20m. Depths greater than 30m were recorded at Sheepstor Dam by Sandeman (1901), and in the adjoining Newleycombe Lake Valley some 10m of weathered granite are exposed in a gully section (Murgatroyd pers. comm. 1978). It seems likely that the depth to solid rock will produce an uneven base, as suggested by the cross-valley profiles at other locations on Dartmoor as outlined in Chapter 2, and on this basis an average thickness of weathered granite of 15m was adopted in the case of the Narrator Catchment. A saturated thickness of 12m has been assumed in the catchment area (Chapters 4 and 3). The observation^ boreholes sunk in the Narrator Catchment revealed lithologies containing large quantities of clay, sandy clay, sands and grits, with gravels and cobbles. All deposits were of varying thicknesses as can be seen from the borehole logs in Appendix 1, and little correlation was evident between sites. This great lateral diversity of materials is hydrogeologically significant since it results in retardation and enhancement of recharge, a factor which is illustrated by the differing responses of the watertable to individual precipitation events in the catchment (Section 8:1:3). Particle size analysis carried out on samples taken from the Narrator Valley show that 24-43% of the materials are gravels, 56-76% sand, and less than 2% silt and clay. Eden and Green's (1971) analysis of in situ growan samples from Dartmoor give a silt and clay content varying from 13.5-28% with clay being from 2-10.5%. Soil materials overlying sedentary growan were found to be composed of 57- 62.5% sand, 26.5-29.5% silt and 11-13.5% clay (Eden and Green 1971).

409 - The samples taken from the Narrator Valley weathered granite aquifer were from depths less than 3m, and as such may not be repre• sentative of the overall aquifer. They do however suggest that there is a larger proportion of coarser materials in the valley than estimated visually from borehole logs, which has significance in terms of groundwater response times and the velocity of groundwater movement. From the analysis of the well hydrographs in Chapter 5, litho- loglcal factors were found to have a significant effect on the fluc• tuation range experienced at each location. Wells in clay materials (Wells 9, 12, 14 and 15) showed the largest fluctuation ranges, while wells in areas of sands, gravels and fissured granite areas, show intermediate ranges (Wells 1, 2 and 7) and the smallest water-level fluctuation ranges are experienced in wells penetrating mixtures of gravel and cobbles adjacent to the stream channels (Wells 3, 4, 6 and 11). The water-bearing horizons penetrated in the Narrator Valley are visualised, in the vertical sense, as a stack of clay lenses. These clay lenses are thought to be off-set in the lateral plane, with intervening zones of higher permeability materials between the lenses. This hypothesis is supported from the distribution of materials illu• strated in the borehole logs, (Appendix 1). The distribution of these clay zones, as determined from groundwater level analysis. Chapter 5, appear to be a feature more characteristic on the south side of the Narrator Valley, although their occurrence cannot be ruled out on the northern slopes. The clay lenses are considered to be everywhere thin and impersistent and may represent a zone of deposition (illuvlatlon lenses) of leached materials of fine clay and silt particles within the weathered granite matrix. Geochemical data collected in the Narrator Valley suggest that individual well waters, particularly on the south side of the valley, are characteristic of isolated water• bearing horizons. Variations in well chemistry are the result of the time taken for the groundwater to move through the weathered matrix, and subsequently of the pathway taken. In the case of the Narrator Valley aquifer, the borehole logs support the interpretation of the aquifer as a series of clay lenses resulting in perched watertables of differing elevations. Most of the

410 - observation wells In the Narrator Valley do not appear to be deep enough to intersect the main groundwater zone, with the possible exception of Wells 1 and 2 on the north side of the valley. Well 7 is in an anomalous position, whose features are interpreted as being a mixture of fissure-flow waters, rapid recharge of waters in the immediate locality of the well and slow leakage from the base of the weathered zone (Chapter 7). 8:1:3 Hydraulic Properties of Aquifer Materials Textural variations in the near surface soil and weathered granite regollth on hill slopes up valley from the groundwater obser• vation network have been shown to have a significant influence on movement of interflow through the weathered granite (Ternan and Williams 1979). The movement of water along textural horizons was found to vary spatially and temporally (Williams et. al» 1981). Such textural variations are likely to be present in downstream locations in deeper horizons and give rise to varying aquifer properties. Table 8:1 illustrates the mean hydraulic conductivity and mean transmissivity values determined at the sites tested in the Narrator Valley. In general the transmissivity and hydraulic conductivity become lower with increasing altitude and increasing clay content in the well profile. This is particularly evident on the south side of the valley where Well 14 on the lower slopes of Sheepstor possesses the lowest hydraulic conductivity value measured. This well also has the greatest thickness of clay in its geological profile. Higher hydraulic conductivity and transmissivity values were expected at Site 1 than were actually measured, particularly since sandy materials are present in the profile at this location. Since repeat tests at this location provided consistent results, some of the lower clay formations at the base of the well screen, and drilling effects are considered to be responsible for impeding groundwater flow. Well 7 aquifer constants are markedly different than any of the others determined using slug tests in the Narrator Brook Catchment. The anomalous conditions at this site are further shown in groundwater chemistry and temperature variations experienced in this locality, and are further discussed in sections 8:1:7 and 8:1:8. The variation of aquifer constants with topographic location would seem to support the idea that the observation wells in the

- 411 Narrator Valley penetrate discrete water-bearing horizons. These horizons have differing lateral extents within the weathered granite« and hence varying contributory areas* From table 8:1 Wells, 12, 6 and 9 have higher hydraulic conductivity and transmissivity values than other wells. This Is again related to lithological conditions. At Veil 6 coarse pebbles and cobbles are present, while at Well 12, although it possesses thicknesses of clay around the screen section, evidence of underlying gravelly material Is apparent from the borehole logs. This will result in higher K values, which has proved to be the case (table 8:1). Pumping tests, (Chapter 6), failed to pump Well 9 dry indicating that this well may have one of the larger contributory areas above a clay lens, on the south side of the valley.

Table 8:1 Aquifer Hydraulic Properties

Topographic Sites T-Mean K-Mean Height (m^ d-^) (m d-^)

248.09 14 0.73 0.12 242.44 12 6.13 1.01 239.51 1 1.72 0.24 237.62 9 8.73 0.77 228.33 7 210.5 26.43 220.24 6 3.32 0.89

Since the Narrator Brook Valley aquifer has proved to be hydro- geologlcally complex, derived storage, transmissivity and hydraulic conductivity values, provide only a tentative guide to the water• bearing and yielding properties of the weathered granite materials. Slug test data plots, presented In Chapter 6, were used to derive aquifer constants. Information concerning the general conditions in an aquifer may also be inferred from these tests. Response times, derived from the slug test data plots indicated unconfined conditions In the valley aquifer. The response times accorded varied between 44-180 seconds which are very small in comparison to response times of 440-2830 minutes as reported by Black (1979) for the Carnmenellis granite, Cornwall. Observation well

- 412 - response times are so rapid that in the case of Wells 6, 7 and 14 they were not measurable. As well response effects are inversely propor• tional to the aquifer storativity, and are unlikely in unconfined aquifers (Black and Kipp 1977), then the Narrator Valley aquifer exhibits characteristic features of an unconfined aquifer. Anomalous behaviour of part of the aquifer in the Narrator Valley has been observed which Indicates the presence of semi-confined conditions in parts of the valley. At Site 11, a combination of lithological and rapid recharge conditions resulted in water under pressure shooting out of the breather hole at the well head. This implies that semi-confined conditions are operational in this area under extreme recharge events* 8:1:4 Water-Level Fluctuations Groundwater contour maps and change in depth to watertable maps are presented in Chapter 5. Wells in the observation network in the Narrator Brook Catchment penetrate aquifer horizons at different depths and locations within the valley, whose individual responses do not suggest a homogenous aquifer. Such an aquifer would be expected to respond in a similar manner to hydrometeorologlcal inputs, and fluctuations at any point in such a homogenous body would be expected to have similar characteristics. Both water-level fluctuations (Chapter 5) and variations in groundwater chemistry (Chapter 7) support the hypothesis of the presence of perched groundwater bodies within the weathered granite matrix in the Narrator Valley. In the Narrator Valley the well hydrographs respond to seasonal variations in recharge. The main features of contrast between the well hydrographs is in the magnitude of the different responses of the water-levels to rainfall input. Distinctive well groupings were defined on the basis of this response. This is further discussed in section 8:2. The change in water-level maps presented in Chapter 5, illustrate that wells on the higher valley slopes and around the reservoir, experience less variability In water-level than those in midslope positions. In the neighbourhood of the reservoir input and subsequent discharge downstream is more or less constant as water-levels do not vary greatly from 0-lm. Groundwater movement downslope from the higher areas of the catchment feeds the discrete flow systems at the

413 middle and base of the slope. The development of a saturated wedge upslope during recharge events may be responsible for the larger variations in the water-level change maps observed at these midslope locations. Water movement in the weathered granite parallel to the Narrator Brook and Burrator Reservoir, may be inferred from the trough in the water-level map (Chapter 5) where changes are in the order of 5m. The spring 12/13, near Site 12 reaches the surface in this trough area. The regions in the Narrator Valley of greatest water-level change are in the locality where observation wells penetrate clay lenses. These clay horizons are regarded as the chief Influences determining the route and direction of drainage waters through the weathered granite matrix in the Narrator Valley. The fact that groundwater levels do not vary greatly on the higher southern slopes of the catchment may be attributed to recharge and the more constant discharge conditions in this area. Over the rest of the catchment at lower topographic elevations, water-level fluctuations are between 0-3m annually. Correlations between water-level fluctuations in the Narrator Valley and the mean weekly fluctuation of the reservoir level produced significant negative correlations. This relationship is thought to be the result of the draining of water in the weathered granite aquifer (and hence the fall in water-levels), and the resultant rise in reservoir level as the saturated granite discharges downstream. This down valley groundwater discharge from weathered granite horizons may also be an important feature in similar valleys on Dartmoor as a whole, and is not likely to be restricted solely to the Narrator Brook Catchment. Large-scale and small-scale textural variations in the weathered granite matrix are considered to produce different contributory flow volumes to the watertable at a given site, dependent upon the relative permeabilities of these horizons. In the Narrator Valley contri• butions to groundwater in the vertical direction on slope locations are thought to be insignificant compared w-th horizontal or lateral groundwater movement. An exception to this general condition is provided by the potential for more rapid infiltration down the annulus of the observation wells, particularly if the well inter -sects a dominant but discrete contributory flowpath. No information however

414 is available on the significance of this process In the Narrator wells. 8:1:5 Surface and Groundwater Discharges It is evident from geological considerations and the deter• mination of aquifer properties in the Narrator Valley that the hydro- geological system is diverse in nature. This diversity is further illustrated in the analysis of streamflow data to determine baseflow, groundwater recession and flow duration characteristics from both Station 11 and Station Cutt (Chapter 4). In an average year (derived from data over the period 1978-1980) approximately 73% of the total recorded stream discharge in the river channel is derived from ground• water reserves. The length of the baseflow recession period, derived from four years data, can be greater than 200 days, further suggesting a substantial groundwater contribution from the weathered granite aquifer. Flow duration analysis also shows from the low flow indices, that baseflow discharge varies little Indicating a more or less constant source, or sources over any water year in the Narrator Brook Catchment. Baseflow represents 90% of the effective rainfall on the catchment leaving a residue of some 10% for deep groundwater recharge. The two sections of the catchment upstream from each of the gauging stations exhibited differing hydrogeological characteristics. These differences are amply demonstrated both by the groundwater recession and storage values listed in table 8:2, and by the variations in influent and effluent sections along the lower part of the river valley (Chapter 4). Over most of the groundwater obser• vation sub-catchment, effluent seepage augments streamflow. In the lower portions of the Narrator Brook however Influent conditions are present from Site E, down towards the reservoir. These influent conditions may be due partly to the greater transmissivity of the coarser valley infill in this area, particularly in the vicinity of Well 6, and to the draining of groundwaters from weathered materials in the catchment, towards the lower elevation of the water-level In the reservoir.

415 - Table 8:2 Hydrogeological Parameters Illustrating the Variations between Upstream and Downstream Sections of the Narrator Valley

Component Station Cutt Station 11

Mean Active 2838x10^ m^ d"^ 1014x10^ m^ d-^ Storage (St)

Average Groundwater 7568x10^ m^ yr"^ 3153x10^ m^ yr-^ Storage per year

Flow Duration Curve 0.10 - 0.14 0.25 - 0.43 - Variability Index

Groundwater 210 days 190 days Recession

The variability indices derived for the streamflows measured at Station Cutt and Station 11, particularly emphasize the hydrogeological differences between aquifer materials at these two locations. High values, 0.25-0.43 at Station 11^are explained as being due to saturated overland flow processes and rapid recharge via clitter slopes and mining spoil-heaps. At Station Cutt indices of 0.1-0.3 are lower, attributable to the larger storage capacity of the weathered granite in the lower stream reaches, which result in a greater uniformity of flow. The two component groundwater recession curves constructed at both Station Cutt and Station 11 again shown differing characteristics. Above Station Cutt 11% of active storage is released in the first 20 days of the recession, while 18% of available active storage is released in the first 20 days of the recession above Station 11. This further demonstrates that the storage capacity of the weathered granite aquifer is greater in the downstream regions and responds in a more uniform manner to recharge and recession than the isolated and widely distributed zones upstream of Station 11. The presence of influent conditions on this lower stretch of the Narrator Brook has some serious repercussions on the reliability of catchment discharge measurements determined at Station Cutt. As can be seen from table 4:S Chapter 4, on average over 30% of the flow recorded at Site E is lost to groundwater before lower flows at Station Cutt are recorded. This value does not take into account

- 416 - errors of measurement in stream discharge estimated by the current metering technique, or the potential discharge errors allocated at Station Cutt by use of a regression to complete a streamflow data set (Chapter 4). However, even with possible discharge measurement errors, it is apparent that a significant amount of streamflow Is lost to the weathered granite aquifer at this location, and as a consequence total catchment discharge as measured at Station Cutt is unrepresentative. 8:1:6 Occurrence and Distribution of Springs Springs and seepage points in a catchment represent important discharge areas (Freeze and Cherry 1979). In the Narrator Brook Catchment perennial springs tend to be located in the valley bottom, while intermittent and ephemeral issues are more commonly found on valley side positions (Ternan and Williams 1979). While this may be true in some parts of the catchment, the generalisation does not apply with regards to the groundwater observation sub-catchment as used in this investigation. The two perennial springs monitored in this study are located on the valley side, and ephemeral issues have been observed at both higher and lower elevations. Spring 12/13, utilises/old sub-surface mine working thus facilitating more rapid flowpath through the weathered granite. Lithological variations within weathered granite matrix result in discrete flows of water at different levels. This is particularly the case in the neighbourhood of clay lenses which give rise to perched watertables and in certain periods may contribute to spring or seep discharges. Spring 11 and Spring 12/13 may be supplied by groundwaters stored on deeper clay layers within the weathered granite, as both springs lose water immediately to the stream channel on emergence. The cooler more constant temperatures recorded from the spring waters support a deeper origin (Chapter 7) for these groundwaters, than those of most of the wells in the valley. The juxtaposition of materials with widely differing hydraulic properties may account for the pattern of spring distribution observed by Ternan and Williams (1979), who noted that springs appear to emerge in distinct areas in the valley, interspersed by regions of no or little spring development.

- 417 Groundwater in altered crystalline rocks tends to be much less uniformly distributed than those waters in sedimentary aquifers. The chances of artesian water occurring Is subsequently lower but conditions of an external character may occur tending to confine groundwater in fissure and joint systems in the granite. Dixey (1950) envisaged beds of clay overlying fractured and weathered granite giving rise to spring developments as Illustrated in figures 8:1a, 8:1b and 8:1c. With reference to figure 8;la, when the watertable is high in the winter months the water which percolates down through fractured granite escapes at point A above the rim of relatively more impermeable materials. When the watertable is low during the drier months water can only be reached by a well at point B. Alternatively, with reference to figure 8:1b water percolates downwards and laterally via joint systems in a granite body until it Issu es at point A. Water can also be held in the porous weathered granite material overlying the coherent granite and may escape as a spring or seepage area at point B caused by the presence of a clay lens or other less permeable textural horizons in the weathered granite matrix at this location. Dixey (1950) also considered the possibility of veins or intrusions in a crystalline mass being less susceptible to weathering than the granite matrix, and as a consequence giving rise to isolated perched watertables which issue as springs at the point of contact with harder materials, (figure 8:1c). Such a mechanism may account for the voluminous water flow observed by Sandeman (1901) during the excavation for Sheepstor dam» Figure 8:2a illustrates how a trench cut through a weathered granite and a vein system supporting perched watertables, may account for a variety of inflow horizons contributing water to the trench. Where the situation as illustrated in figure 8:1a exists, the potential for differential groundwater movement down valley through the varying lithologies, is readily apparent. In addition quantitltes of water may be transmitted vla^leakage from the river bed and by underflow processes. The varying hydraulic properties of the

418 FIGURE 8:\a Mechanisms for Spring Development in a Granite Area

When the water table Is high during a wet period, water escapes to the surface at A, above the rim of the impermeable beds. When the water table is lower in the dry season, water can only be reached by a well through the valley fill at a. f

Jointed and weathered I granite upland

Wafer Table (wet)

Water Table relatively impermeable (dry) clayey materials

~porous materials

•— Granite Bedrock F/GURB 8:]b Mechanisms for Spring Development in a Granite Area

Water percolates downwards and laterally via joint systems in a granite mass until it issues at Also water can be held in porous material overlying relatively impervious granite and escapes as a spring precipitation at s. I -r c I

jointed and weathered granite mass porous materials It/ FIGURE 8'\c Mechanisms for Spring Development in a Granite Area A vein or intrusion may impound the water entering relatively porous layers and allow It to escape at Intermittent springs A.B. C. weathered granite in the Narrator Valley, are demonstrated by the variable discharges and locations of the springs. Iron pan horizons and clay lenses interspersed with more permeable materials down a hillslope in the catchment may give rise to ephemeral springs and seeps. The disposition of such materials and flow paths are Illustrated in figure 8:2a. In extreme recharge events Spring 1 would be the first issue followed In time by Springs 2, 3 and 4. Flow at Spring 1 would cease soon after the recharge event, while Spring 4 will continue to flow for a few days until excess drainage waters ceased to reach this locality. This mechanisms could explain the ephemeral seeps observed on the south side of the Narrator Valley which formed in extreme recharge conditions experienced during this research period. Fracture flow from the granite bedrock may also be a source of sprlngflow in the Narrator Valley. In the farmyard of Deancombe Ruins, there is an overflowing source (possibly the well which provided domestic water to the farm, N.G.R.S.X. 5800 6875) which overflows only in the wetter months of the year. This issue is Inter• preted as being the result of a well penetrating a fracture-fed groundwater system which overflows during the wet season when recharge to the fissured granite above the farm is high. A large head is prod• uced resulting in semi-artesian flow conditions, for a few months until the interconnected fracture systems have drained away to lower elevations, (figure 8:2b). Such drainage of fissure systems may partly account for the two-component nature of the recession curve of the Narrator Brook (Chapter 4), although the effects of varying interflow contributions are important particularly in the upstream locations of the catchment. The groundwater temperature data in relation to depth (Chapter 7) shows strongly that the springs monitored are derived from sources deeper than most of the wells in the catchment. Wells 1 and 2 are the exception to this rule since they may penetrate the main but deeper groundwater zone in the valley. Figure 8:2b Illustrates the likely source of groundwaters from Spring 12/13. This spring issues from an old mine adit whose sub-surface extent is unknown, but temperature data suggests that groundwater is derived from depths of in excess of 17m. The slow drainage of water from fractured and fissured granites intercepted by mine workings may account for such water character• istics.

422 FIGURE 8:2a ORIGIN OF SPRINGS IN THE NARRATOR VALLEY

Ephemeral Springs formed during or after extreme recharge events as a result of the disposition of materials of varying permeability in the weathered granite profile I

IN I Trench intersecting water tables ponded Clay Lenses upon more resistant mineral veins in the weathered granite

c ^Weathpred Granite- Valley Infill

c

• Mineral intrusive ' ' an^ vfein* sysl^m^,

SOLID GRANITE BEDROCK FIGURE 8:2b ORIGIN OF SPRINGS IN THE NARRATOR VALLEY

Wei period Water Table t

Deancombe Well I and Spring

Drier period Water Table

Spring 12/13 issue from old mine workings 8:1:7 Geochemlcal Properties of Surface and Groundwaters

Geochemically derived patterns indicate the hydrogeological complexity of the weathered granite aquifer and overall support the hypothesised distribution of perched groundwater bodies, and nature of the groundwater movement. Chemical analysis of groundwaters in the catchment have Illustrated differences spatially and temporarily, particularly between the north and south sides of the valley. These variations are attributed to the elevation and hydrogeological complexity of individual sites, and to whether the wells penetrated perched groundwater bodies or the main groundwater zone. Individual well-waters, particularly on the south side of the valley are likely to be characteristic of isolated water-bearing horizons.

Large and small scale textural variations In the weathered granite matrix are considered to produce different contributory flow volumes to the watertable at a given site, dependant upon the different relative permeabilities of these horizons. Materials of differing hydraulic properties give rise to the production of the varying lag times observed in groundwaters transmitted to various well locations in the catchment. The correlation of chloride concen• trations in precipitation with those in groundwaters enabled an estimate of the time taken for groundwater recharge. This was found to vary between 5-8 weeks over the groundwater observation sub- catchment, (Chapter 7). Responses to recharge events were observed from watertable fluctuations to be from 1-4 weeks. This disparity in lag times is attributed partially to an increase in hydraulic pressure head in the downslope direction, which resulted in a rise in water- levels before the actual water input reached the observation wells.

In the Narrator Valley the action of translatory flow first pushing out *older' water down gradient may also account for the discrepancy in the lag times determined by using chloride as a tracer and those estimated by water-level responses. The presence of clay lenses and the subsequent ponding of Isolated water bodies above the main saturated zone, which spill over in extreme recharge events, will effectively modify the chloride concentrations determined in the weathered granite aquifer. In effect a mixing will take place between recently recharged and infiltrating waters, those waters held on the higher clay lenses, and that water held at lower positions in the

- 425 weathered granite matrix. The lag times calculated are a result of the pathways taken by the recharge waters. A consideration of soil moisture data collected In the Narrator Valley, In association with seasonal variations In water-levels and precipitation, suggests that recharge Is taking place to the groundwater system before any soil moisture deficit is replenished, (Chapter 5). The soil cover, like the underlying weathered granite materials possesses spatial variations in soil moisture status depending on site llthology.

Water-level data, chemical evidence and calculations of transit time in the aquifer using measured hydraulic properties, show that water movement in the weathered granite aquifer is rapid. Transit time is from 1 week to 3 months depending on the location of a site. Considerations of chemical equilibria in groundwaters from the Narrator Valley support the notion of rapid groundwater movement through the weathered granite aquifer. Groundwaters and surface waters from the catchment plot in the kaolinlte stability field, which is in agreement with observations made in waters from upstream locations of the Narrator Brook Catchment, (Williams et al. 1981). 8:1:8 Temperatures of Groundwaters

Well groups based on the variation of groundwater temperatures over the 18 month monitoring period (Chapter 7) correspond markedly with the depth of penetration of the water-bearing zone in the valley. This feature again illustrates the variation in hydrogeological responses in different parts of a weathered granite aquifer, and demonstrates the usefulness of variations in groundwater temperatures as a method of elucidating depths of groundwater origin in a complex system.

The variation of temperatures recorded down the profile in the observation wells supports other evidence collated from the catchment concerning the penetration of differing aquifer zones. The temperature variations with depth in the wells is related to ground• water flow at each location. Greater mixing takes place at Well 9, as demonstrated by the largest standard deviation of temperature with depth. This has been Interpreted as the result of a larger clay lens catchment zone which acts as a collector for spill-over events from offset clay lenses and clay lenses located at higher elevations.

Information from pump tests conducted at this site (Chapter 6)

A26 - suggested that the contributory area to this veil was larger than those at other locations, a5 the well was not pumped dry. Variations In temperature with depth down the profile at Wells 1 and 2 suggested that these have a more constant groundwater source, a factor which Is substantiated by the more constant chemical character• istics of these wells (table 8:3). Spring 11 and Spring 12/13 exhibited constant temperatures and constant chemical characteristics, also indicative of deeper sources. Well 1, Spring 12/13 and Well 11 are located in areas showing very different water-level variations (Chapter 5) but maintain constant chemical characteristics over the investigation period (table 8:3). This further illustrates the wide variation in the groundwater sources penetrated by the different observation wells. Hydrogeological groupings based on groundwater chemistry are summarised and further discussed in Section 8:2.

Table 8:3 Standard Deviations of Well and Spring Concentrations Measured in the Narrator Valley to Illustrate the Chemical Characteristics of Groundwaters from Different Localities within the Catchment.

Well Temperature Electrical PH Chloride Silica Sodium Potassium 'C Conductivity \iS cm"^ mg L"^

1 0.69 1.69 0.21 1.52 0.66 0.94 0.30

2 0.76 8.11 0.28 2.46 0.68 1.24 1.39

3 2.98 4.88 0.23 2.50 0.86 1.11 0.38

4 3.12 49.2 0.44 4.10 1.36 11.0 0.50

6 2.81 20.55 0.44 2.35 0.70 0.92 0.70

7 0.82 7.60 0.21 9.73 0.68 1.86 0.45

10 1.31 3.01 0.25 3.15 0.73 0.88 0.43

11 2.14 5.38 0.17 1.91 0.45 0.76 0.26 Spll 0.54 4.54 0.18 2.52 0.74 0.92 0.60 12 1.43 3.08 0.22 2.61 0.67 0.78 0.18 Spl2/13 0.71 3.39 0.21 2.36 0.64 0.64 0.79 13 1.22 11.63 0.25 3.25 0.60 0.95 0.44

14 0.93 34.72 0.30 4.40 0.87 12.48 0.55

15 0.99 14.01 0.23 4.85.. .0.60. 0.80. 1.73

- 427 - 8:2 Hydrogeolc^cal Provinces In the Narrator Valley

Llthologlcal evidence (Chapter 2) along vd.th the varying responses of groundwater levels (Chapter 3) and associated hydraulic parameters (Chapter 6), Illustrate the hydrogeologlcal complexity of the weathered granite aquifer In the Narrator Valley. Analysis of groundwater geochemistry (Chapter 7) support a complex recharge and discharge system, with groundwater flow taking place predominantly in the lateral plane with some vertical movement towards the watertable.

From the variety of techniques utilised in this study to assess the potential of the weathered granite aquifer in the Narrator Brook Catchment certain groupings of the observation wells are apparent. Such groupings are presented in table 8:4. The division of wells in the catchment into distinct groups was based on the following criteria:- 1. Lithological and fabric similarities (Chapter 2)

2. Visual assessment of well hydrographs (Chapter 5) 3. Annual range of waterlevels (Chapter 5) 4. Similar responses indicated by the water-level change map (Chapter 5)

5. Standard deviation of water-level fluctuations and the analysis of variance of such movements (Chapter 5) 6. Well groups based on the Stepwise Regression Analysis (Chapter 5)

7. Variations in the hydraulic properties of the weathered granitic materials in the Narrator Valley (Chapter 6) 8. Variations In groundwater temperatures (Chapter 7) 9. Geochemlcally derived patterns of solutes in the groundwaters and the visual assessment of monthly variation plots (Chapter 7)

10. Variation of temperature down the profile in the deeper

wells (Chapter 7) The magnitude of groundwater level responses can in the first instance be related to lithological controls and subsequent hydraulic properties of the saturated zones. The geochemical data provides additional information regarding spatial variations in groundwaters in the catchment, and indicates the relative speed of water movement in weathered materials. The development of temperature stratification down the well profiles and the variation in groundwater temperatures highlights the range of contributing groundwater flow paths, and the overall lateral variation between observation wells,

- 428 - Table 8:4 Summary of Hydrogeological Groupings of Wells in the Narrator Brook Catchment

Criteria Grouping Well Groups Chapter in which No Procedure Grouping is described

1 Lithology 3,4,6 Two 1.2 7 9,10,11,12,13,14,15

2 Visual 3,4,6,11 Five Assessment of 1,2,7,10,12,13 Well Hydrographs 9.1A,,15

3 Annual Range 3,4,6,10,11,13 Five 1,7,2 9,12,14,15

4 Water level 3.4,6,11 Five Change Map 9,14,15 1,2,7,10,12,13

5 Standard 1,2,7,10,12,13 Five Deviations of 9,14,15 Water-levels 3,4,6,11

6 Stepwise 1,10,13,14,15 Five Regression 2,7,6, Procedure 3,4,9,11,12

7 Hydraulic 6,9,12 Six Properties 7, 1,14

8 Temperature 1.2 Seven Variations 7,9,14,15 10,12,13 6,11 3,4

9 Geochemical 3,4,6,11 Seven Variations 1.2 7 10,12,13 9,14,15

10 Down the Profile 9 Seven Temperatures 10 7 1.2 12,14

429 It is evident from table 8:4 that some wells are common to all or most of the other groups determined from the ten grouping procedures.

Such wells which frequently appear together are as follows:-

Wells 1 and 2 Wells 3, 4, 6 and U Wells 9, 14 and 15 Wells 10, 12 and 13 Well 7

A noticeable feature of these groups of wells is their association by mean depth to the watertable, well depth and to a lesser extent elevation as summarised in table 8:5. For the most part wells within these groups tend to be in close proximity.

Table 8:5 Well Depth, Mean Depth to the Watertable and Well Elevations

Wells Mean Depth Well Depth Elevation to Watertable (m) (m) (m)

1 11.75 19.2 239.5 2 6.12 18.8 230.3

3 0.19 2.29 223.1 4 0.22 1.30 223.1 6 0.55 5.63 220.2 11 0.57 4.80 230.9

9 2.62 14.7 237.6 14 2.34 12.8 248.1 15 3.60 14.5 240.9

10 2.90 9.7 234.6 12 2.34 9.4 242.4 13 3.03 10.3 248.1

7 5.27 13.5 . 228.3

Wells 1 and 2 have similar features which result in them being grouped together by most of the well grouping procedures adopted. As discussed in Chapter 7, these are likely to be the only two wells in

430 the groundwater observation sub-catchment which penetrate a deeper saturated zone. The lack of marked temperature variation down the well profile and the constant geochemical characteristics recorded over the research period (Chapter 7) support this hypothesis.

Wells 3, 4, 6 and 11 are similar in their responses due mainly to their shallow depths and their nearness to the river channel, reservoir or spring. By virtue of their shallow depths these wells are not likely to penetrate the main groundwater zone, but intersect locally developed watertables.

Wells 9, 14 and 15 penetrate water-bearing clay lenses which are adjacent to each other, and this together with lithological controls may account for the overall similarities, particularly in their geo• chemlcal characteristics. From pump test data Well 9 appears to penetrate a larger groundwater contributory area than any of the other perched groundwater bodies, and the large scale temperature variations down the well profile supports variable recharge processes in this locality (Chapter 7). The neighbouring clay lenses of Wells 10, 12 and 13 may account for the gross similarities in their well responses. From temperature variations down the well profile Well 10 shows smaller standard deviations than Well 12, suggesting that groundwater flow at Site 10 may be of a more stable nature than that at Well 12. Because of damage to the casing at Well 13, it is difficult to delineate factors responsible for watertable responses. Well 7 has consistently produced responses which differ from any of the other wells in the catchment. Well 7 in particular possesses the highest hydraulic conductivity and transmissivity measured (Chapter 6). Indications from borehole logs (Appendix 1) suggests that the base of this borehole penetrated a more coherent granite material, and as a consequence groundwaters Intersected show different characteristics to those at other localities. Such different geochemical features may be the product of mixing of three sources of groundwater at this site; recharge through fractured and fissured granite; rapid infiltration of rainwaters down the well annulus and contributions from a leakage component from the base of the weathered granite materials. The wide variation in standard deviations of water temperatures down the well profile support the mixing of groundwaters from different sources at this location.

431 - The analysis of groundwater levels by a Stepwise Regression Procedure in the Narrator Valley produced three distinct hydro- geological groupings, which may be related to recharge and discharge zones in the catchment. Province I represents the recharge province (figure 8:3) and is characterised by Wells 1, 10, 13 14 and 15 which are at higher topographic locations within the catchment. Province IX is a discharge province characterised by Wells 3, 4, 9, 11 and 12 and for the most part of at lower topographic elevations within the catchment. Province III represents wells having characteristics which for the time duration analysed cannot be readily delineated as recharge or discharge zones.

In the case of Well 7, the anomalous response is attributed to the fact that it penetrates more coherent, possibly a fissured and fracture granite base underlying weathered materials in the valley. Well 6 penetrates some of the coarsest materials recorded in the aquifer (Chapter 2) being predominantly cobbles and pebbles. Well 2 in general does not show any marked anomalies in its responses, it does however penetrate the largest thicknesses of gravels (7-8 m) recorded in the catchment, which may account for its grouping into Province III. It is evident from table 8:4 that these three provinces compound features determining well groupings by the other methods, and on the whole it is probably unreasonable, in hydrogeological complex areas, to assign wells to rigid groupings. The fact that ten grouping methods produce similar divisions suggests that there are strong controls exerted by topography, elevation, well depth and llthology, on the variations experienced in the groundwater levels In the weathered granite aquifer.

The identification of hydrogeological provinces, typified by wells exhibiting a similar type of response to catchment antecedent conditions, are not likely to be stable features in a heterogenous aquifer. Antecedent catchment conditions will exhibit seasonal variations (particularly with respect to rainfall, evapotransplration and soil moisture) and as such seasonal contrasts are to be expected in the choice of wells and hydrogeological provinces. It is anticipated that in very wet periods a recharge province will consist of Wells 1, 2, 9, 13, 14, 15, a discharge province of Wells 3, 4, 6,

432 - recharge

|\ discharge ni.v.anomalous province

125 m

FIGURE 8:3 HYDROGEOLOGICAL PROVINCES IN THE NARRATOR BROOK CATCHMENT 10, 11, 12, while Well 7 will exhibit characteristically different responses and will hence form the third province. 8:3 Prediction of Groundwater Behaviour in the Narrator Brook Catchment

Data presented in Chapter 4 suggest that hydrogeological variations are apparent within the small sub-catchment used for groundwater level observations utilised in this study. It was anticipated that groundwater conditions at the downstream end of the catchment may be usefully projected to upstream locations, to provide a groundwater contour map for the whole catchment. Several features have emerged from this study to suggest that this would not be appropriate. Firstly, because of the existence of a variety of Isolated groundwater bodies in the Narrator Valley, the observed watertable fluctuations do not record the overall potentlometric surface in the catchment, but merely record the hydraulic head conditions of each isolated zone.

As a consequence of these conditions estimations of hydraulic gradients and the directions of potential groundwater flow may be misleading. Due to the variation in depths to water-bearing horizons in the valley aquifer, extrapolation of groundwater contours constructed is not recommended for the upstream end of the catchment. In upstream locations of the Narrator Brook Catchment the combination of the following additional factors will Influence the positions of potential groundwater surfaces:

(i) the porosity of weathered materials and spoil heaps would be greater, with less clay deposits likely; (11) infiltration would be more rapid through the generally

thinner regolith, and through the clitter slopes and spoil heaps;

(ill) sub-surface piping beneath granite blocks would transmit

water downhill much more rapidly towards the stream enhancing any 'flashy' characteristics of the streamflow; (iv) in some portions of the higher catchment, runoff will be greater due to bare granite tor surfaces, possibly giving rise to greater recharge via fissures to springs, and as a consequence the spring distribution varies;

434 - (v) the effect of the ironpan and the occurrence of soil water- tables is more widespread in the upstream reaches of the catchment; (vi) a greater proportion of old sub-surface adits are present further upstream which may supplement rapid groundwater discharge.

From this discussion it is evident that groundwater predictions in the higher catchment areas can only be conjectural. 8:3:1 Aquifer Response Times

The most useful parameter for describing the groundwater movement in the weathered granite aquifer of the Narrator Valley, was that of measured chloride variations in groundwater and precipitation. As a result of correlating these two measurements in the catchment, lag times, or the time taken for water to move from recharge area to individual wells, was estimated as being in the range of 5-8 weeks over the valley. The more rapid recharge response of the water-levels can be related to the development of a pressure head which is transmitted down valley. Similar features have been observed at Kernick Dam, St. Austell, Cornwall which has been built to store china clay residues. During flood tests water escaped through the jointed granite valley base beneath mica residues and Into the adjoining valley giving artesian back pressures in the valley gravels (Bristow, pers. comm. 1980). A lag time of 3 days occurred after the dam flood tests before the artesian pressures were recorded In the adjoining valley, but Tritium dating on the water gave a 20 day lag time before flooding waters were recorded at sampling sites in the adjoining valley.

In the Narrator Valley the action of translatory flow first pushing out an older 'slug' of water down gradient may also account for the discrepancy in lag times observed by chloride tracing methods. The chloride ion is assumed to have a subdued role in hydro- geological systems, and seems initially to be an ideal natural tracer for determining rates of recharge in weathered granites. However marked anisotrophy in the weathered granite matrix result in unpredictable recharge rates and the usefulness of the chloride tracer may be much reduced, particularly since clays may retard the passage

- 435 of chloride ions through them. In the case of the Narrator Brook Catchment there is little difference in mean chloride concentrations in groundwaters and those in rainfall, and on this basis the chloride ion is conservative, and remains in the groundwater where once in solution it is not removed.

8:3:2 The Use of Slug Tests in the Assessment of Groundwater Behaviour Through the use of slug tests it is possible to ascribe to a specific well local values of transmissivity (T) and storage (S) which will, according to Black and Klpp (1977), differ from the values for the aquifer as a whole. Papadopulus et al. (1973) suggested that a large number of such point (T) and (S) values are often of greater use than a single value of transmissivity obtained from a long-term pump test. The validity of this suggestion has been demonstrated by the use of slug tests in the Narrator Brook Catchment. Because of the variation in the water-bearing zones in the valley, results from a long duration pump test would only be applicable to a small contributory area in the locality of a single well. More numerous slug tests carried out at a variety of locations in the valley, representing a wider spectrum of hydrogeological materials and conditions present, have provided a more realistic range of results for the weathered granite aquifer. Because of the small diameters of the boreholes the slug test technique is the only suitable method employing well responses for estimation of aquifer constants in the Narrator Brook Catchment.

Slug-tests carried out on some 40m of weathered granite to the Caithness area Scotland gave hydraulic conductivity values from 0.02- 1.44 m d~^ as compared to 1.1-4.6 m d"^ determined for the Narrator

Valley. The similarities of these values suggets that slug-test techniques for measuring aquifer constants in unconsolidated materials are very valuable, particularly where conditions for long-duration pumping tests are not suitable.

The slug-test has proved to be a useful means of determining point transmissivity, and to a lesser extent storage and hydraulic conductivity values for the Narrator Valley infill. Due consideration however must be given to the modification on aquifer constants,

436 brought about by drilling and completion procedures, when attempting to interpret derived transmisslvity and storage values.

As the duration of the slug-test is very short the estimated transmisslvlty determined from the test will be representative only of the water-bearing material close to the well, and as such, values so determined are site specific. Vertical permeabilities of most stratified aquifers are only small fractions of their horizontal permeability therefore the induced flow within the small radius of the draw-down cone that develops during the short observation period is predominantly two-dimensional.

As a consequence of this two-dimensional flow, the determined transmissivity approximates the transmlssivity of that part of the aquifer in which the well is screened. Conditions in the near vicinity of the well have in variably been altered to some extent by the drilling and completion procedures, and consequently derived aquifer constants can be modified. Black (1978), noted that where holes were drilled using a mud- flush rotary method, slug-tests gave hydraulic conductivity values lower by more than a factor of ten, than the average value for the aquifer as a whole. This was taken to infer significant mud Invasion giving rise to a substantial 'skin effect'. The compressed alrflush used in the down-the-hole hammer technique as used in the Narrator Brook is not considered to have such a substantial effect, although some smearing of drilled surfaces in the borehole may have occurred as clay with water mixtures were blown to the ground surface. The hydraulic conductivity values determined for Well 1 (Chapter 6) were lower than expected for the amount of sands penetrated at this location. This may be due to a clay coating in the vicinity of the well screen brought about by the drilling or to a greater percentage change in the increasing clay content noted in the basal sand unit of this well (Appendix 1).

Use of the Bouwer and Rice (1976) method of analysis of slug test data has provided aquifer constants compatible with conclusions derived from the analysis of groundwater recession curves (Chapter 4). Parts of the weathered granite aquifer has been shown to possess a large storage capacity, with rapid through flow times. An estimate of

437 transit time, in the catchment, derived using Darcy's Law, gives a value of 3 months from Combshead Tor to Station Cutt. 8:3:3 The Use of Itegression Equations for Groundwater Prediction Purposes

In an attempt to isolate the important controlling independent variables on water-level fluctuations in the Narrator Valley, a stepwise regression procedure was utilised. The prime objective for using a stepwise regression procedure, was to determine the major controls of groundwater fluctuations in the catchment. As a consequence of a multi-aquifer system composed of perched watertables at differing elevations, predictive equations derived from the stepwise regression procedure were not thought to be valid for all cases in the Narrator Brook Catchment. Such results obtained compounded the opinion that in utilising a stepwise regression procedure for predictive purposes in the Narrator Valley, more site specific measurements of soil moisture, rainfall, permeability variations, incidence and duration of site flooding, variations in river stage height, topographic position and elevation factors should be incorporated into the model.

Tests of significance on the selected regression equations, and the subsequent analysis of residuals, indicated that the use of the stepwise regression model in the Narrator Brook was not completely successful. It does however fulfill its exploratory role in that it chooses variables having the strongest relationship with water-level fluctuations. These were soil moisture conditions, antecedent catch• ment conditions, rainfall and seasonal variations in catchment inputs. It also suggests, by virtue of the lack of significant regressions selected, that more catchment and site parameters are required to successfully account for the behaviour of groundwater levels In the Narrator Brook Catchment.

8:4 The Potential of the Narrator Brook Aquifer

Since the Narrator Brook valley aquifer has proved to be hydrogeologically complex, derived storage, transmisslvity and hydraulic conductivity values, form only a tentative guide to the water-bearing and yielding properties of the weathered granite materials. The aquifer contants determined In the Narrator Valley are summarised In Table 8:6 along with estimates of mean saturated

- 438 - thicknesses at each location and the annual watertable elevation range. Water-level ranges and variations in saturated depths again illustrate the hydrogeologicaly diversity of the weathered granite aquifer.

Table 8:6 Mean Aquifer Properties, Waterlevel Ranges and Saturated Thicknesses in the Narrator Brook Catchment

Mean Aquifer Constants Annual Mean Water Saturated K S T Level Thickness Well (m d-1) (n?d-l) Range (m)

1 0.24 10-** 1.73 2.88 7.74 2 - - - 2.62 13.3 3 - - - 0.34 2.0 4 - • - - 0.26 1.06 6 0.89 10-3 3.32 0.75 4.99 . 7 26.43 - 210.47 2.87 8.47 9 0.77 lo-** 8.73 4.04 11.91 10 - - - 1.27 5.6 11 - - 0.91 4.0 12 1.01 10-- ^ 6.13 3.02 7.24 13 - - • - 1.89 5.06 14 0.12 10-^ 0.73 3.80 7.31 15 — — — 5.37 10.09

8:4:1 Groundwater Resources

Groundwater supplies from the main saturated zone in the Narrator Valley, assumed to be at depths greater than 15m in the lower part of the valley, may be a more promising groundwater resource. It is suggested that data obtained from the shallow observation wells used in this investigation, are not adequate enough to describe or attempt to quantify deeper sources in the Narrator Valley. These data do however suggest the potential for groundwater storage in the weathered

granite valley aquifer.

Analysis of discharge characteristics (Chapter 4) using flow duration curves and baseflow recession techniques for the two discharge stations in the catchment have again Illustrated the presence of hydraullcally diverse materials in the valley aquifer. Additionally techniques utilising discharge measurements have also illustrated the importance, magnitude and stability of groundwater

439 - resources in the Narrator Valley. Table 8:7 summarises the main components which highlight the resource potential of the Narrator

Brook aquifer.

Table 8:7 Groundwater Parameters Indicative of Resource Potential in the Narrator Brook

1 Mean stream discharge 64.41 m^ sec"

1 Mean groundwater discharge 50.86 m^ sec" contributing to Streamflow

Assumed mean depth of weathered 15m materials

Assumed mean saturated thickness 12m

Potential storage capacity of aquifer 570 X 10^ m^

Volume of recharge available to deep 557 X 10^ year"^ groundwater resources

Average groundwater component as a 73% percentage of streamflow

. Mean_groundwater.recession.period_ 200.davs

The Narrator Valley possesses a large volume of weathered granite with moderate to high transmissivity values, 5.42 - 36.965 m^ d"^ (Chapter 6). The variety of materials in the Narrator aquifer with different hydraulic properties makes resource assessment difficult. A gross catchment hydraulic conductivity value was derived from Darcy's Law, giving an average value of 3.6 x 10"^ m^ sec"^. It is emphasised that this calculation is merely an attempt to quantify the gross hydraulic characteristics of the weathered and fractured materials in the catchment, as though they were transmitting groundwater as diffuse flow. The real physical situation is most unlikely to be one of homogeneous diffuse flow because of the Joint and fissure systems in the coherent bedrock in the higher reaches of the catchment. Taking a 2km flow path downstream from Combshead Tor to Station Cutt, a transit time for water movement was estimated in the region of 3 months. Again this is only a tentative suggestion since several physical features in the catchment may profoundly alter the estimate. Areas of

440 coarse gravelly deposits In the weathered granite matrix along with mining spoil heaps, will have much higher porosities which will reduce the flow velocities, while the granite bedrock will have negligible values. In addition water-levels recorded at Individual sites in the valley are the result of a specific well intersecting a different groundwater zone, giving a variety of head measurements. As such the hydraulic gradient determinations in the Narrator Valley only approximate gross average conditions over the catchment area. Calculations, based on the average consumption of water of some 310 litres per person per day (Rodda et_ al^. 1976), would suggest that a population of some 1000 people could be supplied continuously for nearly five years from the estimated deep groundwater resources (Table 8:7) from the Narrator Brook Valley aquifer. A mean stream surplus of 3.92 X 10® m^ yr"^ is available which represents that volume of water which could be abstracted without depletion of storage or baseflow contribution to the natural discharge of the Narrator Brook as measured at Station Cutt. Both the above potential resources are theoretical considerations only in the Narrator Valley, since the Narrator Brook discharges into Burrator Reservoir, which supplies domestic and industrial water to Plymouth and its environs. As such the development and potential abstraction from the weathered granite valley aquifer at this locality is curtailed, since both surface and groundwaters eventually find their way to the reservoir basin and hence furnish public water supplies with additional resources. The resource potential of the weathered granite aquifer in the Narrator Valley will have relevance to other weathered granite valleys on Dartmoor, and on the development of crystalline areas in general for public water supplies. 8:4:2 Groundwater Quality A major consideration in the exploitation of a groundwater resource is the quality of the water in terms of its potability for public consumption. The chemical analysis carried out on waters from the groundwater observation network In the Narrator Valley, can only be used as a preliminary guide indicating the suitability of the groundwaters from the weathered granite aquifer, for drinking water supplies. Solute concentrations from groundwaters and surface waters

441 in the Narrator Valley were shown to be well within European Economic Community specified limits for good potable drinking suplles. Iron concentrations In waters from the catchment were not Investigated in the course of this research project, but are known from recordings in Burrator Reservoir to be above 0.1 mg 1~^ (Taylor, pers. comm. 1978). Groundwaters feeding the bogs higher upstream in the catchment become oxygenated upon reaching the surface, and dissolved iron in the groundwater is oxidised to the ferric state. As this form Is insoluble it precipitates and combines with humlc colloids in the bogs. The presence of such colloidal suspension imparts a distinctive reddish brown colouration to the stream, which increases during stormflow as much greater quantities are mobilised. As a consequence the whole stream becomes discoloured and concentrations of iron in the reservoir often exceed the maximum permissible value of Img 1~^ recommended by the World Health Organisation for drinking water. In order to combat this the waters of Burrator Reservoir have, since 1959, been treated with alum to remove Iron in colloidal suspension. 8:4:3 General Assessment of Weathered Granite Resources The variable nature of the physical propetles, and the thickness of valley infill, and the lateral extent of clay lenses all contribute to anlsotrophy in the weathered granite aquifer of the Narrator Valley. Such variations In the thicknesses and disposition of materials In weathered granite aquifers, as Illustrated in the case of the Narrator Brook Valley, are typical of valleys on granite areas in South West England. This can readily be seen from the cross-profiles constructed of several weathered granite valleys in South West England as described In Chapter 2. As such, all weathered granite deposits show hydrogeological complexity to varying degrees, and aquifer constants are likely to vary widely, and those values determined are site specific values only. The similarities between the nature and disposition of weathered granite materials in both the Taw Marsh area (N.E. Dartmoor) and the Narrator Valley are striking. Both locations are characterised by clays, sands and gravels overlying a more coherent granite base, and exhibiting lateral inconsistencies to the extent that adjacent lithologles cannot be readily correlated. As a result all such

- 442 weathered granite zones will exhibit hydrogeological complexity to varying degrees, and aquifer constants will vary widely. Consequently hydraulic conditions evaluated at a particular location are specific to that location alone, and attempts to project such parameters and thereby predict overall aquifer responses are subject to gross error. It seems reasonable to propose that in weathered granite zones on Dartmoor, whether they be valley aquifers like the Narrator Brook Catchment, or basin areas such as Taw Marsh, the groundwater distribution is likely to be determined by site specific properties, as hydraulic gradients depend on local topography as well as geological location. A direct result of this hydrogeological complexity Is that groundwater resource assessment in weathered granite aquifers is beset by three main problems: (i) the determination of the geometry of the weathered granite aquifer; this includes its depth, saturated thickness, lithological variations and hydraulic properties (Chapters 1, 2. 6) (ii) the delineation of recharge areas, including subsurface sources, and fissure-drainage contributions (Chapters 5 and 7) (ill) the determination of the magnitude and significance of losses through leakage, and to estimate underflow volimes (Chapter 4). The resource evaluation for weathered granite areas is more akin to those of alluvial aquifers as outlined by Jeffcoate (1979). The standard technique for estimating recharge volumes infiltrating over a specified areal extent of exposed water-bearing and transmitting lithologies, is not valid in the case of alluvial aquifers or weathered granite areas, because of the diversity of lithologies both in 'outcrop' areas and within the aquifer matrix Itself,

In areas of a granite catchment with a widespread development of ironpan sequences vertical infiltration will be minimal, and lateral flow may contribute directly to the stream channel and overland flow processes. In the localised spoil-heap regions in the Narrator Valley surface runoff is eliminated as these coarser deposits facilitate rapid infiltration, as will sand and gravel lithologies within the weathered granite matrix. Clay lenses and other laterally

- 443 ImpersistenC textural horizons will impede the downward movement of infiltrating waters, resulting In perched groundwater bodies whose volume is difficult to ascertain. As a recharge event continues, clay lenses at certain locations within the weathered materials will spill-over, resulting in either recharge to the main underlying saturated zone, or contributing to the groundwater resources on another isolated clay lens.

Resource assessment in weathered granite aquifers on a general basis is evidently a difficult task. In addition replenishment of the main saturated zone can be afforded through leakage from the stream channel to the aquifer, and contributions from fissure and fracture systems in the coherent granite. It has been shown in the case of the Narrator Valley that certain parts of the stream channel are in hydraulic continuity with the aquifer, and these conditions vary widely within one catchment area. Underflow losses constituting volumes equivalent to some 5% of the discharge in the stream channel at any point in time in the Narrator Brook, may also vary within a catchment, further complicating an already complex hydrogeological assessment. 8:5 The General Form and Properties of Weathered Granite Aquifers It is evident from the material collated from site Investigation and dam excavations, that the occurrence distribution and depth of weathered granite and its potential for water storage, albeit in isolated aquifer units, is not insignificant on the granite areas in South West England. The case of the Narrator Valley weathered granite aquifer, as Investigated here, illustrates some characteristic features of such Isolated water-bearing materials. 8:5:1 The Extent of Weathered Granite Aquifers

The regional boundaries of such aquifers are easily defined, since the uniformity of the underlying geological structure of granite Intrusions like Dartmoor, suggest, that topographic and groundwater divides are near coincident. Most potential weathered granite aquifers are present in valley situations, e.g. Burrator, Fernworthy, Cowsic and Huccaby. The depth to solid, bedrock in all cases reviewed in South West England, as part of this research project, is not uniform and varies from depths of 0.6m to greater than 47m, In some cases this results

444 in greater thicknesses of weathered materials being confined to one location in the valley; as in the case of Hart Tor where greater thicknesses are found on the south east side, on the east side at Cowslc; while at Swincombe, Huccaby, Tawmarsh and Fernworthy on Dartmoor, the greater thicknesses are apparent in the central area of the valley. In some Instances, on the granites of South West England, rotted granite is dispersed in thicker units beneath the valley centres beause erosion has worked preferentially along weaker features. Major decomposition Is associated with mineralisation of the granites in the south west, as is the case in Fernworthy, Burrator, Sheepstor and Taw Marsh. On the whole the thicknesses of unconsolidated materials thin towards the valley sides with the interfluve areas being composed of solid granite, in all the cases examined. This feature Is hydrogeologically significant. If the groundwater is to be stored or transmitted through the weathered granite alone, and assuming no drainage fissures are present in the coherent granite base of a valley aquifer, then such catchments may be considered to be watertight. As a result of fissures, fractures, mineralisation, hydrothermal processes and associated weathering, together with periglacial and alluvial deposits, the thicknesses and lateral extent of such materials overlying a granite base differ markedly. The great variety of materials and variations in thicknesses and lateral extent are illustrated from the borehole logs available for Taw Marsh and the Narrator Valley. As a result, the hydraulic properties of a weathered granite aquifer vary from the very permeable sands and gravels to the relatively Impermeable clays and kaolinised zones. Large quantities of groundwater appear to be transported via fissure and joint systems in the granite in some of the weathered valleys subjected to site investigations. In view of this it would be useful to have access to a detailed survey of joint and fissure trends in granite areas, particularly if attempts are to be made to assess the water resource potential- On Dartmoor there Is considerable local accordance between valley and joint directions, as in parts of the Cowsic, Meavy, Webburn, Taw, Dart and Plym valleys (Brammall 1926). Worth (1930) however suggested that there is an absence of dominant regional trends in jointing on Dartmoor.

445 - Junctions of the weathered rocks with the coherent granite basement are uneven and at various depths depending on the location and Incidence of fissuring and mineralisation processes. At Fernworthy, Burrator and Sheepstor decomposed materials were recorded in the borehole logs as becoming more resistant with depth, and merging into hard rock, in some cases abruptly. Joints were observed to close up with depth at most locations. 8:5:2 Groundwater Movement Hydraulic gradients in any aquifer are caused by variations in the hydraulic potential or 'head* from one point to another. The driving force behind all movement of groundwater Is the potential energy the water possesses by virtue of its elevation above sea level. In weathered granite materials in particular the hydraulic gradient varies from site to site. Porosity and hydraulic conductivity are likely to be within a limited range for a particular type of material in a specific geological location, but the hydraulic gradient, which determines groundwater movement depends on local topography as well as geological location. Calculations of formation constants based on pump tests or slug tests give widely varying results due to the heterogeneous nature of the aquifer, and the results can be taken only as a general guide to conditions. The groundwater recharge to a weathered granite aquifer may be from three sources: (1) replenishment by infiltrating precipitation water onto permeable areas of exposed aquifer material (11) leakage from the river bed (ill) fissure and fracture contributions. Although there may be wide variations in groundwater flow, in direction as well as quantity, the flow pattern will tend towards stability in dry periods relying on baseflow sources with contributions from underflow. This underflow may be compared with the more conventional baseflow component of streamflow in a valley. Unlike baseflow of a river which is normally continuous and Increasing down valley, wide variations may occur in underflow down a weathered granite valley due to differences in aquifer geometry and hydraulic properties. This results in variable Influent and effluent sections along the stream bed, as Illustrated in the case of the Narrator

- 446 Valley. The extent of hydraulic continuity was determined for a study stretch of the Narrator Brook Channel, but it is felt that such conditions will differ greatly within the Narrator valley Itself, although the method of determination will be applicable for use in other catchments. Mining activities may enhance the sub-surface flow pathways in weathered and fractured granite. Old adits and drainage channels may be so utilised. There Is little detailed documented evidence available as to the extent of these activities on parts of Dartmoor, and In the Narrator Valley in particular. As a consequence of this, and of the variety of materials of different hydraulic properties present, estimates of gross catchment parameters such as hydraulic conductivity and porosity must be viewed with caution. 8:5:3 A General Model for Groundwater Conditions In Weathered Granite Aquifers The Narrator Valley aquifer has been shown to be a hydrogeologlcally complex system, and as such is not an ideal situation from which to derive a general groundwater flow model applicable to other weathered granite areas. The disposition and thickness of weathered materials on the granites In South West England are likely to be unique for each location. As such, individual sites will exhibit hydrogeological complexity to varying degrees as Illustrated in Figure 8:i^. As a consequence potential resource assessment by more conventional means is Inappropriate, particularly as Darcian flow may In some zones be superseded by turbulent flow conditions. Todd (1959) noted that non-Darcy flow conditions may be found in rock aquifers, in unconsolidated aquifers with steep hydraulic gradients, or In those aquifer materials containing large diameter openings. Confident predictions, concerning the potential of water resources elsewhere in other weathered granite terrains, are therefore unlikely. It is however apparent from the investigations in the Narrator Valley, that groundwater from storage in weathered granite materials contributes significantly to catchment runoff, and could adequately furnish small isolated supplies for short-term periods.

- 447 FIGURE 8:4 CROSS-VALLEY PROFILE OF A WEATHERED GRANITE AQUIFER ILLUSTRATING THE HYDROGEOLOGICAL COMPLEXITY Shallow depth to bedrock; OF SPECIFIC SITES Well in fractured granite; well-drained solli Groundwater Chemisry diMeis from that in the weathered granite matrix; Cooler water tempoiatures Water Level fluctuation range small.

Well intersects localised perched Deeper wells intersect water tables on clay lenses ; main groundwater zone; Materials have varying hydraulic Greater hydraulic properties; properties well-drained soils; Great variabMity in Water Level Groundwater Chemistry fluctuations reflects stability of Greater depths to bedrock source. Groundwater chemistry indicates mixing of source. Shallow wells intersect high water table Higher soil moisture; effluent slow infiltration; stream l|lMe variation in water levels; channel Groundwater Chemistry reflects surface flooding. ^

GRANITE BEDROCK 8:6 Recommendations and Proposals for Further Research 8:6:1 Aquifer Thickness It is felt that this investigation has fulfilled most of the initial aims as outlined in Chapter 1. One exception is the failure to determine definitive depths to solid bedrock in the Narrator Brook Catchment. The aquifer thickness was not obtained due to the existence of greater depths of unconsolidated material in the valley, and as a consequence the equipment limitations were exceeded. However it is considered that further geoselsmic investigations in the Narrator Valley would be worthwhile, since the vertical and lateral extent of the weathered granite aquifer need to be determined in order to refine the Initial tentative storage estimates derived during this investigation. 8:6:2 Extent of Joint and Fissure Systems It is apparent from the weathered granite valley regions on Dartmoor, which have been sites of dam construction or trial investigations, that joint and fissure systems are important ground• water conduits. It would be useful to have access to documented sources, indicating the main joint and fissure system trends on Dartmoor and other granite intrusions in the South West England. A survey of the joint and fissure systems present in the Narrator Brook experimental catchment area would contribute greatly to the data-base available In this valley, and may throw light upon the origin of some of the springs* 8:6:3 Spring Development The origin of springs In weathered granite areas, relevant to conditions in the Narrator Brook Catchment are attributed to fissure flow in coherent granites., drainage by sub-aerial mine workings, intrusive vein systems, and ephemeral seepages due to the Justapostlon of materials of differing permeabilites. A detailed survey of spring and seepage sites may provide substantial corroborative evidence to support these hypotheses. Such information may be useful and applicable in weathered granite aquifers elsewhere. A tracing technique for use in some of the deeper seated springs may be productive. A spring which developed at the foot of Sheepstor Dam, was initially thought to be the result of a leak from the reservoir, but a chemical analysis showed it to possess distinctly

- 449 different characteristics (Taylor, pers. comm. 1978). If spring waters sampled over a granite area possess different chemical characteristics then it may be possible to determine whether they are from localised shallow origin in weathered materials, from deeper flow in the fractured granite bedrock, or supplied by fissure systems over a regional scale. 8:6:4 Delineation of Groundwater Flow Regimes Granite regions, by virtue of their negligible permeabilities are not normally considered to possess regional groundwater systems in the same sense as those of sedimentary aquifers. However, if deep seated fissure systems are developed within an area then groundwater in the fissures may originate at some distance from its point of emergence. Localised groundwater systems developing in weathered granite valleys are likely to be more readily accessible for monitoring, than any deep fissure flow system. Despite difficulties in assessing weathered granite aquifer potential, due to the variety of materials and differing hydraulic characteristics present, they appear to be the most feasible proposition for the development of groundwater resources in granite areas.

Unless a fissure system terminates at the base of a weathered granite valley aquifer, regional groundwater flow systems, if present in a granite mass, are unlikely to modify the locally developed groundwater systems. This is not always the case in sedimentary aquifers particularly where steep hydraulic gradients are induced by pumping and as a consequence the regional pressure distribution may be modified. If a small isolated weathered granite aquifer were to be periodically 'mined' of its resources, then by virtue of Its water- tightness and the negligible permeability of the granite basement, no deleterious effects would be felt elsewhere. 8:6:5 Mlcrostudies

In the particular case of the Narrator Brook Valley aquifer, it is felt that because of the variety of materials present and the differences in hydraulic properies, it might be appropriate to adopt some more rigorous site specific studies. Such microstudles should attempt to monitor the individual responses of difference aquifer materials to recharge events. Results from such studies may produce more helpful parameters for use in the stepwise regression model, and

450 thereby aid in the prediction of groundwater behaviour in weathered granite aquifers. Because of the physical limitations of the observation wells, and the restricted groundwater horizons penetrated, conventional pump tests were not successful in the catchment. They did however produce some direct evidence concerning the storage effects of the formation stabiliser at each location. Short rapid rises in the early part of the recovery curves in the Narrator Valley were Interpreted as the result of the draining of the well backfill/formation stabiliser. If the storage effect of the formation stabiliser is large in proportion to the slug of water withdrawn during a slug test, then aquifer constants, so derived, are likely to be more representative of the hydraulic properties of the formation stabiliser rather than the aquifer materials. In the published literature available on slug testing no reference is made to the volumes of slugs used in each case, or the appropriateness of a known volume to specific site conditions. In the narrow wells in the Narrator Valley a larger slug volume may produce well responses more characteristic of the aquifer material. If the volume of the slug used is large enough it may Incorporate storage effects from the formation stabiliser, and hence produce a more characteristic well response attributable to the aquifer materials. In the Narrator Brook Catchment two slug sizes were utilised at each site tested in an attempt to verify any changes of aquifer constants. The well responses at each site using two different volume slugs (150ml and 50ml capacity) showed little variation. This is considered to be due to the fact that the individual slug volumes were not large enough, and that the capacity change between the two slugs were not of a large enough magnitude. Due to the physical constraints of the observation well sizes the use of much larger volume slugs was impractical, and time considerations curtailed further equipment developments required to evaluate this approach. It is felt however, that variations of aquifer constant determinations in relationship to slug volume utilised, merits further investigations particularly in areas of unconsolidated aquifer materials, such as weathered granite regions.

451 8:6:6 Future Boreholes The majority of the observation wells in the Narrator Valley only provide information concerning the perched zones in the weathered granite matrix. It is recommended that should further investigations into the occurrence of groundwater in this aquifer be undertaken, then provision should be made for the sinking of at least two deep (>20m) cased and screened wells. This would produce the opportunity to conduct a long duration pump test with the facility of a second borehole for observation purposes. It is suggested that these should be located within the groundwater observation network already instigated for this research investigation. Any future drilling should result in the collection of a definitive log of materials penetrated, and samples from various depths in the aquifer for particle size analysis would be advan• tageous. An alternative drilling technique is required in order to penetrate the variable thicknesses of different materials In the valley infill. A percussion technique would be suitable, a method by which well casing follows the drill tool, thereby enabling the efficient removal of the bit and drill string from the hole. The emplacement of the casing as drilling progresses will provide good access in badly caving materials. Since hydrogeology is a dynamic and Inexact science, the accuracy and reliability of acquired data will usually increase with a longer duration of observation period. The interpretations presented here are based on 18 months of water-level and geochemical data collected In the Narrator Brook Catchment. As such any results can be taken only as a general guide to the hydrogeological conditions. Features observed do however indicate the relative speed of response of weathered granite aquifers to recharge events, and their resource potential.

452 BIBLIOGRAPHY

Acquaye, D., and Tlnsley. J. (1965) Soluble silica in soils. In: Experimental pedology, pub. Butterworth, London, p.p.126-148.

Agerstrand, T. (1966) Groundwater Draft from Earth Layers, in Groundwater Problems, eds. Eriksson, Gustafsson and Nllson, Vol.11 Pergamon Press.

Ahrens, T. (1970) Basic considerations of Well Design. Water Well Journ. 24 (4,5,6,8), Parts I-IV.

Alexander, J. (1977) The Conjunctive Use of Groundwater Resources and Stream Augmentation. MSc Dissertation (Unpub.) Newcastle University.

Alexander, J-, Hall, D.H., and Storey B.C. (1981) Porosity measurements of crystalline rocks by laboratory and geophysical methods. Repv Inst .""GeOl~S'c'i., ENPU 81-10.

Aldrick, R.J. (1978) The Hydrology of the Magnesium Limestone In Yorkshire between the river Wharfe and the river Aire. Q.J. Eng. Geol. London 11(2), 193-201.

Al-Saleh, S., Fuge, R., and Rea, W.J. (1977) The Geochemistry of some Blotite from the Dartmoor Granite. Proc. Ussher Soc. 4(1), 37-47.

Anderson, M.G., and Burt, T.P. (1980) Interpretation of Recession Flow. J. Hydrol. 46, 89-101.

1 - Anon. (1974) The Kernick Dam. Mining Mag. Sept. 1974, 171-175.

Ashley, R.P., and Lloyd, J.W. (1978) An example of the use of Factor Analysis and Cluster Analysis In Groundwater Chemistry Interpretation. J. Hydrol. 39, 355-364.

Atkinson, T.C., and Smith, D.I. (1974) Rapid Groundwater Flow in Fissures in the Chalk: An example from South Hampshire. Q.J. Eng. Geol. London 7, 197-205.

Bachelor, A.S., and Pearson, CM. (1979) Preliminary Studies of dry rock geothermal exploitation in South West England. Tech. Note Trans/section B The Instit. Mln. Metal, 88, B51-B56.

Back, W., and Hanshaw, B.B. (1970) Comparison of Chemical Hydrogeology of the Carbonate Peninsulas of Florida and Yucaton. J. Hydrol., 10, 330-368.

Barnes, B.S. (1939) The' Structure of Discharge-Recession Curves. Trans. Am. Geophys. Union. 20, 721-725.

Barnes, I. (1964) Field Measurement of Alkalinity and pH. U.S.G.S. Wat. Sup. Pap.. 1535-H.

Bell, J.P., and Wellings, S.R. (1980) The physical controls of water movement and storage in the unsaturated zone, a contribution to the Hydrogeological group meeting 8th Jan. 1980, London. Bell, E.G. (1978) Petrographical Factors relating to porosity and permeability in the Fell Sandstone. Q.J. Eng. Geol London, 11(2), 111-126.

Betson, R.P., and Marius, J.B. (1969) Source areas of storm runoff. Wat. Res. Res.. 5, 574-582.

Binnie and Partners (1971) Report on the Water Resources of St. Marys, Westminster, London.

Birtles, A.B. (1978) Identification and Separation of Major Baseflow Components from a Stream Hydrograph. Wat. Res. Res.. 14(5), 791-803.

Black, J.H., and Kipp, K.L. (1977) Observation well response time and its effect upon aquifer test results. J. Hydrol., 34, 297-306.

Black, J.H. (1978) The use of the Slug Test in groundwater investigations. Water Service. 82(985), 174-178.

Black, J.H. (1979) Results of a multiple borehole pumping test in low permeability granite, in Proc. Organiz. Econ. Co-op. Develop. Measurement of low permeability, low flow media, Paris 1979, p.p. 183-195.

Boelter, D.H. (1964) Water storage characteristics of several peats in situ. Proc. Soil Sci. Soc. Am. 28, 433-435.

3 - Boelter, D.H. (1965) Hydraulic conductivity of peats. Soil Sci.. 100(4), 227-231.

Bogle, J.M.L., Ross, R.M., and McMillan, T. (1959) The design and construction of the Avon Dam. Proc. Inst. Civ. Engrs.. 12(1), 83-106.

Boast, C.W., and Klrkham, D. (1971) Augerhole Seepage Theory. Soil Sci. Soc. Amer. Proc. 35(3), 365-374.

Booker, F., and Mildren, J. (1977) The Drought in the South West. Dept. of Economic History, University of Exeter Industrial Archaeology Group. Pub. Penwell Ltd. , Cornwall.

Bonell. M. (1971) An investigation of the movement and fluctuation of groundwater in a small glacial drift catchment in Holderness, East Yorkshire. Unpublished Ph.D. Thesis, University of Hull.

Bonell, M. (1972) An assessment of possible factors contributing to well level fluctuations in Holderness boulder clay. East Yorkshire. J. Hydrol., 16, 361-368.

Bonell, M. (1978) An evaluation of shallow groundwater movements in a small boulder clay catchment in Holderness. University of Hull, Dept. of Geog. Miscellaneous Series, No.18.

Bott, M.H.P., Day, A.A., and Masson-Smlth, D. (1958) The geological interpretation of gravity and magnetic surveys in Devon and Cornwall. Phil. Trans. Roy. Soc. (A), 25, 161-191. Boulton, N,S. (1963) Analysis of data from non-equlllbrlum pumping tests allowing for

delayed yield from 8torage« Proc.' Instn. Civ. Engrs., 26, 469-482.

Boulton, N-S. (1970) Analysis of data from pumping tests In unconflned anisotropic

aquifers. J, Hydrol., 10, 369-378.

Bouwer, H., and Rice, R.C. (1976) A slug test for determining hydraulic conductivity of unconflned aquifers with completely or partially penetrating wells. Wat. Res. Res., 12(3), 423-428.

Bouwer, H. (1978)

Groundwater Hydrology. Pub. McGraw-Hill Book Company 1978.

Bow, C.J., Howell, F.T., and Thompson, P.J. (1970) Permeability and porosity of unflssured samples of Bunter and

Keuper Sandstones of South Lancashire and North Cheshire.

Wat. Wat. Eng., Nov. 1970, 464-466.

Brammall, A., and Harwood, H.F. (1923) The Dartmoor Granite, Its mineralogy structure and petrology.

Mln. Mag., 20. 39-53.

Brammall, A. (1926) The Dartmoor Granite.

Proc. Geor.~Assoc., 37, 251-277.

Brammall, A., and Harwood, H.F. (1932) The Dartmoor Granites; Their genetic relationships, Q."J~Geol. Soc, London, 88, 171-237.

- 5 - Bristow, C,M. (1977)

A review of the evidence for the origin of Kaolin deposits In South

West England.

Proc* 8th Int. Kaolin Symp. and Meeting on Alunlte, Madrid-Rome,

K-2, 1-19.

Broadhead, J.A., and Mackey, P.G, (1972) Use of Trent Alluvial Gravels, In Symp. on advanced techniques In River Basin management: The Trent Research Programme. Inst. Wat. Engrs., London, p.p. 103-131.

Brown, R.H. (1953) Selected procedure for analysing aquifer test data. Journ. Amer. Wat. Works Ass. 45.

Brown, R.H., Konoplyantsev, A,A., Ineson, J., and Konalevsky, V.S.

(eds.) (1975)

Groundwater studies, an International guide for research and

practice.

The UNESCO Press, Paris.

Brunsden, D. (1964) The origin of decomposed granite on Dartmoor, In Dartmoor Essays,

(ed.) Simmons, I.

Dev. Assoc. 97-116.

Brunsden, D., and Gerrard, J. (1977) The physical environment of Dartmoor, in Dartmoor, a new study,

(ed.) Gill, C. 21-52, pub. David and Charles, Newton Abbot.

Buchan, S. (1965) Hydrogeology and its part In the hydrologlc cycle.

Proc. Inst. Civ. Engrs. 31, 428-431.

Burgess, W.G., Clowes, U.R., Lloyd, J.W., and Marsh, J.M. (1976) Aspects of the hydrogeology of St. Mary's, Scllly Isles*

Proc. Ussher Soc. 3, 382-389.

- 6 Burgess, W.G,, and Clowes, U.R. (1975)

Hydrogeological study of St. Mary's, Isles of Scllly,

Msc. dissertation, Dept. Geol. Scl. Birmingham University, 1975.

Burgess, W.G., Edmunds, W.M., Andrews, J.N., Kay, R.L.F., and Lee, D.J. (1982)

The origin and circulation of groundwater in the Carnmenellis Granite: The hydrogeochemical evidence.

Rep. Inst. Geol. Sci. Geophysics and Hydrogeology Division, Investigations of the geothermal potential in the U.K. (in preparation).

Burke, W. (1972) Aspects of the hydrology of Blanket Peat in Ireland, in Hydrology of Marsh-Ridden Areas. Proc. Minsk. Symp. June 1972.

Butler, S.S. (1959)

Engineering Hydrology.

Pub. Prentice-Hall, Inc.

Cainey, T. (1972)

Hydrogeological investigations of the Magnesium Limestone of S.E. Durham, England. J. Hydrol.. 16, 323-340.

Campbell, M.D., and Lehr, J.H. (1973)

Water Well Technology.

McGraw-Hill Book Company, New York, 1973.

Carson, M.A., and Kirkby, M.J. (1972)

Hillslope Form and Process.

Pub. Cambridge University Press.

Charoy, B. (1975)

Ploemeur kaolin deposit (Brittany), an example of hydrothermal alteration. Petrologie 1 (4), 253-266.

7 - Chorley, R.J. (ed.) (1969) Introduction to Physical Hydrology. Pub. Methuen and Co. Ltd.

Chow, V.T. (1952)

On determination of transmisslblllty and storage coefficients from

pumping test data. Transv Amer. Geophy. Union 33, 397-404.

Chow, V.T. (1964) Handbook of Applied Hydrology.

McGraw-Hill Book Company, U.S.A.

Clayden, B., and Manley, J.R. (1964) The soils of the Dartmoor Granite In Dartmoor Essays, ed. Simmons, I-G.

The Dev. Assoc., p.p. 117-140.

Cleaves, E.TJ, Godfrey, A.E.", and Bricker," O.P.' (1970)

Geochemical Balance of a small watershed and its geomorphlc implications. Geol. Soc. of Amer. Bull., 81, 3015-3032.

Coleman, J.D., and Farrar, D.M. (1966) Soil moisture deficit and watertable depth in London clay beneath

natural grassland. Q.J. Roy. Met. Soc, 92, 162-168.

Collins, J.H. (1909) Notes on some features observeable on the Carpalla China Clay Pit, Cornwall. Q.J. Geol. Soc, London 65(1), 155-156.

Connorton, B.J., and Reed, R.N. (1978) A numerical model for the prediction of long-term well yield in an unconfined chalk aquifer. Q.J. Eng. Geol. London 11(2), 127-138.

- 8 - Conway, V.M., and Millar, A. (1960)

The hydrology of some small peat-covered catchments in the North

Pennines.

Journ. Inst. Wat. Engrs., 14, 415-424.

Cook, J.M., Edmunds, W.M., and Miles, D.L. (1979) Water sampling and field chemical measurements for overseas hydrogeological surveys. Institute of Geological Sciences, Hydrogeology Unit. Int. Rep. WD/S.T./79/H.

Cooper, H.H., Bredehoeft, J.D., and Papadopulus, I.S. (1967)

Response of a finite diameter well to an instantaneous charge of

water. Wat. Res. Res., 3(1), 263-269.

Crook, J.M., and Howell, F.T. (1970)

The characteristics and structures of the Permo-Trlassic sandstone

aquifer of the Liverpool-Manchester Industrial region of N.W.

England. Int. Symp. Groundwater, Palermo, Dec. 1970, p.p. 217-255.

Cruse, K. (1979) A review of waterwell drilling methods. Q.J. Eng. Geol. London 12, 79-95.

Cryer, R, (1976)

The significance and variation of atmospheric nutrient inputs in a

small catchment system.

J. Hydrol., 29, 121-137.

Cullen, R.J. (1968) Modern Water Well Drilling Techniques, unpublished B.Sc- dissertation, Southampton University.

Danfors, E. (ed.) (1975) Soil Water Distribution: a state of the art report. Nordic I.H.D. Report No. 9 Oslo.

- 9 Dangerfield, J., and Hawkes, J.R. (1969)

Unroofing the Dartmoor Granite and possible consequences with regard to mineralisation.

Proc. Ussher. Soc, 2(2), 122-131.

Davis, A.W. (1971)

Changes in river quality associated with storm hydrograph.

Unpublished M.Sc. Thesis, Newcastle University.

Davis, S.N. (1964) Silica in streams and groundwater. Amer. Journ. Scl., 262, 870-891.

Davis, S.N., and Turk, L.J. (1964) Optimum depth of wells in crystalline rocks. Groundwater 2(2), 6-11.

Davis, J.C. (1973)

Statistics and Data Analysis in Geology. Pub. John Wiley and sons Inc. London.

Davis, S.N., and DeWlest, R.J.M. (1966)

Hydrogeology.

Dearman, W.R. (1964) Dartmoor, its geological setting, in Dartmoor Essays, (ed.) Simmons, I.G. Dev. Assoc.

De Geer, J. (1966)

Some hydrogeologlcal aspects on aquifers especially Eskers, in Groundwater Problems. Vol. II, (eds.) Eriksson, Gustafsson and Nilsson. Pub. Pergamon Press.

De la Beche, N.I. (1839) Report on the Geology of Cornwall, Devon and West Somerset.

Mem. Geol. Surv., 28, H.M.S.O. London.

10 - Demek, J. (1965) Slope development in granite areas of Bohemian Massif

(Czechoslovakia) Zeit. fUr Geomorph., Supp. 5, 82-106.

De Ridder, N.A., and Wit, K.E. (1965) A comparative study of the hydraulic conductivity of unconsolidated

sediments. J. Hydrol. 3, 180-206.

Desmedt, F., Van der Beken, A., and Demaree, G. (1977)

Investigations of the hydrologlcal balance in a peat swamp.

J. of Hydrol., 34, 151-160. .

JHnes, H.G. (1950) The metalliferous mining region of South West England.

Mem. Geol. Survey, 1, 1-508 and 2, 509-759 London.

Dixey, F. (1950) A practical handbook of Water Supply.

Pub. Thomas Murby and Co. London.

Draper, N., and Smith, H. (1980)

Applied Regression Analysis. Pub. J. Wiley and Sons Inc. Chichester, 2nd ed.

Downing, R.A., and Williams, B.P. (1969) The groundwater hyrogeology of the Lincolnshire limestones with special reference to the groundwater resources. Wat." Resources Board. Pub. No.9.

Dudarenko, H. (1981) Factors Influencing fluctuations in groundwater levels and temperatures in an unconfined aquifer. Narrator, Dartmoor, England.

Unpublished Undergraduate Dissertation, Department of Geographical Sciences, Plymouth Polytechnic.

- 11 - \gri^r>u C^^-'-^V ^^-f^u-^t £^U.v^d JrJ^. pui>. W.A..^.O. ^o.a

Edmonds, E.A., McKeowan, M.C., and Williams, M. (1975) British Regional Geology: South West England* 4th ed. pub. H.M.S.O. London.

Edmunds, W.M., Owen, M., and Tate, T.K. (1976) Estimation of Induced recharge of river water Into chalk boreholes at Taplow using hydraulic analysis, geophysical logging and geochemlcal methods.

Rep. Inst. Geol. Scl. 76/5 pub. H.M.S.O. London.

Ellis, E.E. (1906) Occurrence of water In crystalline rocks. U.S.G.S. Wat. Sup. Paper, 1494-E, 19-28.

Englund, J.0,^(1980) Groundwater Discharge through Springs with well-defined outlets, a case study from Jel^ya-Moss, South Norway. Nordic Hydrol., 11, 145-158.

Enslln, J.F. (1943) Basins of decomposition In Igneous rocks. Their importance as underground water reservoirs and their location by the electrical relstivlty method. Trans> Geol. Soc. South Africa, XLVI, 1-13.

Eriksson, E. (1960) The yearly circulation of Chloride and Sulphur In nature; Meteorological geochemlcal and pedological implications. Tellus 2 (12), 63-109.

Eriksson^ E. (1955) Airborne salts and the chemical composition of river waters. Tgll'us 7, 243-250.

Eriksson, E., and Khunakasem, V. (1968) The Chemistry of groundwaters la: Groundwater problems, Eriksson, E., Gustasfsson, Y., and Nllsson, K. (eds.). Pergamon Press, Oxford, p.p. 110-146.

- 12 - Exley, C.S. (1958) Magmatic differentiation and alteration in the St. Austell Granite.

Quart. Journ. Geol. Soc, 114, 197-230.

Exley, C.S. (1976)

Observations on the formation of kaolinite in the St. Austell

Granite, Cornwall.

Clay Minerals, 2, 51-63.

Ezekiel, M., and Fox, K.A. (1959)

Methods of correlation and regression.

3rd ed. pub. John Wiley and Sons Inc., New York, 1959.

Farrar, R.E.S. (1969)

Slblyback Reservour, Cornwall.

Civil. Eng. Dams Feature Nov. 1969, 1121-1124.

Farquharson, F.A.F., Mackney, D., Newson, M.D., and Thomasson, A.J. (1978) Estimation of runoff potential of river catchments from soil Indices, Soil Survey, Special Survey, No.6, Rothamsted, Harpenden

Feller, M.C., and Kimmins, J.P. (1979)

Chemical Characteristics of small streams near Haney in South

Western British Columbia. Wat. Res. Res., 15(2), 247-258.

Feth, J.H., Robertson, C.E., and Polzer, W.L. (1964)

Sources of Mineral Constituents in Water from Granitic Rocks, Sierra Nevada, California and Nevada. U.S.G.S. Wat. Sup. Paper 13531.

Fetter, C.W. (1980)

Applied Hydrogeology. Pub. C.E. Merrll Publishing Co., Columbus Ohio. rcn-i^.T.C-, and Knocks, Ji3- (J^^l-)

Ferris, J.G., Knowles, D.B., Brown, R.H., and Stallman, R.W. (1962)

Theory of aquifer tests.

U.S.G.S. Wat. Sup. Paper 1536-E.

- 13 - Ferris, J.G., and Knowles, D.B. (1963)

The slug-injection test for estimating the coefficient of transmisslblllty of an aquifer in "Methods of Determining Permeability, Transmissibllity and Drawdown", (ed.) Bentall, R. U*S.G.S. Wat. Sup. Paper 1536-1.

Flnlayson, B.L. (1978) Suspended solids transport in a small experimental catchment.

Zelt. fUr Geomorph., 22, 192-210.

Fitch, F.J., and Miller, J.A. (1964)

The age of the paroxysmal Variscan Orogeny in England, in the

Phanerozolc Time-Scale, Quart. Journ. Geol. Soc, 120 S, 159-175.

Fltzpatrick, E.A. (1956) An indurated soil horizon formed by permafrost.

Journ. Soil Scl., 7, 248-254.

Fitzpatrick, E.A. (1963) Deeply weathered rock in Scotland, its occurrence, age and contribution to the soils. Journ. Soil Sci., 14(1), 33-43.

Fookes, P.G., Dearman, W.R., and Franklin, J.A. (1971) Some Engineering aspects of rock weathering with field examples from Dartmoor and elsewhere* Q. J. Eng. Geol., 4, 139-185.

Foster, S.S.D-, and Milton, V.A. (1974)

The permeability and storage of an unconfined chalk aquifer.

Hydrol. Scl. Bull., 19, 485-500.

Foster, S.S.D., and Milton, V.A. (1976)

Hydrologlcal basis for large-scale development of groundwater storage capacity in the East Yorkshire chalk. Rep^. Inst. Geol. Sci., Report 76/3 pub. H.M.S.O. London.

14 Foster, S.S.D., Parry, E.L., and Chilton, P.J. (1976)

Groundwater resource development and saline water intrusion in the chalk aquifer of North Humberslde. Rep.~ InstGeol. Sci., Report 76/4 pub. H.M.S.O. London.

Foster, I.D.L. (1978) A multivariate model of storm-period solute behaviour*

J. Hydrol. 39, 339-353.

Foster, I.D.L. (1979) Intra-catchment variability in solute response, an east Devon

example, Earth Surface Processes, 4, 381-394.

Fox, I.A., and Rushton, K.R. (1976) Rapid recharge In a limestone aquifer.

Groundwater, 14(1), 21-27*

Freeze, R.A., and Cherry, J*A. (1979) Groundwater Pub. Prentice-Hall Inc. New Jersey.

Freund, J.E. (1974) Modern Elementary Statistics. 3th ed. pub. Prentice-Hall International Editions.

Garrels, R.M., and Christ, CL. (1965) Solutions, Minerals and Equilibria, pub. Freeman Cooper and Company, California.

Garrels, R.M*, and MacKenzle, F.T. (1967) Origin of the Chemical Composition of some Springs and Lakes. Advances in Chemistry Series, Amerr Chemr~5oc* 67, 222-242*

Gear, D*J. (1951) Underground water supplies in the granitic and gnelssose rocks of eastern Uganda. Union Geodesique et Geophysique. Inter. 2, 252-261.

15 - Gibb, A. (1980) Cemeteries sought for nuclear waste. New Sci., 85 (1194), 488-491.

Gibson, M.J. (1977) The place of resistivity surveying in solving environmental problems, BSc (unpublished) Dissertation, Geographical Sciences, Plymouth Polytechnic.

Glendlnlng, S.J. (1980) A preliminary account of the hydrogeology of Altnabreac, Caithness. Rep. Inst. Geol. Sci., ENPU 80-6.

Godwin, H. (1931)

Studies in the Ecology of Wicken Fen, (1) The Groundwater Level of the Fen. Journ. Ecol., 19, 449-473.

Golterman, H.L., and Clymo, R.S. (1971) Methods for chemical analysis of fresh waters. I.B.P. Handbook No.8, Blackwell Scientific Publications, Oxford, 3rd ed.

Gomez, B., and Sims, P.C. (1981) Overbank deposits of the Narrator Brook, Dartmoor England. Geol.-Mag., 118 (1), 77-82.

Gorham, E. (1961) Factors influencing the supply of major ions to inland waters, with special reference to the atmosphere. Geol~Soc'r"Amerr"Bull., 72, 795-840.

Granant, L. (1972) On the relation between pH and the chemical composition in atmospheric precipitation. narr-t-tet~Instr-Rept., AC-18, Stockholm.

16 - Granneman, N.G., and Sharp, J.M. (1979)

Alluvial hydrogeology of the Lower Missouri River Valley.

J. Hydrol., 40, (1/2), 85-99.

Gray, D.A. (1969) The groundwater hydrology of the Yorkshire Quae River Basin,

Wat. Sup. Pap. I.G.S., Hydrogeological Rep. No.4, pub. N.E.R.C.,

London.

Green, CP., and Eden, M.J. (1971)

Gibbsite in weathered Dartmoor granite.

Geoderma 6, 315-317.

Gregory, K.J-, and Walling, D.E. (1974) Fluvial processes In instrumental watersheds, Inst. Brit. Geog., London.

Groundwater Manual - A Water Resources Technical Publication. U.S.

Dept. of the Interior, Bureau of Reclamation, 1st ed. (1977).

Groundwater in the U.K., Year 1971-1973.

H.M.S.O. (1978).

Grundy, F. (1951) The groundwater depletion curve, its construction and uses. Assemblee Generale~de~Bruxelles,"~InternT"d*~TIydrbro^ 2, 213-217.

Gulllermo, D., Sewell, J.I-, and Shelton, C.H. (1968) An application of Principal Components Analysis and Factor Analysis in the Study of Water Yield. War.-Res.-^Res., 4(2), 299-306.

Gustafsson, Y. (1966) The influence of topography on groundwater formation, in "Groundwater Problems" Vol.11, (ed.) Eriksson, E., Gustafsson, Y. and Nilsson, K. Pub. Pergamon Press.

17 Haan, C.T., and Allen, D.M. (1972)

Comparison of Multiple Regression and Principal Component

Regression for Predicting Water Yields in Kentucky,

Wat. Res. Res., 8(6), 1593-1596.

Halton, T. (1954) Hydrological conditions In the chalk at Compton, West Sussex.

Journ. Instn. Wat. Engrs., 8, 138-144.

Hall, D.G. (1967) The Assessment of Water Resources In Devon, England using limited

hydrometrlc data, Internat. Assoc* Scl. Hydr. Pub. 76, 110-120.

Hall, F*R* (1968) Baseflow Recessions - A Review Wat. Res. Res., 4(5), 973-983.

Hamrol, A. (1961) A quantitative classification of the weathering and weatherability

of rocks.

Proc . 5 th Int.' Conon Soll^'Mechanl'cs"'and'"Foundat l'on~Eng, 2

771-774.

Hammond, R., and McCullagh, P.S. (1978)

Quantitative Techniques in Geography,

pub. Clarendon Press, Oxford, 2nd ed.

Hantush, M.S., and Jacob, C.E. (1955) Non-steady radial flow in an Infinite leaky aquifer.

Trans. Amer. Geophys* Union 36, 95-100.

Harris, H. (1972) Indii'^trlal Archaeology of Dartmoor

Pub. David and Charles, Newton Abbot.

- 18 Harrod, T.R*, Hogan, D*V., and Staines, S.J. (1976)

Soils in Devon II, Sheet SX65 (Ivy Bridge) Soil Survey Record,

No.39, Soil Survey Memoir, Harpenden*

Haslam, H.W* (1975) Geochemlcal survey of stream waters and stream sediments from the

Cheviot Area, Rep* Inst* Geol. Scl*, 75/6*

Hauser, D.P. (1974) Some problems in the use of Stepwise Regression Techniques in

geographical research, Canadian Geographer XVIII, 2, 148-159.

Hawkes, J.R., and Chaperlln, K. (1966)

The Dartmoor Granite* A summary of the Geological Surveys

findings to date in relation to the Cornish bodies of granites in

general.

Pr'Q'fc-rnJsflherT-Soc., 1, 206-207.

Headworth, H.G. (1972)

The analysis of natural groundwater level fluctuations in the chalk

of Hampshire. Journ* Instn. Wat. Engrs*, 26, 107-124.

Headworth, H.G. (1978)

Hydrogeologlcal characteristics of Artesian boreholes in the chalk

of Hampshire. Q.J. Eng. Geol. London 11(2), 139-144*

Heath, R*C., Trainer, F*W* (1968) Introduction to groundwater hydrology. Pub* John Wiley and Sons, London.

Hem, J.D. (1970) Study and interpretation of the chemical characteristics of natural water.

U.S.G.S* Wat.^Sup. Pap-. 1473.

- 19 Henwood, W. (1843) Observations on subterranean temperatures in the mines of Cornwall and Devon. Transv R. Geol. Soc. Cornwall. 5, 387-412.

Hertzler, R.A. (1939)

Engineering aspects of the influence of forests on mountain streams. Civ. Eng., 9 (8), 487-489.

Hewlett, J.D., and Hlbbert, A.R. (1967) Factors affecting the response of small watersheds to precipitation in humid areas, in Proceedings of the International Symposium on Forest Hydrology, 1965, Pennsylvania State University, pub. Pergamon.

Hlbbert, A.R., and Cunninghan, G.B. (1967)

Streamflow data processing opportunities and application, in Soper,

W.E.", Lull] W.WJ (edsO International Symposium on Forest Hydrology, Oxford, p.p. 725-736.

Horton, R.E. (1933) The role of infiltration in the hydrologic cycle, Transv Amer. Geophys.' Union, 14, 446-460.

Howcroft, H.J. (1977) The hydrology of a small catchment Flerd-Studles, 4(4), 555-574.

Hvorslev, M.J. (1951) Timelag and Soil Permeability in Groundwater Observations, Bull. 36, U.S.—Corps, of Engrs. Waterways Experimental Station, Vlcksburg, Miss.

20 Illsley, A.E., Wakeling, T.R.M., and Humphreys, J.D. (1976) The Kernlck and Portworthy Mica Residue Disposal Schemes for E.C.L.P. in Cornwall and South Devon, U.K. Douzlerme CongrSs des Grand's Barrages, Mexico, Q4A, 437-461.

Ineson, J. (1953) Some observations on pumping tests carried out on chalk, wells. Journ. Instn. Wat. Eng., 7, 215-225.

Ineson, J. (1959) Yield-depression curve** of discharging wells with particular reference to chalk wells and this relationship to variations in transmissibillty, Journ. Instn. Wat. Eng., 13, 119-163.

Ineson, J. (1962) A hydrogeological study of the permeability of chalk, Journ. Instn. Wat. Eng., 16, 449-463.

Ineson, J., and Downing, R.A. (1964) The groundwater component of river discharge and its relationship to hydrogeology, Journ. Instn. Wat. Eng., 18, 519-541.

Ineson, J., and Downing, R.A. (1965) Some hydrogeological factors in permeable catchment studies. Jour" I' l.'Stn. Wat. Eng., 19, 59-80.

Ineson, J. (1967) Groundwater conditions in the coal measures of the South Wales coalfields. Wat. Sup. Papers Hydrogeological Report No.3, GeolT''Survey of Great Britain.

Jacks, G. (1973) Chemistry of some groundwaters in igneous rocks. Nordic Hydrol., 15, 207-236.

- 21 Jacob, C.E. (1940) On the flow of water in an elastic artesian aquifer. TrahQ. Amer* Geophys. Union, 21, 574-586.

Jacob, C.E. (1963) The recovery method for determining the coefficient of transmissibillty, in Methods of Determining Permeability, Transmlssibility and Drawdown, ed. Bentall, R. U.S.G.S. Wat. Sup. Pap. 1536 I, 283-292.

Jeffcoate, P.R. (1977) Review of alluvial aquifers in England and Wales, Central Water Planning Unit, Reading.

Johnson, E.E. (1975) Groundwater and Wells. Pub. Johnson Division, U.O.P. Inc.

Jones, A. (1971) Soil Piping and Stream Channel Initiation, War.-Resv Res., 7(3), 602-610.

Jones, G.P. (1970) Development of Shallow Non-Consolidated Aquifers. Pro'cv Instn. CivEngrs., 47, 449-455.

Jones, G.P. (1977) Utilisation of groundwaters in Britain: in Papers and Proceedings W.R.C. Conference on Groundwater Quality - Measurement, Prediction and Protection, Reading, p.p. 1-16.

Kazmann, R.G. (1977) River infiltration as a source of groundwater supply, Procv Am. Socv Civ. Engrs., 73(6), Part 1, 837-853.

Keller, H.M. (1970) Factors affecting water quality of small mountain catchments. J.-Hydrol. (N.Z.), 9(2), 133-141.

22 Kennard, J., and Lee, J. (1947) Some features of construction of Fernworthy Dam, Journ. Instn. Wat* Engrs,, 1, 11-38,

Kennard, M.F,, and Knlll, J,L, (1969) Reservoirs on limestone, with particular reference to the Cow Green Scheme, Journv'Instn. Wat. Engrs., 23, 87-136.

Kent, M. (1979) Regional evaluation for colliery spoil reclamation. Ph.D. Thesis (unpub-) Dept. Landscape Architecture, University of Sheffield.

Kent, M., and Wathern P. (1980) The vegetation of a Dartmoor Catchment. Vegetatlo., 43, 163-172.

Kikkawal K.] and Kawanlshl^ H." (1967) Sudden rise of groundwater table caused by severe rainfall, Special contributions. Geophysical Institute, Kyoto University, No.7, 119-124.

Kilmartln, R.F., and Peterson, J.R. (1972) Rainfall-runoff regression with logarithmic transformations and zeros in the data, Wat. Res. Res., 8, 1096-1099.

Kirbyi C. (1979) Water in Great Britain, Pub. Pelican, 1979.

Kirkby, M.J-, and Chorley, R.J. (1967) Throughflow, overland flow and erosion, Bi IT.- Int. Assoc. Sci. Hydr., 12, 5-21.

- 23 Kirkby, M,J. (1978) Hlllslope hydrology. Pub. John Wiley and Sons*

Kirkham, D. (19A7) Studies of hillside seepage In Che lowan Drift Area, Proc. Soil Scl. Soc*' Ainer*, 12, 73-170,

Kltchlng, R. (197A) Infiltration studies on the Lower Greensand outcrop near Llss, Hampshire, Rep. Inst* Geol* Sci., 74/10, pub* H.M.S.O* London*

Knlll, J.L* (1972)a Engineering geology in reservoir construction in South West England, Proc. Ussher* Soc*, 2(5), 359-371.

Knill, j'.h'. (1972)b Assessment of Reservoir Feasibility, Q.J. Eng,^ Gaol., A, 355-372.

Kruel, W*F*J.M., and Leifrlnlck, F.A. (1946) Recent groundwater investigations in the Netherlands - Monograph on the progress of research in Holland. Pub. Elsevier Pub. Co. Amsterdam.

Krumbein, W.C, and Grayblll, F.A. (1965) An introduction to statistical models in geology. Pub. McGraw-Hill, New York.

Kruseman, G.P., and De Rldder, N.A. (1970) Analysis ad evaluation of pumping test data. International Institute for Land Reclamation and Improvement, Wageningen, The Netherlands.

Kulandlswamy, V.C, and Seetharaman, S* (1969) A note on Barnes* Method of hydrograph separation* J. Hydrol,, 9, 222-229.

24 - Kunkle, G.R. (1962) The baseflow duration curve, a technique for the study of groundwater discharge from a drainage basin. Jourri. Geophy. Res.. 67(4), 1543-1554.

Kurtz, L.T., and Melsted, S.W. (1972) Movement of chemicals in soils by water. Soil Sci.. 115(3), 231-239.

Land, D.H. (1966) Hydrogeology of the Hunter Sandstone in Nottinghamshire. Hydrogeological Report 1, Institute Geological Sciences London.

Lahermo, P. (1970) Chemical Geology of Groundwater and Surface Waters in Finnish Lapland. Geol. Survey of Finland, Bull. 242, 1-105.

Landers, R.A., and Turk, L.J. (1973) Occurrence and quality of groundwater in crystalline rocks of the Llano area, Texas, Groundwater. 2(1), 5-10.

Langbein, W.B. (1938) Some Channel Storage Studies and their application to the determination of infiltration Trans. Amer. Geophys. Union, 19, 435-447.

Larson, I. (1966) Groundwater in Precambrian Rocks in Southern Sweden, in Groundwater Problems, Vol.11, ed. Eriksson, E., Gustafsson, Y., Nilsson, K. Pub. Pergamon Press, London, 1966.

Laurenson, E.M. (1961) A Study of Hydrograph Recession Curves of an Experimental Catchment, Joum. Instltn. of Engr. Australia, July-Aug., 253-258.

25 Legrand, U.E. (1949) Sheet Structure, a major factor in the occurrence of groundwater in the granites of Georgia. Economic Geol., 44, 110-118.

Legrand, H.E., and Strlngfleld, V.T. (1973) Karst Hydrology - a reviw, J. Hydrol., 20, 97-120-

Am. Sea. so, 5" -^3^ Lewis, D.C., and Burgy, R.H. (1964) Hydra-llc characteristics of Fractured and Jointed Rocks. Groundwater, 2, 4-9.

Liakopoulos, A.C. (1965) Retention and distribution of moisture in soils after infiltration has ceased. Bull. Internat. Assoc. Scl. Hydr. 10, 58-69.

Llndblom, U.E., LundstrOm, L., and Stille, H. (1978) Large-scale permeability tests in the granite of the Strips Mine Sweden. Proc. of the Organisation for Economic Co-operation and Development, Seminar on in situ heating experiments in geological formations, Stripa, Sweden, p.p. 207-219.

Linsley, R.K-, and Franzini, J.B. (1972) Water-Resources Engineering, 2nd Edition, pub. McGraw-Hill, London, 1972.

Linsley, R.K., Kohler, M-A., and Paulus, J.L.H. (1949) Applied Hydrology. Pub. McGraw-Hill Book Co., London, 1949.

Linton, D.L. (1955) The Problem of Tors. Geogr. Journ., 121, 470-487.

26 - Lohman, S.W. (1972) Groundwater Hydraulics. U.S.G.S. Prof. Paper 708

Lovelock, P.E.R., Price, M., and Tate, T.K. (1975) Groundwater conditions In the Penrith Sandstone at Cllbum, Westmorland. Journ. Instn. Wat. Eng. Sci., 29, 157-173-

Lumb, P. (1962) The properties of decomposed granites, Geotechnlque, 12, 226-243.

MacMereth, F.J.H., Heron, J., and Tailing, J.T. (1978) Water Analysis - some revised methods for Limnologists. Freshwater Biological Association, Scientific pub. No- 36. ^okhir P.fA. 0U<*) , , , , .

Mather, J.D., Chapman, N.A., Black, J-H., and Lintern, B.C. (1982) The Geological disposal of high level radioactive waste - a review of the Institute of Geological Sciences Research Programme. Nucl. Energy 21, No.3, 167-173.

McGowan, M. (1975) Depths of water extraction by roots - application to soil-water balance studies, I.A.E.A.-S.M. 176 (17), 435-445.

Meier, 0., and Petersson, S-G. (1951) Water suplies with Archaen bedrocks of Sweden. Int.-Assoc. Sci. Hydr., Pub. No.33, 252-61,

Melnzer, O.E. (1923) Outline of Groundwater Hydrology, with Definitions. U.S.G.S. Wat. Sup. Paper, No.494.

27 - Melnzer, O.E. (1923) The occurrence of groundwater in the United States. U.S.G.S. Wat. Sup. Paper, No.A89.

Miller, J.A., and Mohr, P.A. (196A) Potassium-argon measurements on the granites and some associated rocks from South West England. Geol. Journ., A, 105-26.

Monkhouse, R.A. (1975) Design of short-duration Aquifer Tests, Central Water Planning Unit, Tech. Note. No.10, Reading.

Monkhouse, R-A., and Phillips, S. (1978) The design, construction and maintenance of recharge wells. Central Water Planning Unit, Tech. Note. No.25.

Monkhouse, R.A* (1978) Alluvial Aquifers In England and Wales, Central Water Planning Unit, Tech. Note. No.27.

Moore, I.C-, and Gribble, CD. (1980) The suitability of aggregates from weathered Peterhead granites, Q.J, eng. Geol., 13, 305-313.

Morey, G.W., Fournier, R.O., and Rowe, J.J. (1962) The solubility of quartz in water in the temperature Interval from 25°C - 300**C.

Geochem. cosmochlm. Acta., 26, 1029-1043.

Morton, E. (1973) A review of the Influence of geology on the design and construction of impounding dams, Journ. "Inst. Wat. Engrs. Scl., 27 (5), 243-27lJ

Moye, D.G. (1955) Engineering Geology for the Snowy Mountain Scheme, Journ;-Instn. Engrs. Australia, 27, 287-298.

- 28 - Myers, L.E., and Van Bavel, C.H.M. (1963) Measurement and evaluation of watertable elevation. Int.' Com. on 'Irrig\~'and"Drain., Tokyo, Japan, Q17, 109-119.

Murgatroyd, A.L. (1980) Fluvial transport in the Narrator Brook, Devon: a summary of sources, dynamics and controls. Ph.D. Thesis, C.N.N.A. Plymouth Polytechnic.

Naylor, J.A. (1974) The groundwater resources of the river gravels of the middle Thames Valley. Water Resources Board, Reading 1974.

Nilsson, L.Y. (1966) Short-time variations of the groundwater and its reasons, in Groundwater Problems, (eds.) Eriksson, E., Gustafsson, Y., Nilsson, K. p^p. 57-72, pub. Pergamon Press, London, 1966.

Nutbrown, D.A-, and Downing, R.A. (1976) Normal Mode Analysis of the Structure of Baseflow Recession Curves, J. Hydrol. 30, 327-340.

Oilier, CD. (1969) Weathering pub. Oliver and Boyd, Edinburgh.

Painter, R.B. (1971) A hydrological classification of the soils of England and Wales. Procv Instn. Civ. Engrs., 48, Tech. Note 29, 93-95.

Palmer, J., and Neilson, R.A. (1962) The origin of granite tors on Dartmoor, Devonshire. Proc.- York Geol. Soc., 33, 315-40.

29 Palmer, W.C., and Havens, A.V. (1958) A graphical technique for determining evapotransplration by the Thornthwalte Method, Mon^-'-ly Weather Review 86, 123-8.

^ Pr^u^dowo J of \c^r^ d.a«^r Papadopulos, I.S., Bredehoeft, J.D., and Cooper, H.H. (1973) On the analysis of Slug-Test Data, Wat. Res. Res., 9, 1087-1089.

Parker, G.G., and Stringfield, V.T. (1950) Effect of earthquakes, trains, tides, winds and atmospheric pressure changes on water in the geologic formations of Southern Florida, Economic Geology 45, 441-460.

Parsons, M.L. (1970) Groundwater Thermal Regime in a Glacial Complex. Wat. Res. Res., 6 (6), 1701T1720.

Pearson, R., and Money, M.S. (1977) Improvements in the Lugeon or packer permeability test, Q.J.-Eng.- Geol. 10, 221-239.

Peck, A.J. (1960) The watertable as affected by atmospheric pressure, Journ. Geophys. Res., 65 (8), 2383-2388.

Phillips, S. (1978) Prediction of Groundwater Levels, Tech. Note No.21, Central Water Planning Unit, Reading.

Pickering, W.F. (1977) Pollution Evaluation - the Quantitative Aspects, Environmental Science and Technical Series, Vol.2, pub. Marcel Dekker Inc., New York.

- 30 Piper, A.M. (1944) A graphic procedure In the geochemlcal Interpretation of water analyses, Amer. Geophys. Union Trans., 25, 914-923.

Pltty, A.F. (1966) An approach to the study of Karst Water, University of Hull, occasional papers in Geog. No.5.

Pitty, A.F. (1973) Introduction to Geomorphology, pub* Methuen and Co. Ltd.

Pltty, A.F., Halllwell, R.A., Ternan, J.L., Whittel, P.A-, and Cooper, R.G. (1979) The range of water temperature fluctuations in the Limestone Waters of the Central and Southern Pennines, J* Hydrol., 41, 157-160*

Pricej M.'i and Foster^ S^S.D^ (1974) Water supplies from Ulster Valley Gravels, Proc.'Instnv Civ;' Engrs•, 57 (2), A51-A66.

Prickett, T.A. (1965) Type-curve Solution to Aquifer Tests under Watertable Conditions, Groundwater, 3, 5-1A.

Proctor, J.F., and Marine, I.W. (1965) Geologic, Hydrologlc and Safety Considerations in the Storage of Radioactive Wastes In a Vault Excavated in Crystalline Rock, Nuclear Science and Eng., 22, 350-665.

Rainwater, F.H., and Thatcher, L.L. (1960) Methods of Collection and Analysis of Water Samples U.S.G.S. Wat. Sup. Paper No.lA5A*

Ralston, A., and Wllf, S- (1960) Mathematical Methods for Digital Computers, pub. John Wiley.

31 - Rasmussen, W.C, and Andreasen, G*E. (1959) Hydrologlc Budget of the Beaverdam Creek Basin, Maryland. U.S.G.S* Wat. Sui^. Paper No.1472.

Read, H.H., and Watson, J* (1968) Introduction to Geology, Vol.1, Principles, pub. Macmillan.

Reeve, R.C., and Klrkham, D. (1951) Soil Anlsotropy and Some Field Methods for Measuring Permeability, Tran°. Amer. Geophys. Union 32, 582-590.

Reeves, M.J,, Skinner, A.C., and Wilkinson, W.B. (1975) The relevance of aquifer-flow mechanisms to exploration and development of groundwater resources, J. Hydrol., 25, 1-22.

Reeves, M,J., Parry, E.L., and Richardson, G. (1978) Preliminary evaluation of the groundwater resources of the western part of the Vale of Pickering, Q.J. Eng. Geol,, 11 (3), 253-262.

Reid, C, Barrow, G., Sherlock, R.L., MacAlister, D.A., Dewey, H., and Brownhead, C.N. (1912) The Geology of Dartmoor Mem. Geol. Survey London.

fSf^r. lycGr "^J,^*^- rr-ot^m. tSfl'^f-^ Research Panel (1969) Observation Boreholes, Construction and Use, final report of the Research Panel. Journv Instn. Wat. Engrs., 23 (9), 369-381.

Rice, R.M. (1967) Multivariate Methods Useful in Hydrology. Proc-.~Inter"HydroX.""Symp., Fort Collins U.S.A. 1967, p.p. 471- 478.

- 32 - Richard, S.J., Willard, L.S., Davis, S., and Spencer, J.R. (1973) Tensiometer use in Shallow Groundwater Studies, Journ* Irrig. and Drain Div.A.S.C.E., 99 (IRA), 457-464,

Rlggs, H.C. (1964) The Baseflow Recession Curve as an Indicator of Groundwater, rnrr-AssocT-Sci~.' Hydro., Pub. No.63, 352-363-

Rippon, J.E. (1978) Studies of Acid Rain on Soils and Catchments, Central Electricity Generating Board Report, 1978.

Rodda, J.C., Downing, R.A., and Law, F.M. (1976) Systematic Hydrology pub. Butterworth and Co. Ltd. London, 1976.

Rose, C.W. (1966) Agricultural Physics pub. Pergamon Press.

Rorabaugh, M-I. (1964) Estimating Changes in Bank Storage, Int. Assoc. Sei. Hydro., 63, 432-441.

Rushton, K.R. (1978) Estimating transmlssivity and storage coefficient from abstraction well data. Groundwater, 16 (2), 81-85.

Sage, R.C., and Lloyd, J.W. (1978) Drift deposit influences on the Trlassic Sandstone Aquifer of North West Lancashire as inferred by hydrochemlstry. Q.-J-.-Eng.- Geol., 11 (3), 209-218.

Sandeman, E. (1901) The Burrator works for the water supply of Plymouth Minutes Proc. Inst. Civ. Engrs., 146, 2-42-

33 Satchell, R.L.H*, and Edworthy, K*J. (1972) Bunter Sandstone: Artificial recharge. Water Resources Board, Trent Research Programme, 7, pub. H*M.S.O. London.

Schlcht, R.J., and Walton, W.C. (1961) Hydrologlc Budgets for three small watersheds in Illinois. Illinois State Water Survey, Urbana, U.S.A.

Schneider, R. (1962) An application of thermometry to the study of groundwater, U*S.G*S* Wat* Sup. Paper No.l5AA-B, 1-16.

Searcy, J.K. (1959) Manual of Hydrology; Part 2, Low Flow Techniques* U*S>G.S. Wat. Sup. Paper* No.l5A2-A.

Sherrell, F.W. (196A) Some aspects of groundwater in South West England, Pri'c~rvke^ckr^So'ci, ij 129-130."

Sherrell, F.W. (1970) Some aspects of the Trlasslc aquifer in East Devon and West Somerset, Q*J. Eng* Geol.; 2\ 255-286.

Shorter, A.H., Ravenhlll, W.L.D., and Gregory, K.J. (1969) South West England, pub. Thomas Nelson and Sons Ltd.

Siever, R., (1969) (Silicon) Abundance in natural waters. In: Handbook of Geochemistry 2-1, pub. Springer-Verlag, Berlin.

Sims, P.C. (1975) A hydrologlcal study of the Narrator Brook Catchment, Dartmoor, England. M.Sc. Dissertation (unpub.) University College London.

U.S. C

- 34 - Sloan, C.E. (1970) Prairie Potholes and the Watertable, Geological Survey Research 1970, Prof. Parer 700B, B227-B231.

Smith, K. (1966) Percolation, Groundwater discharge and Streamflow in the Nldd Valley, Journ. Instn. Wat. Engrs. Sci., 20 (7), 459-471.

Smith, K. (1972) Water in Britain, pub. MacMlllan.

Snyder, W.M. (1962) Some possibilities for Multivariate Analysis in Hydrologlc Studies, Journ. Geophys. Res., 67 (2), 721-729.

Stallman, R.W. (1963) Methods of collecting and interpreting groundwater data, U.S.G.S. Wat. Sup. Paper No. 1544-H.

Stallman, R.W. (1965) Effects of watertable conditions on waterlevel changes near pumping wells, Wat. Res. Res., 1 (2), 295-312.

Steele, T.D. (1976) A blvariate-regression model for estimating chemical composition of streamflow or groundwater. Int. Assoc. Sci. Hydrol. Bull., 21, 149-161.

Stephens, N. (1966) Some Pleistocene deposits in North Devon, Bioletyn Peryglacjalny 15, 103-113.

35 Stevenson, CM. (1968) An analysis of the chemical composition of rainwater and air over the British Isles and Eire for the years 1959-1964, Roy. Met. Soc. Quart. Journ., 94, 56-70.

Stewart, J.W. (1962) Water yielding potential of weathered crystalline rocks at the Georgia Nuclear Laboratory, U.S.G.S. prof. Paper No.450-B, 1-145.

Stiff, H.A. (1951) The interpretation of chemical water analysis by means of patterns, Journ. Petroleum Tech., 3 (10), 15-17.

Stow, G. (1962) Modern Water Well Drilling Techniques in use in the U.K., Proc. Inst. Civ. Engrs., 23, 1T14.

Tardy, Y. (1971) Characterisation of the principal weathering types by the geochemistry of waters from some European and African crystalline massifs. Chemical Geology, 7 (4), 253-271-

Te Punga (1957) Periglaciatlon in Southern England, Tl'jdschr. Kon.'Ned. Aardr. Gen-, (Rotterdam) 74, 401-12*

Ternan, J.L-, and Williams, A.C (1979) Hydrological pathways and granite weathering on Dartmoor, in Geographical approaches to Fluvial Processes^ (ed.) Pltty, A.F., p.p. 5-30. "^^^^^ ^^V\u\^o. h^^^ -{^ l6w*rt'^ of-tK, p.^o.^tr,c i-rfc^ ^U^ r«U a.xJ dum t.On

Thels, CV. (1938) The significance and nature of the cone of depression in groundwater bodies, Econ. Geol., 33, 889-902.

36 - Thornthwaite, C.W. (1948) An approach towards a rational classification of climate, Geog, Review 38, 55-94.

Toebes, S.L., and Strang, D.D. (1964) On Recession Curves, 1., Recession Equations, J. Hydrol. (New Zealand) 3 (2), 2-15.

Todd, D.K. (1959) Groundwater Hydrology, pub. Toppan Company Ltd., Japan.

Toth, J. (1971) Groundwater Discharge - A common generator of diverse geologic and morphologic phenomena. Bull, Int. Assoc. Sci. Hydro. XVI (1,3), 7-24.

Trafford, B.D., and Rycroft, D.W. (1973) Observations on the soil water regimes in a drained clay soil, Journ. Soil Sci. 24, 380-391.

Truesdale, V.W., and Smitch, C.J. (1976) The automatic determination of silicate dissolved in natural freshwaters by means of procedures involving the use of either a or p molybdosilic acid. Analyst. Jan. 1976, 19-31.

Turk, L.J. (1975) Diurnal fluctuations of watertables induced by atmospheric changes, J. Hydrol. 26, 1-16.

Uhl, V.W,, and Sharma, G.K. (1978) Results of Pumping Tests in Crystalline-rock Aquifer, Groundwater 16 (3), 192-203.

- 37 - Uhl, V.W. (1979) Occurrence of Groundwater In the Satpura Hills Region of Central India, J. Hydrol. 41 (1/2), 123-141.

-Uhl, V.W., Nagabhushanam, K,, and Johansson, J.O. (1979) Hydrogeology of Crystalline Rocks: Case Studies of two areas in India, Nordic Hydrol, 10, 287-308,

U.S. Pipt- of VKA =r-o(>oor (l«i7?^ / f x i -» Ussher, W.A.E. (1912) The Geology of the Country around and Modbury, Geol. Surv, Memoir.

Van Der Kamp, G. (1976) Determining Aquifer Transmissivity by means of Well Response Tests: The Underedamped Case, Wat. Res. Res. 12 (1), 71-77.

Van Everdingen, R.O. (1968) The influence of the South Saskatchewan Reservoir on the local groundwater regime - a prognosis, Geol. survey of Canada, Dept. of Energy, Mines and Resources 1968.

Walling, D.E. (1971) Streamflow from instrumented catchment in South-East Devon, in Gregory, K.J., Ravenhill, W.L.D. (eds) Exeter Essays in Geography, Exeter, p.p. 55-81.

Wallis, J.R. (1965) Multivariate statistical methods in Hydrology - a comparison using data of known functional relationships. Wat. Res. Res. 1 (4), 447-461.

Walton, W.C. (1970) Groundwater Resource Evaluation, pub. McGraw Hill.

38 Ward, R.C. (1962) Some Aspects of the Hydrology of the Thames Flood plain near Reading, Ph.D. Thesis, (unpub.). University of Reading.

Ward, R.C. (1963) The effect of site factors on water-table fluctuations, J. Hydrol. 1, 151-165.

Ward, R.C. (1975) Principles of Hydrology, 2nd ed. pub. McGraw Hill.

Ward, S.D., Jones, A.D., and Manton, M. (1972) The vegetation of Dartmoor, Field Studies 3, 505-533.

Waters, R.S. (1957) Differential Weathering and erosion on Oldlands. Geog. Journ. 123,. 503-509.

Waters, R.S. (1964) The Pleistocene Legacy to the Geomorphology of Dartmoor, In Dartmoor Essays, ed. Simmons, I.G. Devonshire Association p.p. 73-96.

Waters, R.S. (1971) The Significance of Quarternary Events for the Landforms of South West England, in Exeter Essays in Geography, (eds.) Gregory, K.J., Ravenhill, W.L.D., pub. University of Exeter.

Webber, N.B. (1961) The Base-Flow Recession Curve: Its derivation and application. Journ: Instn. Wat. Eng. Sci. 15, 368-377.

Werner. P.W., and Sundquist, K.J. (1952) On the i^roundwater Recession Curve for Large Water-sheds, Internat; Assoc. Sci. Hydr, Pub. No.33, p.p. 202-212.

39 Weyman, D.R. (1970) Throughflow on slopes and its relation to the stream hydrograph, Internat, Assoc. Sci. Hydr. Bull. 15, 25-33.

Weyman, D.R. (1973) Measurements of the downslope flow of water in soil, J. Hydrol. 20, 267-288.

Weyman, D.R. (1974) Run-off processes, contributing area and streamflow in a small upland catchment, in Fluvial Processes in Instrumental Watersheds, eds. Gregory, K.J., Walling, D.E., Institute of British Geographers, Special Pub. No.6, p.p. 33-43.

Whipkey, R.Z. (1965) Subsurface flow from forested slopes, Internat. Assoc. Sci. Hydr. Xth Annee Bull. 2, 74-85.

Whitten, D.G.A., and Brooks, J.R.V. (1972) The Penguin Dictionary of Geology, pub. Penguin Books.

Williams, A.G., Ternan, J.L., and Kent, M. (1981) Hydrochemlcal characteristics of a Dartmoor Hillslopoe, Paper presented at conference of Inter. Geophy. Union. Exeter University August 1981.

Williams, B.P.J., Downing, R.A., and Lovelock, P.E.R. (1971) Aquifer properties of the Bunter Sandstone in Nottinghamshire, England, Int. Geol. Congr. 24th, Montreal, 11, 169-176.

Wllson, H.B. (1975) The construction and operation of gravel well sources with particular reference to those of the Nene and Ouse Water Board, Journ. Instn. Wat. Eng. Sci. 29 (3), 123-138.

40 Wisler, CO., and Brater, E.F. (1959) Hydrology, 2nd ed. pub. Wiley, New York.

Wheildon, J., Francis, M.F., and Thomas-Betts, A. (1977) Investigations of the South West England Thermal Anomaly Zone, Seminar on Geoghermal Energy, Brussels, EUR5920, 1, 175-188.

Wright, C.E. (1974) Catchment Regression Models: Southern Water Authority, Technical Note No.l, Central Water Planning Unit.

Worth, R.H. (1930) Worth's Dartmoor.

Pub. David and Charles, Newton Abbot, 2nd Impression 1971.

W.W.A. (19754)

Yazor Brook Gravels Investigation; Final Report 1975, Welsh Water Authority.

- 41 Appendix 1

Drilling Logs, Narrator Catchment Dartmoor Devon

Drilling logs. Taw Marsh N.E. Dartmoor

la- DRILLING LOGS - No>^mber_J978

NARRATOR CATCHMENT

(BURRATOR) Dartmoor, Devon

KEY

Vertical scale: 1 cm = 1 metre

Horizontal scale : Not to scale

Depth below surface (in metres) LITHOLOGY DESCRIPTION On Topsoil, vegetation etc. O o o o o o

o t Sand -ooi qO O o

Clay 5H

X Sand and clay with pebbles and cobbles

104 Mud

+ + - + + + + Granitic material

Water table at time of drilling Sand and gravel Total depth lined 15 SITE 1

LITHOLOGY DESCRIPTION Ground level Topsoil

O O Brown soil with dry sandy material

O O ^ + + + + Granitic boulder 7 + + + o o o q o o G[ o o o Grey sand with Poo b o o o a yellow tinge o o <] boo ^ = 0% o o o y° P+ n Q+ + Boulder 7 + + 10 o o Yellow/pinky grey sand

+ + + + + Boulder 7

> O O

t> o o

o o o 15H °V Orange sand passing o o o o o o downwards into

Ooo yellow sand with increasing clay content O o o o d o%3 <^ o o

20

-3dr SITE 2

LITHOLOGY DESCRIPTION Ground level Dry topsoil

b «° o _ o o ° <^ Some pebbles passing P oo into yellow-grey sand

+ + + Granitic rock o o Grey dry sand

+ + + + + Granitic rock + + +

o o

Grey sands 10 H o o

passing

Into

I5i sands and gravels

Km Grey-yellow sands and gravels

20-* SITE 3

LITHOLOGY DESCRIPTION Ground level On VTTTTTA'* Topsoil Mud and gravels Cobbles (Format ion- badly caving)

5-1

SITE 4

LITHOLOGY DESCRIPTION Ground level Oi Topsoil Mud Gravel and m cobbles (Formation- badly caving) 5H

--5

LITHOLOGY DESCRIPTION Ground level Topsoil Mud gravel pebbles and cobbles

(Formation- badly caving) 5H

10 J SITE 7

LITHOLOGY DESCRIPTION On Ground level mzm^— Topsoil

o % - Followed by dry sand, oo light-grey to yellow ° s ° o o in colour o I o, Yellow-grey sand changing to a ? OO slight orange colour

OO Yellow-grey sand with slight clay OO content ,o o o O Granitic boulder 7

Pale grey clay

10 H

Granitic material possibly bedrock 7

15^

-7. SITE 9

LITHOLOGY DESCRIPTION Ground level mm Topsoil

Dry sandy material

Granitic boulder Sand

5H Granitic rock

Yellow sand

10- Yellow sandy material with some clay

Yellow clay with some sand and grit

15 SITE 10

LITHOLOGY DESCRIPTION LITHOLOGY DESCRIPTION (a) 0-1 V77777A Ground Topsoiland Topsoil and ievel pebbles dry sandy material Yellow sand Yellow sand Yellow^sand with pink staining

Yellow sandy clay 5H Yellow sandy clay

Yellow sandy clay

Yellow clay with 10 i sand and grits Possibly caving formation at base

IB-"

(a) and (b) are approximately 2 metres apart SITE 11

LITHOLOGY DESCRIPTION Ground level Topsoil

+ + Granitic boulder Yellow sands with some clay

Yellow sandy clay

Yellow sandy clay with grits and gravel

{Formation^ ^ badly caving) 10 i

IB--

•10. SITE 12

LITHOLOGY DESCRIPTION Ground level Topsoil Rock fragments

Grey-brown sand

Grey-yellow sandy clay

Yellow sandy clay

10 H Yellow sandy clay with some gravel (Formation - caving at base)

IS-" SITE 13

LITHOLOGY DESCRIPTION 0-1 7777777 Ground level Topsoil

Brown sandy material

5H

Yellow sandy clay

Granitic boulder 10

Yellow clay with some sand and grits 15 H + + Granitic boulder

20

- '2; SITE 14

LITHOLOGY DESCRIPTION LITHOLOGY DESCRIPTION (a) 0-1 V77777i Ground Topsoil Topsoil level Dry brown Dry sandy sandy material material Granite boulder

Yellow sand passing down into reddish 5H Yellow sand sandy material and yellow clay

Clay concentration higher in sand

Yellow sandy clay

Yellow-clay with some sand 10 H

Yellow clay with some sand

15-"

(a) and (b) are approximately 2 metres apart

'3, SITE 15

LITHOLOGY DESCRIPTION Ground level TTTTTA Topsoil and brown sandy material

Dry sandy material 5H

Clay nodules containing feldspar

10- Soft yellow sand with some clay

Yellow clay with some sand

15^ Taw Marsh - Borehole Logs

No.l 356 m O.D.

Depth Lithology Thickness 0.30 m Surface soil 0.3 1.52 Clay + powdered granite 1.22 2.69 Sandy clay 1.07 3.96 Stoney sand 1.37 6.40 Sandy clay 2.44 8.99 Gravel + small stones 2.59 12.45 Coarse gravel 3.46 12.8 Fine sandy gravel 0.35 14.93 Granite boulder 2.13 16.15 Coarse sands and stones 1.22 17.83 Clayey gravel 1.68 19.63 Running gravel + granite 1.80 21.18 Fairly hard granite 1.55 23.01 Hard granite 1.83 26.36 Hard granite 3.35 27.88 Grey granite 1.52

: No.lA 364.8 m O.D.

Depth Lithology Thickness 0.30 Turf 0.3 1.52 Gravel + small boulders 1.22 9.75 Gravel + large boulders 8.23 13.56 Decomposed granite 3.81 14.32 Red + grey rock granite 0.76 17.98 Grey rock granite 3.66

15a- Borehole No.2 349.5 m O.D.

Depth Llthology Thickness 0.15 m Turf 0,15 1.21 Boulders and soil 1.06 2.74 Loamy gravel 2.74 5.48 Grey sandy clay 1.83 7.31 Brown sandy clay with layers 3.2 of stones 10-51 Brown sandy clay 0.46 10.97 Dark granite boulder 6.17 17.14 Coarse sand + stones 1.19 18.33 Light red granite 1-12 19.45 Hard light red sand 2.19 21.64 Hard grey sand 0.40 22.04 Grey granite 1.58 23.62 Hard grey sand 5.48 29.10 Grey granite 1.4 30.50 Hard grey granite

16. Borehole No.3 356-8 m O.D.

Depth Hthology Thickness 0,76 m Soil 0.76 1.82 Boulders, clay + granite 2.14 3.96 Decomposed granite 1.22 5.18 Brown sandy clay 6.25 11.43 Brown sandy clay with layers 0.76 of stones 12.19 Coarse sand 7.92 20.11 Gravel stones + sand 0.76 20.87 Clayey sand 0.31 21.18 Stones 2.13 23.31 Grey sand 2.59 25.90 Loamy yellow sand 0.92 26.82 China clay, yellow + brown sand 1.83 28.65 Clayey yellow sand 2.13 30.78 Clayey dark brown gritty sand 3.56 34.34 Yellow + grey sand 1.16 35.50 Soft rock 3.97 39.47 Granite 1.67 41.14 Hard brown granite 0.61 41.75 Hard grey granite

17a- Borehole No.4 359.2 m O.D.

Depth Lithology Thickness 0.3 m Turf 0.3 m 1.21 Dark sandy gravel 0.91 3.50 Light sandy gravel 2.29 3.96 Dark clayey gravel 0.46 4.26 Granite boulder 0.30 4.41 Brown gravel 0.15 6.85 Crushed granite 2.44 11.12 Grey/brown gravel 4.27 13.56 Greyish/brown soft, gravel 2.44 15.54 Crushed granite 1.98 16.15 Grey gravel and layers of clay 0.61 and granite 23.77 Brown clayey gravel 7.62 24.77 Hard distorted mottled gravel 1.14 28.24 Hard grey sand with fine gravel 3.33 32.22 Rock spar and granite 4.98

- 18, Appendix 2

Mean Daily Rainfall Narrator Brook 1976-1979 HEAN DAILY RAINFALL 1976 HEAD WEIR NARRATOR BROOK

(to ,

FEB MR APR flAT JUL I SEP OCT MEAN DAILV RAINFALL 1977 „r-,^^.>

L HEAD WEIR NARRATOR BROOK

ceo

I NOV JUL AUG SF.P OCT JAM FEB .1AR APR MEAN DAILY RAINFALL HEAD WEIR NARRATOR BROOK

ill Uk JI__ILL I JAN [fCD !1AR APR JUL AUC ccr NOV DEC MEAN DAILY RAINFALL 1979

•ox HEAD WEIR NARRATOR BROOK —

1L_J .1 , .n Mi FEB flAR APR HAT AUf. CEP ocr Appendix 3

Streamflow at Discharge Sites and Stations Narrator Brook

Flow Duration Curve Tabulations

Mean Daily Discharge Narrator Brook (St, Cutt) 1976-1979

Discharge Data Spring 12/13 Streamflow at Discharge Sites and Discharge Stations, Narrator Brook (m^ sec"^)

Date St. 11 St. Cutt

20/10/78 0.046 0.04 0.05 0.07 0.15 7/12/78 0.156 0.091 0.082 0.064 0.21 12/1 0.170 0.244 0.304 0.276 0.22 13/2 0.33 0.42 0.44 0.39 0.31 20/2 0.19 0.26 0.27 0.26 0.24 27/2 0.19 0.218 0.217 0.218 0.24 19/3 0.27 0.192 0.359 0.336 0.22 9/A 0.17 0.189 0.233 0.209 0.22 19/A 0.16 0.158 0.176 0.182 0.22 24/4 0.18 0.240 0.247 0.201 0.23 18/5 0.12 0.108 0.090 0.123 0.20 25/5 0.13 0.123 0.098 0.135 0.20 5/6 0.17 0.159 0.202 0.174 0.22 20/6 0.13 0.128 0.175 0.137 0.20 3/7 0.09 0.090 0.105 0.116 0.18 10/7 0.07 0.075 0.113 0.096 0.17 24/7 0.06 0.064 0.080 0.062 0.16 31/7 0.20 0.102 0.145 0.173 0.24 7/8 0.09 0.083 0.086 0.079 0.18 14/8 0.45 0.216 0.447 0.322 0.38 21/8 0.12 0.143 0.112 0.124 0.20 4/9 0.10 0.103 0.093 0.147 0.19 11/9 0.09 0.084 0.098 0.099 0.18 25/9 0.09 0.070 0.115 0.107 0.18 2/10 0.09 0.066 0.072 0.078 0.18 9/10 0.09 0.076 0.083 0.106 0.18 16/10 0.07 0.067 0.074 0.090 0.17 6/11 0.19 0.157 0.213 0.256 0.24 13/11 0.23 0.154 0.284 0.105 0.26 20/11 0.22 0.134 0.272 0.249 0.25 18/12 0.38 0.331 0.411 0.432 0.34 22/1/80 0.15 0.266 0.246 0.286 0.21 5/2 0.26 0.398 0.412 0.433 0.27 Flow Duration Curve 1976

Qm^/sec Number of days % of total % Exceeded with this discharge

0.61 - 0.63 - — 0.58 - 0.60 - - 0.55 - 0.57 - - 0.52 - 0.54 - - 0.49 - 0.51 - - 0.46 - 0.48 - - 0.43 - 0.45 - - 0.40 - 0.42 2 0.54 0.54 0.37 - 0.39 1 0.27 0.81 0.34 - 0.36 5 1.36 2.17 0.31 - 0.33 8 2.18 4.35 0.28 - 0.30 11 3.00 7.35 0.25 - 0.27 16 4.37 11.72 0.22 - 0.24 32 8-74 20.46 0.19 - 0.21 12 3.27 23.73 0.16 - 0.18 13 3.55 27.28 0.13 - 0.15 11 3.00 30.28 0.10 - 0.12 26 7.10 37.38 0.07 - 0.09 30 8.19 45.57 0.04 - 0.06 58 15,84 61.41 0.01 - 0.03 141 38.52 99.83

TOTAL 366 a Flow Duration Curve 1977

Qm^/sec Number of days % of total % Exceeded with this discharge

> 0.76 1 0.27 0.27 0.73 - 0.75 1 0.27 0.54 0.70 - 0.72 0.67 - 0.69 0.64 - 0.66 0.61 - 0.63 1 0.27 0.81 0.58 - 0.60 1 0.27 1.00 0.55 - 0.57 1 0.27 1.35 0.52 - 0.54 2 0.54 1.89 0.49 - 0.51 0.46 - 0.48 0.43 - 0.45 3 0.82 2.71 0.40 - 0.42 5 1.36 4.07 0.37 - 0.39 6 1.64 5.71 0.34 - 0.36 12 3.28 8.99 0.31 - 0.33 6 1.64 10.63 0.28 - 0.30 12 3.28 13.91 0.25 - 0.27 17 4.65 18.56 0.22 - 0.24 53 14.52 33.08 0.19 - 0.21 40 10.95 44.03 0.16 - 0.18 38 10.41 54.44 0.13 - 0.15 34 9.31 63.75 0.10 - 0.12 26 7.12 70.87 0.07 - 0.09 50 13.69 84.56 0.04 - 0.06 56 15.34 99.9 0.01 — 0.03 Total 365 Flow Duration Curve 1978

Qm^/sec Number of days % of total % Exceeded with this discharge

0.76 1 0.27 0.27 0.73 - 0.75 0.70 - 0.72 1 0.27 0.54 0.67 - 0.69 1 0.27 0.81 0.64 - 0.66 1 0.27 1.08 0.61 - 0.63 1 0.27 1.35 0.58 - 0.60 2 0.54 1.89 0.55 - 0.57 1 0.27 2.16 0.52 - 0.54 2 0.54 2.7 0.49 - 0.51 3 0.81 3.51 0.46 - 0.48 4 1.09 4.6 0.43 - 0.45 4 1.09 5.69 0.40 - 0.42 5 1.36 7.05 0.37 - 0.39 6 1.63 8.68 0.34 - 0.36 6 1.63 10.31 0.31 - 0.33 10 2.73 13.04 0.28 - 0.30 9 2.45 15.49 0.25 - 0.27 18 4.91 20.4 0.22 - 0.24 21 5.73 26.13 0.19 - 0.21 30 8.19 34.32 0.16 - 0.18 17 4.64 38.96 0.13 - 0.15 23 6.28 45.24 0.10 - 0.12 20 5.46 50.7 0.07 - 0.09 64 17.48 68.18 0.04 - 0.06 116 31.69 99.87 0.01 — 0.03 Total 366 Flow Duration Curve 1979

Qm^/sec Number of days of total % Exceeded with this discharge

> 0.76 4 1.09 1.09 0.73 - 0.75 1 0.27 1.36 0.70 - 0.72 1 0.27 1.63 0.67 - 0.69 0.64 - 0.66 1 0.27 1.9 0.61 - 0.63 1 0.27 2.17 0.58 - 0.60 0.81 2.98 0.55 - 0.57 1 0.27 3.25 0.52 - 0.54 1 0.27 3.52 0.49 - 0.51 1 0.27 3.79 0.46 - 0.48 3 0.81 4.6 0.43 - 0.45 4 1.09 5.69 0.40 - 0.42 3 0.81 6.5 0.37 - 0.39 3 0.81 7.31 0.34 - 0.36 6 1.63 8.94 0.31 - 0.33 14 3.82 12.76 0.28 - 0.30 14 3.82 16.58 0.25 - 0.27 22 6.01 22.59 0.22 - 0.24 38 10.38 32.97 0.19 - 0.21 32 8.74 41.71 0.16 - 0.18 47 12.84 54.55 0.13 - 0.15 38 10.38 64.93 0.10 - 0.12 47 12.84 77.77 0.07 - 0.09 64 17.48 95.25 0.04 - 0.06 17 4.64 99.09 0.01 - 0.03 Total 366 Flow Duration Curve Tabulations Narrator Brook St. Cutt 1976-1979

Qm^/sec Number of days % of total % Exceeded with this discharge

> 0.76 4 0.27 0.27 0.73 - 0.75 0 - - 0.70 - 0.72 1 0.06 0.33 0.67 - 0.69 0 - - 0.64 - 0.66 1 0.06 0.39 0.61 - 0.63 1 0.06 0.45 0.58 - 0.60 2 0.13 0.58 0.55 - 0.57 2 0.13 0.71 0.52 - 0.54 4 0.27 0.98 0.49 - 0.51 6 0.41 1.39 0.46 - 0.48 10 0.68 2.07 0.43 - 0.45 5 0.34 2.41 0.40 - 0.42 10 0.68 3.09 0.37 - 0.39 33 2.25 5.34 0.34 0.36 30 2.05 7.39 0.31 - 0.33 54 3.69 11.08 0.28 - 0.30 98 6.69 17.77 0.25 - 0.27 182 12.33 30.21 0.22 - 0.24 166 11.34 41.55 0.19 - 0.21 214 14.62 56.17 0.16 - 0.18 267 18.25 74.42 0.13 - 0.15 357 24.40 98.82 0.10 — 0.12 16 1.09 99.91 Total 1463 Flou Duration Curves Narrator Brook St. 11 1976- •1979

Qm^/sec Number of days % of total % Exceec with this discharge

>0.76 6 0.41 0.41 0.73 - 0.75 2 0.13 0.54 0.70 - 0.72 2 0.13 0.67 0.67 - 0.69 1 0.06 0.73 0.64 - 0.66 2 0.13 0.86 0.61 - 0.63 3 0.20 1.06 0.58 - 0.60 6 0.41 1.47 0.55 - 0.57 3 0.20 1.67 0.52 - 0.54 5 0.34 2.01 0.49 - 0.51 4 0.27 2.28 0.46 - 0.48 7 0.47 2.75 0.43 - 0.45 11 0.75 3.5 0.40 - 0.42 15 1.02 4.52 0.37 - 0.39 16 1.09 5.61 0.34 - 0.36 29 1.98 7.59 0.31 - 0.33 38 2.59 10.18 0.28 - 0.30 46 3.14 13.32 0.25 - 0.27 73 4.98 18.3 0.22 - 0.24 144 9.84 28.14 0.19 - 0.21 114 7.79 35.93 0.16 - 0.18 115 7.86 43.79 0.13 - 0.15 106 7.24 51.03 0.10 - 0.12 119 8.13 59.16 0.07 - 0.09 208 14.21 73.37 0.04 - 0.06 247 16.88 90.25 0.01 - 0.03 141 9.63 99.88 Total 1463 hEAN DAILY DISCHARGE: 19' NARRATOR BROOK

UJ

(J

:3 o

a: I

SFP NCV DEC nEAN DAILY DISCHARGE 1977 NARRATOR BR02:<

I

\ MEAN DAlur DISCHARGE 1 9/(! NARRATOR BRDDK MEAN DAILY OISCMARGE: 1979 NARRATOR BROCK (Litres sec~^)

Date Q

5/3/79 224.9 12/3 - 19/3 284.7 26/3 317.7 9/4 - 23/4 - 30/4 93.2 7/5 63.7 14/5 93.2 21/5 93.2 28/5 93.2 4/6 173.4 11/6 173.4 18/6 150.5 25/6 93.2 2/7 63.7 9/7 110.5 16/7 40.6 23/7 0.3 30/7 0.3 6/8 0.15 13/8 0.20 20/8 110.5 3/9 51.4 10/9 40.6 17/9 51.4 24/9 0.30 1/10 0.30 8/10 51.4 15/10 0.30 22/10 0.30 29/10 51.4 5/11 110.5 12/11 224.9 19/11 224.9 26/11 224.9 3/12 173.4 10/12 224.9 17/12 430.2 24/12 352.9 31/12 430.2 7/1/80 352.9 14/1 284.7 21/1 224.9 28/1 173.4 4/2 253.7 11/2 430.2 18/2 317.7 25/2 253.7 Appendix 4

Analysis of Variance Tables to test the significance of the Regressions chosen by the Step-Wise Regression Procedure ANOVA (8 variables) Narrator Brook

Critical F value at 0.01 significance level from statistical Tables = 3.12

Source of Degrees of Sum of Mean Calculated Decision Variation Freedom Squares Square F-value

Well 1 Regression 7 0.0202 0.00288 9.53 Reject Ho. Residual 39 0.0118 0.000302 (Regression is signifi• cant)

Well 2 Regression 7 0,0974 0.0139 0.572 Accept Ho. Residual 39 0.948 0.0243 (Regression Is not significant)

Well 3 Regression 7 2.6516 0.3788 5.255 Reject Ho. Residual 39 2.8112 0.0720 (Regression is signifi• cant)

Well 4 Regression 7 0.068 0.009 13.13 Reject Ho- Residual 39 0.029 0.0007

Well 6 Regression 7 0.8061 0.115 16.56 Reject Ho. Residual 39 0.2711 0.0069

Well 7 Regression 7 0.4141 0.0591 2.790 Accept Ho- Residual 39 0.8268 0.0212 (Regresslon is not significant)

Well 9 Regression 7 0.8753 0.1250 4.971 Reject Ho. Residual 39 0.9306 0.0251

Well 10 Regression 7 0.0702 0.0100 13.348 Reject Ho. Residual 39 0.0293 0.0007

38< ANOVA (continued)

Source of Degrees of Sum of Mean Calculated Decision Variation Freedom Squares Square F-value

Well 11 Regression 7 1.1102 0.1586 6.95 Reject Ho. Residual 39 0.8893 0.0228

Well 12 Regression 7 0.6892 0.0984 7.036 Reject Ho. Residual 39 0.5457 0.0139

Well 13 Regression 7 0.1035 0.0147 35.81 Reject Ho. Residual 39 0.0161 0.0004

Well 14 Regression 7 0.1781 0.0254 12.32 Reject Ho. Residual 39 0.0805 0.0020

Well 15 Regression 7 0.6158 0.0879 10.95 Reject Ho. Residual 39 0.3133 0.0080