An assessment of recent changes in catchment sediment sources and sinks, central ,

Australia

Andrew Owen Hughes

2008

A thesis submitted in fulfillment of the requirements

for the degree of Doctor of Philosophy

School of Physical, Environmental and Mathematical Sciences (Geography)

Australian Defence Force Academy

University of New South Wales

i

ii ABSTRACT

Spatial and temporal information on catchment sediment sources and sinks can provide an improved understanding of catchment response to human-induced disturbances. This is essential for the implementation of well-targeted catchment-management decisions.

This thesis investigates the nature and timing of catchment response to human activities by examining changes in sediment sources and sinks in a dry-tropical subcatchment of the Great Barrier Reef (GBR) catchment area, in northeastern Australia.

Changes in catchment sediment sources, both in terms of spatial provenance and erosion type, are determined using sediment tracing techniques. Results indicate that changes in sediment source contributions over the last 250 years can be linked directly to changes in catchment land use. Sheetwash and rill erosion from cultivated land (40–60%) and channel erosion from grazed areas (30-80%) currently contribute most sediment to the river system. Channel erosion, on a basin-wide scale, appears to be more important than previously considered in this region of Australia.

Optically stimulated luminescence and 137Cs dating are used to determine pre-and post-

European settlement (ca. 1850) alluvial sedimentation rates. The limitations of using

137Cs as a floodplain sediment dating tool in a low fallout environment, dominated by sediment derived from channel and cultivation sources, are identified. Low magnitude increases in post-disturbance floodplain sedimentation rates (3 to 4 times) are attributed to the naturally high sediment loads in the dry-tropics. These low increases suggest that previous predictions which reflect order of magnitude increases in post-disturbance sediment yields are likely to be overestimates. In-channel bench deposits, formed since

European settlement, are common features that appear to be important stores of recently eroded material.

i The spatially distributed erosion/sediment yield model SedNet is applied, both with generic input parameters and locally-derived data. Outputs are evaluated against available empirically-derived data. The results suggest that previous model estimates using generic input parameters overestimate post-disturbance and underestimate pre- disturbance sediment yields, exaggerating the impact of European catchment disturbance. This is likely to have important implications for both local-scale and catchment-wide management scenarios in the GBR region. Suggestions for future study and the collection of important empirical data to enable more accurate model performance are made.

ii DECLARATION OF ORIGINALITY

I hereby declare that this is my own work and to the best of my knowledge it contains no material previously published or written by another person, nor material which to a substantial extent has been accepted for the award of any other degree or diploma at

UNSW or any other educational institutions, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual property of this thesis is the product of my own work, except to the extent that assistance from others in the project’s design and conception or in style, presentation and linguistic expression is acknowledged.

Signed

Date 14 December 2008

iii ACKNOWLEDGEMENTS

This research was completed with the financial support from the Australian Research Council (grant number: DP0449886), CSIRO Water for Healthy Countries Flagship, the Marine and Tropical Sciences Research Facility, and the Australian Rivers Institute.

I would like to thank my supervisors Associate Professor Jacky Croke (UNSW@ADFA) and Professor Jon Olley (Griffith University) for their support. Jacky, thank you for allowing me to pursue my interests and for providing advice and feedback above and beyond the call of duty. Jon, thanks for your valuable input, particularly with the Monte Carlo mixing model and the 137Cs advection-diffusion modelling. Thanks also to Dr Ingrid Takken (UNSW@ADFA) for stepping into the breach and securing further funding at a crucial time.

Lucy McKergow (NIWA) provided assistance with field and laboratory work and provided many constructive comments on aspects of this thesis. Tim Pietsch (CSIRO Land and Water) provided field work assistance, access to the CSIRO Land and Water OSL laboratory and help with single-grain OSL data interpretation. Ian Webster (CSIRO Land and Water) provided advice that ensured the 137Cs advection-diffusion modelling functioned correctly. Heiko Timmers (UNSW@ADFA) and Chris Leslie (CSIRO Land and Water) provided technical advice on gamma spectrometry. Ken McMillan (CSIRO Land and Water) prepared sediment samples for OSL analysis.

Many thanks to Chris Thompson (UNSW@ADFA) whose unsurpassed field skills were critical to the completion of this research. Thanks also Eugene Wallensky (ANU) and David Purvis-Smith (formerly of UNSW@ADFA) for field work assistance.

Maurice Conway (Queensland Department Primary Industries) generously provided the use of a vehicle-mounted soil coring rig. Cameron Dougall, George Bourne, Chris Carroll, and Bob Packett (all of the Queensland Department of Natural Resources and Water) provided spatial and hydrological data for Theresa Creek and comments on drafts.

Thanks to the PEMS staff at UNSW@ADFA, especially Brian Lees, Peter Palmer, Ali Arezi and Ray Lawton.

iv Thanks to Kathryn Amos (University of Adelaide), Rebecca Bartley (CSIRO Land and Water), Gary Caitcheon (CSIRO Land and Water), Peter Hairsine (CSIRO Land and Water), Gary Hancock (CSIRO Land and Water), Graham McBride (NIWA), Ian Prosser (CSIRO Water for Healthy Countries), and Peter Wallbrink (EWater) for constructive discussions and technical assistance.

Thanks to the landowners within Theresa Creek. Special thanks to Robin Cross, David and Rose Staal, Jim Purvis, and Greg Purvis for property access and sharing their local knowledge.

Thanks to my family and friends for love and support and for keeping me sane during the rollercoaster known as a PhD. Special thanks to Tanya, Geoff, Karen, Ian, Janelle and Hamish for housing me while in Canberra. Thanks to my son Cadel for his infectious smile and distracting me from my PhD worries. Finally, thanks to Lucy for your love, support, patience and strength, this thesis is dedicated to you

v

.

vi

“Today's scientists have substituted mathematics for experiments, and they wander off through equation after equation, and eventually build a structure which has no relation to reality.”

Nikola Tesla (1856 – 1943)

vii TABLE OF CONTENTS

CHAPTER 1- Introduction...... 1 1.1. Introduction...... 2 1.2. Catchment sediment sources ...... 4 1.3. Catchment sediment sinks...... 5 1.4. Human-induced catchment disturbance in Australia ...... 8 1.5. Research aims ...... 12 1.6. Thesis structure...... 12

CHAPTER 2 - Sediment source changes over the last 250 years in a dry-tropical catchment, , Australia...... 15 2.1. Introduction...... 16 2.2. Study site ...... 19 2.3. Methods ...... 23 2.3.1. Sediment sources...... 23 2.3.2. Temporal changes in sediment sources...... 30 2.4. Results ...... 33 2.4.1. Geochemical differentiation of geological source areas ...... 33 2.4.2. Fallout radionuclide differentiation of sediment sources ...... 34 2.4.3. Sources of fine river bed sediment...... 37 2.4.4. Floodplain sediment ages and accretion rates ...... 41 2.4.5. The spatial provenance of floodplain deposits over the last 250 years.42 2.5. Discussion...... 44 2.6. Conclusions...... 50

CHAPTER 3 - Determining floodplain sedimentation rates using 137Cs in a low fallout environment dominated by channel- and cultivation-derived sediment inputs, central Queensland, Australia ...... 53 3.1. Introduction...... 54 3.2. Study Site...... 55 3.3. Using 137Cs to determine rates of floodplain sedimentation...... 57 3.3.1. Total 137Cs inventory ...... 58 3.3.2. Caesium-137 depth profiles ...... 58 3.4. Sample collection and laboratory analysis ...... 60 3.5. Results ...... 62 3.5.1. Expected total 137Cs inventories...... 62 3.5.2. Reference cores ...... 63

viii 3.5.3. Floodplain cores ...... 67 3.6. Discussion...... 71 3.7. Conclusions...... 72

CHAPTER 4 - Changes in the rates of floodplain and in-channel bench accretion in response to catchment disturbance, central Queensland, Australia...... 73 4.1. Introduction ...... 74 4.2. Study Site...... 76 4.3. Methods ...... 79 4.3.1. Caesium-137 depth profiles...... 79 4.3.2. Optical dating...... 80 4.4. Results...... 82 4.4.1. Floodplain and in-channel bench sediment dating ...... 82 4.4.2. Estimated floodplain accretion rates ...... 86 4.4.3. In-channel bench accretion rates...... 89 4.5. Discussion...... 90 4.5.1. Patterns of floodplain sedimentation...... 90 4.5.2. Increased rates of floodplain accretion as an indicator of the impact of catchment disturbance on river sediment flux...... 93 4.5.3. The significance of in-channel benches as stores of post-disturbance generated sediment ...... 98 4.6. Conclusions...... 100

CHAPTER 5 - Validation of a spatially distributed erosion/sediment yield model (SedNet) with empirically-derived data ...... 103 5.1. Introduction ...... 104 5.2. SedNet background...... 106 5.3. Study area...... 108 5.4. Methods ...... 110 5.4.1. SedNet model...... 110 5.4.2. Model input data...... 116 5.4.3. Model validation data...... 120 5.5. Model application ...... 127 5.5.1. Post-disturbance models ...... 127 5.5.2. Pre-disturbance models...... 129 5.6. Results and discussion...... 132 5.6.1. Evaluation of modelled sediment loads...... 132 5.6.2. Evaluation of modelled estimates of sediment source contributions... 134

ix 5.6.3. Floodplain accretion rates and prediction of pre-disturbance catchment sediment yield ...... 137 5.6.4. Future model development ...... 138 5.7. Conclusions...... 141

CHAPTER 6 - Summary and Conclusions...... 145 6.1. Summary and conclusions...... 146 6.2. Future research...... 152

REFERENCES……………………………………………………………………………… 155

APPENDIX 1 – Monte Carlo model mean difference data…………………………. 183

APPENDIX 2 – Sediment source geochemistry data…………..…………………… 185

APPENDIX 3 – Summary of over-bankfull discharge data for gauging station

130210A……………………………………………...……………………………………… 187

APPENDIX 4 – 137Cs data for floodplain and reference site cores ………….…… 189

APPENDIX 5 – Channel cross-sections…………………………………………….… 191

APPENDIX 6 – Radial plots for single-grain OSL data……...……………………… 195

APPENDIX 7 – Site data for OSL samples…………………………………………….. 203

x LIST OF FIGURES

Figure 1.1. Great Barrier Reef catchments, river networks, and mean annual rainfall. Also illustrated is the location of the study catchment, Theresa Creek...... 10

Figure 2.1. Theresa Creek catchment, showing the location of the sediment sampling sites and gauging stations referred to in the text...... 19

Figure 2.2. Annual rainfall from Clermont Post Office climate station (22.83˚S., 147.64˚E.) for the period 1889 - 2006 (BoM, 2007). Solid line is a 5-year running average. Dashed horizontal line is the mean annual rainfall for the entire 118 year period (649 mm y-1)...... 21

Figure 2.3. (A) Geology of Theresa Creek showing the principle geology classes and (B) simplified land use map of Theresa Creek (Calvert et al., 2000)...... 22

Figure 2.4. Concentrations of nine elements (as selected by stepwise linear discriminant analysis) in the sediment samples collected from the four principal source areas, floodplain core, and river bed. Major elements are plotted against SiO2. Trace elements are plotted against ThO2...... 35

Figure 2.5. A contour bank typical of those found throughout the intensively cultivated basaltic-derived Vertisol areas...... 37

Figure 2.6. Sediment source contributions to each river bed sampling site. Each sampling site has two associated pie graphs, the left graph illustrates the predicted relative contribution (%) from each rock type source area, and the right graph indicates the predicted relative contribution (%) from each erosion type...... 39

Figure 2.7. An example of severe gully erosion from the metasediment-underlain upper Sandy Creek area...... 40

Figure 2.8. The relative contributions from each of the major geological source areas to the floodplain core (0 - 60 cm) from lower Theresa Creek. The 1954 date was determined by analysis of a 137Cs depth profile for the site. All other dates were obtained by OSL dating of single grains of quartz...... 43

Figure 3.1. Deposition of nuclear weapon testing-derived 137Cs in the northern and southern hemispheres from 1954 - 1989 (source: UNSCEAR, 2000)...... 54

Figure 3.2. Theresa Creek catchment and tributaries showing the floodplain and reference site sampling sites, local climate stations and gauging station 130210A. .... 56

Figure 3.3. The expected 137CS fallout accumulation curves for the 10 - 20° and 20 - 30° south latitudinal bands based on UNSCEAR (2000) data, and that estimated for Theresa Creek catchment (23° S) using the UNSCEAR (2000) and measured soil core inventories...... 63

Figure 3.4. Caesium-137 depth profiles for the three reference core sites: A) REF1, B) REF2 and C) REF3. The error bars represent measurement precision (95% confidence limits)...... 64

xi Figure 3.5. Total 137Cs inventories for the reference (REF1, REF2, REF3) and floodplain cores (CC-F, UTC-F, LTC-F) collected from within and around Theresa Creek. REF average is the mean average of the reference site total 137Cs inventories. The error bars for each core represent measurement precision (95% confidence limits). The error bars for the mean reference core value represent one standard error...... 66

Figure 3.6. Annual rainfall (1889 – 2006) for climate stations within, and local to, the Theresa Creek catchment. Locations of climate stations are indicated on Figure 3.2. 66

Figure 3.7. Caesium-137 depth profiles and total inventories for A) Capella Creek (CC- F), B) upper Theresa Creek (UTC-F) and C) lower Theresa Creek (LTC-F) floodplain sampling locations. The error bars represent measurement precision (95% confidence limits). The smoothed curves indicate 137Cs concentrations as determined by the advection-diffusion model...... 68

Figure 4.1. Theresa Creek catchment showing the locations of the floodplain core sites and gauging stations...... 76

Figure 4.2. Example of an in-channel bench on the main tributary of Theresa Creek. 78

Figure 4.3. Caesium-137 depth profiles for (A) upper Theresa Creek (UTC-F), (B) lower Theresa Creek (LTC-F) and (C) Capella Creek (CC-F) floodplain sampling locations. The error bars represent measurement precision (95% confidence limits)...... 84

Figure 4.4. Age-depth relationships for (A) upper Theresa Creek (UTC-F), (B) lower Theresa Creek (LTC-F), and (C) Capella Creek (CC-F) floodplain sampling locations. The vertical dashed line represents ~ 150 y BP; the time of first European settlement within the catchment...... 88

Figure 4.5. Sand wave at the upper Theresa Creek (UTC) site. Note the fence post in the centre of the photograph for scale. Photograph was taken looking upstream...... 90

Figure 4.6. Photograph looking downstream at the upper Theresa Creek (UTC) site. Photograph was taken standing on the in-channel bench. Note the well-established vegetation indicating channel stability...... 97

Figure 5.1. Theresa Creek catchment, showing the location of river bed sediment sampling sites and gauging stations referred to in the text...... 109

Figure 5.2. A schematic river network showing links, Shreve magnitude of each link and the internal catchment area (shaded) of a magnitude one and magnitude four link (from Prosser et al., 2001a)...... 110

Figure 5.3. Conceptual diagram of sediment budget for a river link and its associated subcatchment (light shade). Solid black arrows represent link budget input terms while open arrows represent link budget loss terms. For the illustrated river link storage loss is in the form of floodplain (grey shade) deposition; however, this can also be reservoir deposition where a link ends in a reservoir (from McKergow et al., 2005)...... 111

Figure 5.4. Suspended sediment rating curves for gauging stations (A) 130207A (Sandy Creek@Clermont) and (B) 130210A (Theresa Creek@Valeria). The data (47 samples) for 130207A were collected between 1974 and 2001. The data (38 samples) for 130210A were collected between 1973 and 2006...... 121

xii Figure 5.5. Flow duration plots for the Sandy Creek@Clermont (130207A; 1965 - 2008) and Theresa Creek@Valeria (130210A; 1971 – 2008) gauging stations. The horizontal dashed lines indicate the maximum flow for the QDNRW suspended sediment sampling datasets (47 m3 s-1 for 130207A and 517 m3 s-1 for 130210A)...... 123

Figure 5.6. (A) Sediment source contributions to each river bed sampling site as determined by radionuclide tracing, (B) land use (source: Calvert et al., 2000), and (C) gully density (source: Trevithick et al., 2008) for Theresa Creek...... 125

Figure 5.7. Post-disturbance sediment yields for gauging stations 130210A (Theresa Creek@Valeria) and 130207A (Sandy Creek@Clermont) calculated from historical suspended sediment sampling records and predicted by parameterisations (A - D) of the SedNet model...... 133

Figure 5.8. Relative contribution of channel, cultivation and non-cultivations sources to channel sampling sites (A) SC1 (= SC2 and SC3), (B) TC1, and (C) TC2 as estimated by radionuclide tracing and four parameterisations (A - D) of the SedNet model...... 136

Figure A5.1. Channel cross-section at the upper Theresa Creek site (UTC) showing the sediment core positions………………………………………………………………. 192

Figure A5.2. Channel cross-section at the lower Theresa Creek site (LTC) showing the sediment core positions…………………………………………………………………… 192

Figure A5.3. Channel cross-section at the Capella Creek site (CC) showing the sediment core positions…………………………………………………………………… 193

Figure A6.1. Radial plot of single-grain De estimates (Gy) for sample LTC-B-38…... 196

FigureA6.2. Radial plot of single-grain De estimates (Gy) for sample LTC-B-95…... 196

Figure A6.3. Radial plot of single-grain De estimates (Gy) for sample LTC-F-22….. 197

Figure A6.4. Radial plot of single-grain De estimates (Gy) for sample LTC-F-35..… 197

Figure A6.5. Radial plot of single-grain De estimates (Gy) for sample LTC-F-50….. 198

Figure A6.6. Radial plot of single-grain De estimates (Gy) for sample LTC-F-95….. 198

Figure A6.7. Radial plot of single-grain De estimates (Gy) for sample UTC-B-40…. 199

Figure A6.8. Radial plot of single-grain De estimates (Gy) for sample UTC-B-90….. 199

Figure A6.9. Radial plot of single-grain De estimates (Gy) for sample CC-F-20…… 200

Figure A6.10. Radial plot of single-grain De estimates (Gy) for sample CC-F-40…. 200

Figure A6.11. Radial plot of single-grain De estimates (Gy) for sample CC-F-60…. 201

Figure A6.12. Radial plot of single-grain De estimates (Gy) for sample UTC-F-15... 201

Figure A6.13. Radial plot of single-grain De estimates (Gy) for sample UTC-F-60.. 202

Figure A6.14. Radial plot of single-grain De estimates (Gy) for sample UTC-F-100 .202

xiii LIST OF TABLES

Table 2.1. Geochemical properties that provide the best differentiation of geological source areas by stepwise linear discriminant analysis...... 34

137 210 Table 2.2. Mean Cs and Pbex concentrations of the < 10 µm fraction of sediment from uncultivated pasture, cultivated land, and channels (errors are one standard deviation)...... 36

Table 2.3. Dating results and accretion rates for the floodplain sediment core dated from lower Theresa Creek (LTC-F)...... 41

Table 2.4. Qualitative assessment of the relative importance of each of the source types by geological source area as determined by the geochemical and radionuclide tracing results (*** = high, ** = moderate, * = low)...... 46

Table 3.1. Results of the total 137Cs inventory and 137Cs depth profile methods for determining accretion/erosion for the three sampled floodplain cores. The mean reference site inventory is 246 ± 12 Bq m-2 (n = 3; uncertainty is equivalent to one standard error)...... 67

Table 3.2. Total accretion and accretion rates as determined by the advection-diffusion model for the three floodplain cores. The accretion rates are for the period between the year of the first known overbank discharge event (1954) after the first year of significant fallout (1951) to 2006, the year of core extraction...... 71

Table 4.1. Characteristics of the floodplain/in-channel bench sampling sites...... 77

Table 4.2. Burial ages for the sediment obtained from floodplain and bench locations at the three sampling locations. Also indicated is whether or not 137Cs was present in the material surrounding the dated quartz grains. The detection of 137Cs indicates that sediment was likely to have been deposited since ca. 1954...... 85

Table 4.3. Rates of pre- and post-disturbance alluvial accretion from previous selected studies...... 95

Table 5.1. Range of gully parameter values used by previous applications of SedNet. See text for description of terms...... 113

Table 5.2. Sources of spatial and hydrological data used as inputs to the SedNet model for Theresa Creek...... 117

Table 5.3. Regionalised hydrological parameters calculated for use in SedNet...... 119

Table 5.4. Mean suspended sediment load estimates for gauging stations 130207A (Sandy Creek@Clermont) and 130210A (Theresa Creek@Valeria)...... 123

Table 5.5. Floodplain accretion rates and increases since catchment disturbance commenced (ca. 1850), as determined by OSL dating and 137Cs depth profiles for the three floodplain sampling sites...... 126

Table 5.6. Summary of the post-disturbance SedNet model parameterisations...... 127

Table 5.7. Summary of pre-disturbance SedNet model parameterisations...... 131

xiv Table 5.8. Increases in floodplain accretion rates between each of the pre-disturbance model parameterisations and post-disturbance model D. Also indicated is the catchment sediment yield predicted by each of the pre-disturbance models...... 138

Table A1.1. Mean difference between Monte Carlo model estimates and measured geochemical and radionuclide tracing data ………….…………………… 184

Table A2.1. Basic descriptive statistics for the sediment source geochemistry data 186

Table A3.1. Summary of over-bankfull discharge data for gauging station 130210A …………………………………………………………………………………… 188

Table A4.1. Caesium-137 data for floodplain and reference core sites …………… 190

Table A7.1. Site data for OSL samples………………………………………………… 204

xv SYMBOLS AND ABBREVIATONS

A link catchment area (SedNet) Af floodplain area (SedNet) AD Anno Domini ANOVA analysis of variance B link bank erosion contribution (SedNet) Bq becquerel BP before present C suspended sediment concentration Ca upslope contributing area CSIRO Commonwealth Scientific and Industrial Research Organisation D link gully density (SedNet) Db burial dose De equivalent dose Dr dose rate E coefficient of efficiency F SedNet link floodplain width GBR Great Barrier Reef GIS geographical information system GPS Global Positioning System Gy gray H link bank height (SedNet) HSDR hillslope deliver ratio L link river length (SedNet) LDA linear discriminant analysis NLWRA National Land and Water Resources Audit OSL optical stimulated luminescence ρs sediment bulk density ρw density of water Pfs fine sediment proportion (SedNet) PET potential evaporation PR link proportion of intact riparian vegetation (SedNet) Q stream discharge Qf floodplain discharge QBF Bank full discharge QMA Mean annual flow QMO Median overbank discharge RF mean annual rainfall ROC mean annual runoff coefficient RUSLE revised universal soil loss equation S link channel slope (SedNet) SAR single-aliquot regenerative-dose τ gully age v settling velocity XRF x-ray fluorescence y year

xvi CHAPTER 1

Introduction

Chapter 1 - Introduction

1.1. Introduction

Assessment of anthropogenic activities and their impacts on river systems is a common research focus in fluvial geomorphology (e.g., Douglas, 1967; Knox, 1977; Brooks and

Brierley, 1997; Walling, 1999; Syvitski et al., 2005). Increased erosion and delivery of sediment to rivers are arguably the most significant consequences of human-induced catchment disturbance. The effects are experienced both within rivers (Brierley et al.,

1999; Bramley and Roth, 2002) and in downstream environments, such as lakes and coastal waters (McLaughlin et al., 2003; Leahy et al., 2005; Restrepo et al., 2006). It has been demonstrated that human disturbances, principally in the form of overgrazing, deforestation, and intensive agriculture, have increased erosion rates and delivery of sediment to many rivers (Wasson et al., 1998; Knox, 2001; Pasternack et al., 2001).

Much emphasis has been placed on quantifying the impact of human activities on erosion rates (e.g., Walling and He, 1999; Chaplot and Le Bissonnais, 2000) but there remains a poor understanding of the relative importance of sediment sources (e.g., sheetwash and rill, channel or mass movement), and the extent to which they reflect human-induced land disturbance (Walling, 1983; Collins et al., 1997; Syvitski et al.,

2005).

Furthermore, the impact of human disturbance on end-of-catchment sediment yield appears to be variable, with large increases estimated for some rivers (e.g., Wasson et al., 1998; McCulloch et al., 2003) while others show reductions or limited change (e.g.,

Knox, 2001; Walling and Fang, 2003). In most catchments, long-term measurements of sediment yield, such as those derived from stream monitoring, are non-existent. The detection of changes in catchment sediment yield is, therefore, often inferred by other means, such as the sedimentary record contained within catchment sediment sinks (e.g.,

2 Chapter 1 - Introduction floodplains, lakes and reservoirs). Catchment sediment sinks contain material eroded primarily from the upper catchment and are important sources of information on changes in sediment sources and flux as a result of anthropogenic activities (Oldfield and Clark, 1990; Owens et al., 1999; Hudson, 2003).

Because of the long-history of large-scale human impact in many parts of the world

(e.g., Xu, 2003; Houben et al., 2006) it is often difficult to determine the precise timing and scale of catchment changes from pre-disturbance through to present-day conditions.

The recent nature of large-scale catchment disturbance in Australia (since the early nineteenth century) provides a suitable timeframe for the quantification of the impacts of large-scale agricultural development. However, because of the scale of many

Australian catchments, there has been a bias towards the use of large-scale erosion/sediment yield models to assess the impact of human-induced catchment disturbance (e.g., Simons et al., 1996; Viney and Sivapalan, 1999; Prosser et al., 2001;

Vertessy et al., 2001). While such models are useful tools, they are often applied using poor resolution empirical data with limited or no corroboration of their predictions

(Lane et al., 1997; Refsgaard, 1997; Kirchner, 2006). Important catchment management decisions are being made on the basis of these models, and there is an urgent need to invest more effort in validating their outputs.

This imbalance provides suitable motivation for this study. This thesis seeks to improve our understanding of catchment responses to anthropogenic activities by examining changes in sediment sources and sinks. The overall aim of this study is to improve both our understanding, and data availability, on spatial and temporal patterns of erosion and deposition in a dry-tropical catchment in northeastern Australia.

3 Chapter 1 - Introduction

1.2. Catchment sediment sources

Human-induced disturbances, such as intensification of agricultural activities and urbanisation have been linked to the degraded water quality of rivers and receiving environments, such as lakes, estuaries and coastal zones (Bramley and Roth, 2002;

Ahearn et al., 2005; Fabricius, 2005; Hunter and Walton, 2008). Sediment and sediment-bound nutrients and contaminants, such as heavy metals and agricultural chemicals, have been associated with many water quality issues (e.g., Cooper, 1993;

Lenat and Crawford, 1994; Kennish, 2002). If the generation and delivery of sediment and associated contaminants from catchments is to be effectively reduced, then more information is required on the major sources of sediment, both in terms of spatial provenance and erosion type (Collins et al., 1997; Walling, 2005; Douglas et al., 2007).

Sound information on catchment sediment sources assists in the efficient allocation of, often scarce, catchment rehabilitation resources (Matisoff et al., 2002; Wasson et al.,

2002). It is also necessary to construct catchment sediment budgets and assists in the interpretation and modelling of suspended sediment yields (Collins and Walling, 2004).

A number of approaches are available to determine the sources of catchment sediment

(Loughran and Campbell, 1995; Collins and Walling, 2004). Arguably, sediment source tracing (or “sediment fingerprinting”) is one of the most versatile methods available because of its ability to account for both sediment mobilisation and delivery.

It is also useful for determining longer-term trends in sources (Collins et al., 1997;

Owens et al., 1999; Collins and Walling, 2004). Briefly, sediment source tracing involves determining the relative importance of sediment sources, in terms of spatial provenance and/or erosion process, by comparing the properties of suspended (or deposited) sediment to sediment acquired from major sources areas (Collins and

4 Chapter 1 - Introduction

Walling, 2004; Walling, 2005). The source tracing approach has been applied using a wide range of sediment properties, including geochemistry (Collins et al., 1997), mineral-magnetism (Foster et al., 1998), radionuclides (Olley et al., 1993) and grain size

(Kurashige and Fusejima, 1997).

Applying source tracing techniques to sediment obtained from flood events or deposited in river channels provides information on current or recent sediment sources (Russell et al., 2001; Collins and Walling, 2002; Dirszowsky, 2004; Wallbrink, 2004). However, tracing of floodplain or lake sediment, when utilised in conjunction with a chronology provided by sediment dating techniques (e.g., Collins et al., 1997; Owens and Walling,

2002; Wasson et al., 2002), extends the versatility of the approach by providing a temporal dimension. To date, most sediment tracing studies sourcing floodplain sediment have used caesium-137 (137Cs) and lead-210 (210Pb) to provide core chronologies and are, therefore, restricted to a timeframe of up to ~ 100 years. There is further potential to extend the timescale over which this method can be applied by the use of more precise and temporally extensive dating techniques. This study combines sediment tracing techniques with luminescence dating to obtain a detailed temporal representation of changes in catchment sediment sources.

1.3. Catchment sediment sinks

The sedimentary record contained within catchment sediment sinks is an important source of information on changes in catchment sediment flux (Owens et al., 1999).

Floodplains, in particular, are one of the most important sinks within catchments, with up to 50% of a catchment’s annual sediment load being stored within them (Trimble,

1983; Phillips, 1991). Many studies have utilised the information contained within floodplain sediment to determine catchment response to changes in climate and land use

5 Chapter 1 - Introduction over varying time scales (e.g., Ritter et al., 1973; Nanson et al., 1995; Nott et al., 2002;

Asselman et al., 2003; May, 2003). Despite this, there are few fine-resolution Holocene floodplain chronologies (e.g., Rustomji and Pietsch, 2007) that can be used to assess the response of catchments to human disturbance (Brooks and Brierley, 2004). The work of

Rustomji and Pietsch (2007) used over 80 optically stimulated luminescence dates to construct a detailed record of the impact of human catchment disturbance on floodplain and in-channel bench sedimentation in the Lake Burragong catchment in southeastern

Australia.”

To a large degree, the lack of fine-resolution floodplain chronology data can be attributed to the difficulty in obtaining precise depositional age data for about the last

200 years, the period catchments have been under most pressure from anthropogenic activities. Radiocarbon dating, used widely in the dating of floodplain sediment (e.g.,

Langedal, 1997; Makaske et al., 2002; Leigh et al., 2004; Keesstra et al., 2005), is unable to provide precise dates for young deposits (Lian and Roberts, 2006; Pierson,

2007). Obtaining dateable material suitable for radiocarbon dating can also be problematic (Rittenour et al., 2003; Rowland et al., 2005). Other techniques, such as heavy metal profiling (Lecce and Pavlowsky, 2001; Middelkoop, 2002) and inference from texture variations (Brooks and Brierley, 1997; Gomez et al., 1998; Carson, 2006) have been utilised to good effect, but they are dependent on specific conditions, such as the existence of historical mining activities and readily distinguishable floodplain stratigraphy.

Fallout radionuclides (i.e., 137Cs, 210Pb) have been widely used to determine recent rates of floodplain sedimentation and provide satisfactory results in many locations (e.g.,

Goodbred and Kuehl, 1998; Walling et al., 2003; Pavlovic et al., 2005; Yeager et al.,

6 Chapter 1 - Introduction

2005). Caesium-137 and 210Pb are, however, limited to providing sedimentation rates for the last ~ 50 years and ~ 100 years, respectively. Low or variable fallout rates also limit the applicability of fallout radionuclides as dating tools in some locations (Bishop et al., 1991; Chappell, 1999). As a result, most applications have been carried out in regions characterised by uniform and high fallout rates (e.g., Walling and He, 1997;

Goodbred and Kuehl, 1998; Owens et al., 1999; Ritchie et al., 2004). This study tests the applicability of 137Cs dating to a dry, low latitude, southern hemisphere location, an environment the technique has seldom been applied in. The only known published study to use 137Cs to determine floodplain sedimentation rates in a dry, low latitude, southern hemisphere location (i.e., Amos et al., 2009) found that, depending on the method used, low fallout rates made determining reliable sedimentation rates difficult.

Luminescence dating has the greatest potential to provide detailed floodplain chronologies in a wide range of environments (Lian and Roberts, 2006). In particular, optically stimulated luminescence (OSL) dating has been widely applied to fluvial sediments (Wallinga, 2002; Zhang et al., 2003). Recent developments in the application of the OSL technique to single grains of quartz has shown that accurate ages can be obtained for very young sediments (Olley et al., 2004; Rowland et al., 2005; Rustomji and Pietsch, 2007). This, together with the ability to easily obtain samples at multiple depths through a floodplain profile, makes it a versatile technique that can be used to construct fine-resolution chronologies. This study uses the recently refined method of

OSL dating of single grains of quartz (Galbraith et al., 1999; Olley et al., 2004), in combination with 137Cs dating, to obtain detailed floodplain core chronologies. These chronologies provide information on changes in catchment sediment flux from before significant human catchment disturbance through to the present.

7 Chapter 1 - Introduction

While floodplains are probably the most important sink of fine sediment within most lowland rivers (Walling et al., 1996), the role of in-channel stores of sediment has largely been ignored (Collins and Walling, 2007). For example, extensive in-channel lateral deposits of sediment, known as in-channel benches (cf. Taylor and Woodyer,

1978) have been reported in many rivers, although they appear to be particularly common in Australian catchments (e.g., Erskine and Livingstone, 1999; Carthew and

Drysdale, 2003; Kemp, 2004; Sheldon and Thoms, 2006; Rustomji and Pietsch, 2007).

In Australia, they have been described as post-disturbance landforms that formed in response to the massive influx of gully-derived sediment delivered to many rivers since

European settlement (Carthew and Drysdale, 2003; Rustomji and Pietsch, 2007). In- channel benches are considered transient features that are reworked during large flood events (Erskine and Livingstone, 1999). However, recent OSL dating of benches in southeastern Australia demonstrated that they may be stable features (Rustomji and

Pietsch, 2007). In-channel benches may function primarily as floodplains in some environments due to the confinement of contemporary flows within large relict channels

(Nanson and Croke, 2002). Further investigation of these features is required, particularly with regards to their potential as stores of post-disturbance generated sediment. This study uses OSL dating to determine the age and rate of deposition of selected in-channel benches in the study area. An assessment is also made of the importance of these features as stores of recently eroded sediment.

1.4. Human-induced catchment disturbance in Australia

In Australia, the focus of previous research on the impacts of human-induced catchment disturbance has been on the more populated southeastern region (e.g., Eyles, 1977;

Prosser et al., 1994; Wasson et al., 1998; Fryirs and Brierley, 1999; Olley and Wasson,

8 Chapter 1 - Introduction

2003; Gale and Haworth, 2005; Page et al., 2007). The largely temperate Murray-

Darling basin (MDB) and the coastal catchments of New South Wales (NSW) and

Victoria have received most attention (Figure 1.1). Land use disturbance in this region, mainly in the form of vegetation clearance and the introduction of grazing animals, resulted in an extensive phase of river channel and gully incision shortly after settlement in the early nineteenth century (Eyles, 1977; Prosser and Winchester, 1996). Sediment yield estimates for the initial post-settlement period suggest increases of up to one to two orders of magnitude (Wasson et al., 1998; Olley and Wasson, 2003; Rustomji and

Pietsch, 2007). Although the main phase of channel expansion had ended by the mid- twentieth century (Eyles, 1977; Prosser and Winchester, 1996), gullies continue to contribute the majority of sediment to rivers (Wallbrink et al., 1998; Wasson et al.,

1998) and suspended sediment loads remain above pre-disturbance conditions (Olley and Wasson, 2003).

It remains unclear whether the catchment response in other parts of Australia has been similar, given differences in climate, flow regime, terrain, geology, and land use history.

Recently more attention has been paid to other regions of Australia where post-

European land use changes have significantly altered natural systems (e.g., Wasson et al., 2002; Bartley et al., 2007). In particular the catchments of northeastern Australia that drain to the Great Barrier Reef (GBR; Figure 1.1) have been the focus of targeted research over the last decade (e.g., Bramley and Roth, 2002; Neil et al., 2002; Brodie and Mitchell, 2005; McKergow et al., 2005). There is growing concern that the increased delivery of sediment and sediment-bound nutrients and contaminants are

9 Chapter 1 - Introduction

Figure 1.1. Great Barrier Reef catchments, river networks, and mean annual rainfall.

Also illustrated is the location of the study catchment, Theresa Creek.

10 Chapter 1 - Introduction adversely affecting the sensitive ecosystems of the receiving marine environment (van

Woesik et al., 1999; Fabricius and De'ath, 2004). The significance of the problem was highlighted in 2001 as part of the Australian National Land and Water Resources Audit

(NLWRA). The NLWRA concluded that the “biggest total contribution to soil erosion in Australia is from the vast semi-arid woodlands and grazing lands in northern regions” (NLWRA, 2001).

Large-scale model-based findings, suggest that since European settlement (ca. 1850) the sediment yield from the GBR catchments has increased by up to an order of magnitude

(e.g., NLWRA, 2001; Neil et al., 2002; Furnas, 2003; McKergow et al., 2005).

Increased sediment yield has been directly attributed to land degradation in response to large-scale vegetation clearance and overstocking (Brodie et al., 2003; Furnas, 2003).

Limited evaluation of sediment sources suggests that hillslope (sheetwash and rill) erosion contributes the majority of sediment to the GBR catchments (NLWRA, 2001;

Dougall et al., 2005; McKergow et al., 2005), although recent research suggests that channel sources have probably been underestimated (Bartley et al., 2007; Brooks et al.,

2007).

Focused attempts have also been made more recently to improve erosion and sediment budget predictions by improving model inputs (e.g., Dougall et al., 2005; McKergow et al., 2005). While this work has been invaluable, many of the model estimates are largely unvalidated by empirically-derived data.

The use of model outputs, without any validation with field-derived data, is risky. Use of erroneous results may lead to the implementation of poor catchment management decisions and the inefficient allocation of catchment rehabilitation resources. In this study, the acquisition of field-based catchment sediment source and sink data provides

11 Chapter 1 - Introduction the opportunity to test the performance of SedNet, a catchment-scale erosion/sediment yield model. SedNet has been applied widely throughout Australia, including the GBR catchments. To date, there have been very few attempts to validate the results of this modelling with empirically-derived data.

1.5. Research aims

The specific aims of this study are:

1. Determine the relative importance of major catchment sediment sources, both in

terms of spatial provenance and erosion process, and identify how sediment

sources have changed as a result of post-European land use changes.

2. To investigate the application of 137Cs dating techniques for determining rates of

recent floodplain sedimentation in a low fallout, southern hemisphere

environment.

3. To compare and contrast pre- and post-disturbance rates of alluvial

sedimentation and interpret how changes in rates of sedimentation provide

evidence of changes in catchment sediment flux.

4. Use empirically-derived data to validate a spatially distributed erosion/sediment

yield model with a view to providing improved estimates of pre- and post-

disturbance catchment sediment yield.

1.6. Thesis structure

The GBR catchment area is extensive (423 000 km2) and encompasses environments ranging from small, wet coastal catchments with high energy rivers through to large inland grassland and woodland dominated catchments that have intermittent flow

12 Chapter 1 - Introduction regimes. Over 70% of the catchment area can be classified as dry tropical (mean annual rainfall < 800 mm; Furnas, 2003). Over 80% of this area is occupied by two catchments; the Fitzroy River (~ 140 000 km2) and Burdekin River (~ 130 000 km2)

(Figure 1.1). The size of catchments in the GBR clearly prevents detailed field estimates from entire catchment areas. The approach of this research, therefore, is to undertake the majority of the field-based work in a single catchment.

This study focuses on Theresa Creek, a 6000 km2 catchment within the Fitzroy River basin, central Queensland, Australia (Figure 1.1). Theresa Creek was selected as it is a mixed-land use (agriculture/grazing) catchment and is representative of the heavily impacted headwaters found throughout the dry-tropics of northeastern Australia.

Theresa Creek is characterised by both high rates of hillslope (Carroll et al., 1997) and gully erosion (Hughes et al., 2001). Anecdotal reports also suggest that human land use disturbance in the catchment has resulted in considerable within-channel and floodplain deposition. In addition, focussing research on an upland catchment, in close proximity to significant areas of erosion, is likely to provide the greatest opportunity to detect the impact of human-induced catchment disturbance.

This thesis consists of six chapters: this introduction, four main research chapters, and a summary and conclusions chapter. The main chapters are presented as a series of independent research papers that have been either submitted or accepted for publication in peer-reviewed journals (as outlined at the beginning of each chapter). Appendices are used in this thesis to present data which, although considered essential for a complete understanding of the results, are regarded as extraneous to the main argument of the text.

13 Chapter 1 - Introduction

Chapter 2 uses a multi-faceted approach to investigate the changes in sediment sources in the study catchment, both in terms of spatial provenance and erosion type. The relationship between erosion type and spatial provenance is described and inferences made on how the relative contribution of hillslope and channel sediment sources have been altered by post-European settlement land use changes.

Chapter 3 investigates the applicability of 137Cs for determining sedimentation rates in a dry, low-latitude, southern hemisphere environment. The two most commonly used methods for determining floodplain sedimentation rates using 137Cs, depth profiles, and total inventories, are assessed.

Chapter 4 examines floodplain and in-channel bench sedimentation within Theresa

Creek. Changes in pre- and post-settlement sedimentation rates are analysed to evaluate how catchment disturbance has affected alluvial sedimentation rates. Assessment is also made of whether previously published estimates of large increases in post- disturbance sediment yield are supported by the floodplain sedimentation record.

Chapter 5 applies the spatially distributed erosion/sediment yield model SedNet (Prosser et al., 2001) to the study catchment. The model’s ability to predict accurate end-of- catchment sediment loads and spatial representations of major sediment sources is evaluated using available suspended sediment/discharge monitoring data and radionuclide tracing of sediment sources. An assessment is also made of SedNet’s predictions of pre-European settlement sediment yield.

Chapter 6 summarises the overall findings of the thesis as assessed against the initial aims. Recommendations for future research are made within the context of the acknowledged limitations of the study.

14 Chapter 1 - Introduction

15 CHAPTER 2

Sediment source changes over the last 250 years in a dry- tropical catchment, central Queensland, Australia

Gully erosion in upper Theresa Creek.

A version of this chapter has been published in Geomorphology as: Hughes, A.O.,

Olley, J.M., Croke, J.C., McKergow, L.A., 2008. Sediment source changes over the last

250 years in a dry-tropical catchment, central Queensland, Australia. doi:10.1016/j.geomorph.2008.09.00 Chapter 2 - Sediment source changes over the last 250 years

2.1. Introduction

Many erosion studies focus on sheetwash and rill erosion as the primary sources of sediment, with limited consideration of the contribution of river bank and gully erosion

(e.g., Lal, 2003; Seidel and Mäckel, 2007). In some environments river bank and gully erosion contribute the majority of a catchment’s total sediment yield (Olley et al., 1993;

Wallbrink et al., 1998; Wasson et al., 1998; Poesen et al., 2003). It is, therefore, important that all potential sources of sediment are considered. Improved information on sediment sources enhances the ability to construct accurate catchment sediment budgets, especially in large catchments where it may be inappropriate to up-scale data from small catchment or plot-scale studies (Walling, 2005). Sediment source information can also support catchment management decisions and ensures that limited rehabilitation resources are targeted toward major problem areas (Wasson et al., 2002).

Few empirically-based studies exist that assess the relative importance of hillslope and channel sources in rivers draining to the Great Barrier Reef (GBR). This is despite the realisation that European land use changes over the last ~ 150 years may have increased river sediment yields with resultant adverse effects on the receiving marine environment

(Neil et al., 2002; McCulloch et al., 2003; McKergow et al., 2005). Regional and catchment scale studies suggest that since European settlement (ca. 1850) sediment yields from the GBR catchments have increased by up to an order of magnitude (Neil et al., 2002; Furnas, 2003; McKergow et al., 2005). Increased sediment yields have been directly attributed to land degradation which has occurred as a result of vegetation clearance and overstocking (Brodie et al., 2003; Furnas, 2003).

To date, most of our knowledge of sediment sources and increased catchment sediment yields in the GBR catchments is based on large-scale modelling studies (e.g., Belperio,

16 Chapter 2 - Sediment source changes over the last 250 years

1983; Moss et al., 1992; McKergow et al., 2005). While these studies provide valuable information on the response of large catchments to land disturbance, knowledge gaps remain, particularly with respect to sediment sources.

The issue of the relative contribution of hillslope and channel sources within Australian catchments was reviewed by Prosser et al. (2001), who stated that while there is information for some areas, particularly southeastern Australia, there are still significant regional gaps in our knowledge. For the purpose of this study hillslope erosion refers to sheetwash and rill processes and channel erosion comprises both riverbank and gully erosion. Prosser et al. (2001) proposed that as catchment size increases, the likelihood of channel sources becoming more dominant also increases due to hillslope sources being increasingly buffered by sediment storage. Because of the size of the two main dry tropical catchments (Burdekin River ~130 000 km2; Fitzroy River ~140 000 km2) draining into the GBR lagoon, in addition to their low terrain, dry-tropical climate and land use history, channel sources are probably more significant than previously identified.

Very few studies examine the sediment provenance and/or the relative contributions of different erosion processes from GBR catchments. McKergow et al. (2005) used the spatially distributed sediment budget model SedNet to determine the patterns and relative contribution of different erosion sources for all GBR catchments. Their results suggested that although channel erosion dominated in some catchments, hillslope erosion accounted for over 60% of the sediment being transported by GBR catchments.

While this is likely to be the case for the steeper, wet coastal catchments, it is less certain for the extensive low-lying dry-tropical zones. Douglas et al. (2005) identified the relative contribution of soils derived from different rock types to the Fitzroy River

17 Chapter 2 - Sediment source changes over the last 250 years using the geochemical properties of sediments collected from dams, weir pools and flood deposits; however, they did not make any comments about dominant erosion processes. The study by Bartley et al. (2005) used field-based data to construct a sediment budget for a 13.5 km2 catchment within the Burdekin River basin. Bartley et al. (2005) identified gully erosion as a significant source of sediment to the river system.

However, this study was carried out during a period of below average rainfall and did not experience any large rainfall events. Using spatial modelling, with support from fallout caesium-137 (137Cs) concentration data, Bartley et al. (2004) identified hillslope erosion as the dominant erosion process in the coastal Herbert River catchment.

Field-based erosion studies within GBR catchments have focused almost exclusively on the effects of grazing or cultivation on hillslope erosion (e.g., McIvor et al., 1995;

Carroll et al., 1997; O'Reagain et al., 2005) with little attention given to the importance of channel erosion sources. In the more widely studied southeastern Australian catchments channel sources can contribute up to 90% of the total sediment yield (e.g.,

Olley et al., 1993; Wallbrink et al., 1998; Wasson et al., 1998). Furthermore, recent findings from northern Australia suggest that gully erosion may be a more significant contributor of sediment in tropical locations than previously thought (Wasson et al.,

2002; Brooks et al., 2007).

This study examines the sediment sources from Theresa Creek, a 6000 km2 subcatchment of the Fitzroy River basin in central Queensland, Australia (Figure 2.1).

The Fitzroy River basin (143 000 km2) has been identified by several regional-scale modelling studies as the single largest contributor of sediment to the GBR (Neil et al.,

2002; McKergow et al., 2005). Fallout radionuclide concentrations are used to determine the relative contribution of channel and hillslope sediment sources.

18 Chapter 2 - Sediment source changes over the last 250 years

Geochemical properties of river bed and floodplain sediments are used to determine the recent and historical (ca. 250 years BP) spatial provenance of transported sediment.

Optical dating and a 137Cs depth profile provide information on the timing of floodplain sediment deposition. The relationship between erosion type and spatial provenance is described and inferences made on how the relative contributions of hillslope and channel sources have been affected by post-European settlement land use changes.

Figure 2.1. Theresa Creek catchment, showing the location of the sediment sampling sites and gauging stations referred to in the text.

2.2. Study site

Theresa Creek is located within the Nogoa River subcatchment in the western Fitzroy

River basin (Figure 2.1). The main population centre in the catchment is Clermont

(population ~ 2000), which lies 300 km west of Rockhampton (population ~ 65 000),

19 Chapter 2 - Sediment source changes over the last 250 years the largest urban centre in the Fitzroy River basin. Theresa Creek joins Retreat Creek ~

10 km north of Emerald, which eventually joins the main channel of the Nogoa River.

This study focuses on a 6000 km2 subcatchment of Theresa Creek. The terminus of the catchment, for the purpose of this study, is located ~ 20 km upstream of the confluence with Retreat Creek. Catchment elevation ranges between 160 m at the catchment outlet to over 800 m at the catchment divide.

The climate is dry-tropical (Köppen climate classification: BSh). The mean annual rainfall at Clermont is 649 mm (1 σ = 240 mm; BoM, 2007; Figure 2.2) with most rainfall occurring between November and March. The long-term rainfall record shows a cyclic pattern of periods of above-average rainfall followed by periods of drought

(Figure 2.2). Such a pattern may be related to the El Niño Southern Oscillation (ENSO) and/or the Pacific Decadal Oscillation (PDO; Power et al., 1999; Mantua and Hare,

2002). Mean annual potential evaporation at Clermont is ~ 2080 mm (BoM, 2007).

Runoff is highly variable; the Nogoa River has an annual flow coefficient of variation of

1.34, one of the highest measured in Australia (Finlayson and Brizga, 1993). Most large runoff events are a result of tropical low pressure systems that periodically make landfall.

The study catchment is comprised of two main tributaries, Sandy and Theresa Creeks.

Sandy Creek drains the eastern and northern parts of the catchment, while Theresa

Creek drains the western catchment. The rock types in the catchment provide distinct geological differentiation of the major tributaries (Figure 2.3A). The eastern side of the catchment is almost completely comprised of deeply weathered Tertiary basalts. The geology of the western catchment is more complex: Devonian granites, Neoproterozoic

20 Chapter 2 - Sediment source changes over the last 250 years to Cambrian metasediments and early Carboniferous sandstones and siltstones are the most spatially abundant geologies. The headwaters of Sandy Creek in the northern catchment are almost exclusively Neoproterozoic to Cambrian metasediments. Tertiary to Quaternary clastic strata are also located throughout the catchment, particularly adjacent to the contemporary floodplain alluvium of the main tributaries.

1400

1200

1000

800

600

Rainfall (mm) Rainfall 400

200

0 1889 1899 1909 1919 1929 1939 1949 1959 1969 1979 1989 1999 Year

Figure 2.2. Annual rainfall from Clermont Post Office climate station (22.83˚S., 147.64˚E.) for the period 1889 - 2006 (BoM, 2007). Solid line is a 5-year running average. Dashed horizontal line is the mean annual rainfall for the entire 118 year period (649 mm y-1).

Catchment topography is closely related to rock type. The Tertiary basalt-dominated eastern catchment is largely gently sloping, while the Devonian granite and

Tertiary/Quaternary clastic sedimentary areas tend to be low rolling terrain. The

Neoproterozoic to Cambrian metasediment and Carboniferous clastic areas are (in the most part) steep and deeply dissected.

The streams draining the basalt terrain tend to be relatively shallow (< 3 m) and in some locations the channels become almost indistinguishable from the surrounding floodplain. Field observations indicate that most of the sediment transported by these

21 Chapter 2 - Sediment source changes over the last 250 years channels consists of silt and clay with very little sand or gravel. The main channels in the metasediment and granite areas are generally deep (~ 10 m) and are characterised by large volumes of sand- to gravel-sized sediment. Gauging station data for Theresa

Creek (gauging station 130210A; 1971 - 2005) indicate that, apart from some permanent waterholes, the river is dry for over 70% of the time.

Figure 2.3. (A) Geology of Theresa Creek showing the principal geology classes and (B) simplified land use map of Theresa Creek (Calvert et al., 2000).

As with topography, land use is closely related to the underlying geology (Figure 2.3B).

The eastern side of the catchment, which is underlain by basalt-derived black clays

(Vertisols), is largely cultivated with wheat (dry-season), sunflower, and sorghum (wet- season) (Carroll et al., 1997). The remainder of the catchment is grazed by beef cattle.

The “easier” country, associated with the Devonian granites and Tertiary/Quaternary

22 Chapter 2 - Sediment source changes over the last 250 years clastic sediments, has been extensively cleared of woody vegetation, while the steeper, less accessible areas remain largely vegetated with grassy woodland.

2.3. Methods

2.3.1. Sediment sources

2.3.1.1. Radionuclide tracing

Radionuclide tracing has been used to determine the relative contribution of different erosion processes in catchments (e.g., He and Owens, 1995; Wallbrink et al., 1998).

The method involves characterising potential sources of erosion (e.g., cultivated land, uncultivated pasture, and channels) on the basis of their radionuclide (e.g., caesium-137

137 226 210 ( Cs), radium-226 ( Ra), and excess lead-210 ( Pbex)) concentrations. These concentrations are then compared with the radionuclide concentrations of samples from downstream sediment deposits (e.g., river bed) using a numerical mixing model to determine the relative contribution of each of the sources (e.g., Walling, 2005). This method is dependent on the documented observation that the concentration of some

137 210 radionuclides (such as Cs, Pbex) varies significantly with soil depth (Wallbrink and

210 Murray, 1993; He and Owens, 1995). Caesium-137 (half-life 30.2 years) and Pbex

(half-life 22.3 years) are both fallout radionuclides that on deposition bind closely with sediment particles. Caesium-137 is a product of atmospheric nuclear testing that

210 occurred from the 1950s to 1970s, while Pbex is a naturally occurring radionuclide that is generated by the decay of Radon-222 (222Rn) in the atmosphere (He and Owens,

137 210 1995; Wallbrink et al., 1998). At undisturbed sites Cs and Pbex are concentrated in the top 10 cm of soil, while at cultivated sites they are mixed to ploughing depth, and accordingly are found in lower concentrations (He and Owens, 1995). Sediment derived from channel banks and gully side walls/head-cuts generally have low or undetectable levels of fallout radionuclides (Wallbrink and Murray, 1993).

23 Chapter 2 - Sediment source changes over the last 250 years

Sediment source samples were collected throughout the study catchment from each of the three principal source types (cultivated land, uncultivated pasture, and channels).

For both uncultivated pasture and cultivated sources, 10 colluvial toeslope sites were identified and spot samples (~ 1 kg) from the top 2 cm of soil were collected. Sampling focussed on colluvial toeslopes adjacent to the drainage network as sediment from these zones best represents sediment that has been mobilised from hillslopes and is in the process of being transported to the river network. Ten channel source sediment samples were obtained directly from actively eroding gully sidewalls and headcuts, and exposed river banks. River bank and gully erosion both contribute subsurface-derived sediment, therefore, the radionuclide signals of the two sources are indistinguishable and assessment of their relative contribution is not possible using this method.

Most applications of sediment tracing have involved the analysis of suspended sediment from flood events (e.g., Collins et al., 1997a; Russell et al., 2001), although it has also been successfully applied to river bed sediments (e.g., Olley and Caitcheon, 2000;

Dirszowsky, 2004). In the case of Theresa Creek, river bed sediments were sampled as there is no flow for most of the year and no flood events occurred during the study period. A number of limitations with this approach are recognised. Firstly, the analysis of river bed sediments provides information on the provenance of sediment from the most recent flow or series of flows. Localised rainfall events generating runoff in one tributary, but not another, may mean that sediment on the river bed may not be representative of catchment-wide sources. The close proximity of the headwater areas in Theresa Creek reduces the likelihood of this occurring, although it cannot be discounted as a possibility. Secondly, river bed sediment is likely to be representative of sources in the latter part of a flood and this material may not necessarily be

24 Chapter 2 - Sediment source changes over the last 250 years representative of the sources throughout the entire flood event. Lastly, there is also the possibility that river bed sediment may be influenced by the input of sediment derived from channel sources with high connectivity during smaller flow events. Despite these limitations, geochemical and radionuclide tracing of river bed sediments provides useful information on the relationship between sediment spatial provenance and erosion processes.

Five river bed sites were sampled on two occasions between August and November

2006. The sites were located on the main channel of Sandy Creek (SC1, SC2, and

SC3), Theresa Creek (TC1), and the confluence of Sandy and Theresa Creeks near the outlet of the catchment (TC2) (Figure 2.1). Each sample site was a section of river ~ 1 km long where ~ 20 sub-samples (~ 1 kg each) were collected. Sub-samples were collected from a range of bed-forms to a depth of ~ 30 cm.

All samples were prepared for analysis by separating out the < 10 µm fraction, grinding and ashing. Samples were wet sieved to retrieve sediment < 63 µm, then fractionated by water column settling to obtain the < 10 µm fraction (clay and very fine silt). The samples were oven-dried then homogenised by grinding in a tungsten ring mill.

Organic material was removed by ashing the samples at 450 ˚C for 48 h. Geochemical analysis of a very narrow particle size range (i.e., <10 µm) and removal of organic material by ashing obviates the need for particle-size and organic content correction factors as has been done in some previous studies (e.g., Collins et al., 1997b; Motha et al., 2003). Although the < 10 µm fraction makes up a small proportion of the channel bed sediment (~ 10%) it does contribute a much higher proportion of the sediment transported during flood events and comprises ~ 50% of the sediment found on

25 Chapter 2 - Sediment source changes over the last 250 years floodplains. Also this very fine particle size range has the potential for the greatest impact on the GBR receiving environment.

Radionuclide analysis using high resolution gamma spectrometry, as described by

Murray et al. (1987), was carried out at the CSIRO Land & Water gamma spectrometry laboratory in Canberra.

2.3.1.2. Geochemical tracing

Geochemical tracing involves characterising the spatial provenance of eroded sediment on the basis of its geochemical signature (Walling, 2005). The method is dependent on contrasting chemical signatures of the underlying bedrock of different source areas.

Catchments are usually divided into broad geological units and the geochemical signature of the soils developed on these units are characterised and compared to the geochemical signature of suspended or deposited sediments. As with the radionuclide tracing technique, a mixing model is used to determine the relative contribution of the different spatial sources. Geochemical tracing has been used extensively and in a wide variety of settings to determine the provenance of in-transport or deposited sediment, including floodplain core sediments (e.g., Owens et al., 1999; Dirszowsky, 2004).

For the purpose of characterising source areas, the catchment was classified into four main geological units: Tertiary basalts (33%), Devonian granites (15%), Neoproterozoic to Cambrian metasediments (13%), and Tertiary to Quaternary clastic sediments (13%; see Figure 2.3A). Other units of sizeable spatial extent that were not sampled include the Quaternary alluvium (9%) and Carboniferous sandstones and siltstones (4%). As

Quaternary alluvium is primarily located on the contemporary floodplains, it was not considered to be a significant source of sediment. The Carboniferous unit was not sampled because of its limited areal extent and because remotely sensed data (e.g.,

26 Chapter 2 - Sediment source changes over the last 250 years aerial photography and QuickBird satellite imagery) suggested it was unlikely to be a major source area.

Sediment samples used to characterise the four geological source areas were collected from the beds of first- or second-order channels that drained only one geological unit.

Sites were identified via desk-top GIS analysis and were located in the field by a hand- held mapping GPS unit (Thales Mobile MapperTM) equipped with ArcPadTM software.

At each site 10 sub-samples (0.5 kg each) were obtained from a channel reach. The collection of channel bed samples was preferred over hillslope soil samples as channel sediments are likely to be more representative of source areas as the effects of local source area heterogeneity are likely to be averaged out by river transport processes

(Olley and Caitcheon, 2000).

Deposited sediment samples were obtained from both river beds and a floodplain core collected from the lower reaches of Theresa Creek (Figure 2.1). The river bed sampling sites were the same as those used for the radionuclide tracing. The floodplain core was extracted from a site below the confluence of Theresa and Sandy Creeks within 10 m of the true left bank. Flood events large enough to result in overbank flow at this site are likely to be the result of rainfall throughout the catchment, therefore, most parts of the catchment will be contributing. Analysis of available gauging station data from two locations (gauging stations 130208A and 130211B; Figure 2.1) on opposing sides of the catchment show a similar runoff response during larger runoff events. For the period

2 between 1976 and 1988 there was a strong relationship (r = 0.78) between discharge volumes at the two gauging stations when daily discharge exceeded 3000 megalitres.

Floodplain sediment at the floodplain core site is, therefore, assumed to be representative of sediment transported during large flood events and that these large

27 Chapter 2 - Sediment source changes over the last 250 years events will contribute sediment from throughout the catchment. The floodplain core was obtained using a vehicle-mounted hydraulic soil coring rig that extracted 80 cm long cores with a diameter of 76 mm.

The floodplain core was sectioned into 2 cm increments to a depth of 60 cm. Each of the source/river bed samples was bulked and split into samples of ~ 500 g. Samples were prepared using the same methods as the radionuclide tracing samples; the < 10 µm fraction was ground and ashed.

Major and trace element concentrations of the processed samples were determined using a PANalytical Axios-Advanced wavelength dispersive X-ray fluorescence (XRF) system at the CSIRO Land and Water XRF laboratory in Adelaide, Australia. Samples were prepared by fusing ~ 1 g of each sample with 4 g of lithium borate flux at 1050°C

(Norrish and Hutton, 1969). XRF analysis determined the concentrations of major elements (%) (SiO2, AlO2, MgO, Fe2O3, CaO, Na2O, K2O, TiO2, P2O5, MnO, SO3) and minor elements (ppm) (ZnO, CuO, SrO, ZrO2, NiO, Rb2O, BaO, V2O5, Cr2O3, La2O3,

CeO2, PbO, Y2O3, CoO, Ga2O3, U3O8, ThO2, As2O5, Cl) for each of the source and deposited (river bed and core) samples.

Stepwise linear discriminant analysis (LDA) was used to determine the combination of geochemical properties that provide the best differentiation between geological source areas. Stepwise selection was made on the basis of the smallest p value. The combination of geochemical properties that successfully classified all source samples with the lowest Wilks’ lambda value was chosen. Wilks’ lambda is a multivariate test statistic that measures the significance of the discriminatory power of the model.

Wilks’ lambda values close to zero indicate high discriminatory power while values close to one indicate low discriminatory power.

28 Chapter 2 - Sediment source changes over the last 250 years

2.3.1.3. Mixing model

A Monte Carlo mixing model was used to predict the relative contribution of each of the sources to the channel and floodplain deposit samples. The mixing model, a modification of the approach outlined by Olley and Caitcheon (2000), was used for both the geochemical and radionuclide tracing. This particular model was used as it was easily solved within Microsoft Excel® using the Solver tool and was able to be applied to both the radionuclide and geochemical tracing data. In the mixing model, individual sample concentrations are denoted by Ci,j,k, where i = source index (i = 1, …, I; I = 3 for erosion sources and I = 4 for rock type sources), j = sample number index (j = 1, …, J; J is 10 for both the geochemical and radionuclide tracing) and each sample has k constituent concentrations (k = 1, …, K; K = 2 for radionuclide tracing and K = 9 for geochemical tracing).

For each Monte Carlo iteration (l) and for each source (i), j is randomly selected and

Ci,j,k,l is used to calculate source-weighted composite concentrations:

I C C ,lk = ∑ ρˆ ,, kjii ,l (2.1) i=1

Where l = 1, …, 1000 and ρˆi is the proportion contributed from each erosion type/rock

type source. The relative contribution of each erosion type/rock type source, ρˆi must

I meet the following constraints: 0 ρˆ i ≤≤ 1 and ∑ ρˆ i = 1. i=1

For each geochemical property/radionuclide tracer, the average concentration of C is calculated over 1000 iterations using:

29 Chapter 2 - Sediment source changes over the last 250 years

1000 CC k = ∑ ,lk 1000 l=1 (2.2)

The best estimate of the relative contribution ( ρˆ i ) of each erosion type/rock type was determined by minimising the sum of squares of the deviations of the concentration calculated in Equation 2.2 from the measured geochemical property/radionuclide tracer concentration of the deposit (Cd):

2 K  − CC   dk  (2.3) ∑ C  k=1  d 

The mixing model goodness-of-fit was assessed by determining the relative difference between the predicted and measured tracer concentrations of each tracer property (see

Collins et al., 1997b). The relative differences were averaged to determine a mean relative difference for each core increment/river bed sample. Mean relative differences were typically between ± 9 and ± 25% for the geochemical tracing and ± 14 and ± 30% for the radionuclide tracing (see Appendix 1 for mean relative differences for each channel bed and core sample).

2.3.2. Temporal changes in sediment sources

To determine the timing of changes in catchment sediment sources, a chronology for the floodplain core was established, principally by optical dating of quartz grains. Some additional information was also provided by a floodplain core 137Cs depth profile.

2.3.2.1. Optical dating

Optical dating has proved to be a reliable method for dating fluvial sediment, and recent advances in single-aliquot techniques have improved the confidence in dating young

30 Chapter 2 - Sediment source changes over the last 250 years sediment (Wallinga, 2002; Olley et al., 2004). The premise of optical dating is that, when buried, quartz grains begin to accumulate a trapped-charge population that increases in a measurable and predictable way in response to the ionising radiation dose to which the grains are exposed. Exposure to sunlight releases the light-sensitive trapped charge, thereby resetting the OSL signal; a process commonly referred to as

‘bleaching’. The time elapsed since sediment grains were last exposed to sunlight may be estimated by measuring the OSL signal from a sample of sediment, determining the equivalent dose (De) that this represents (for which the SI unit is the gray, Gy), and estimating the rate of exposure of the grains to ionising radiation averaged over the period of burial. The latter parameter of interest is termed the dose rate (Dr). The burial age of well-bleached grains may then be obtained from the following equation:

Burial age = De / Dr (2.4)

The OSL signal of clean quartz grains exposed directly to sunlight is reduced to a negligible level within a few seconds (Wintle, 1997; Aitken, 1998). However, incomplete or non-uniform bleaching is commonplace in many depositional environments (Murray and Olley, 2002), due to surface coatings on the grains and/or insufficient exposure to sunlight during sediment transport. This results in grains being deposited with a heterogeneous distribution of residual trapped charge, and a correspondingly wide range of measured De values. For such sediments, the population of grains with the lowest measured De values provides the most accurate estimate of Db: the burial dose to which those grains that were well bleached at deposition have been exposed since the most recent transport event (Olley et al., 2004).

31 Chapter 2 - Sediment source changes over the last 250 years

Sediment samples for optical dating were collected by driving steel tubes into the sidewall of a pit at the floodplain core site (Figure 2.1). Samples were extracted from various depths and sealed to remain light-proof. Sand-sized grains of quartz (180 - 212

µm) were extracted from the tubes using the methods described by Olley et al. (2004).

A modified version of the single grain regenerative dose protocol (SAR), as developed by Olley et al. (2004), was used to obtain the De distributions for each sample. Sample burial doses (Db) were estimated using a probability density summation method described by Rustomji et al. (2006). The contribution of lithogenic radionuclides to the

Dr was obtained by high-resolution gamma spectrometry analysis (Murray et al., 1987) of the sediment immediately surrounding the sampled quartz grains. Total dose rate was determined by adding the cosmic ray contribution, as determined from Prescott and

Hutton (1994), to the lithogenic dose rate component. The dry dose rates were adjusted, following Aitken (2005), assuming water content of 7.5 ± 5% for all samples.

2.3.2.2. Caesium-137 dating

Caesium-137 soil depth profiles have been used widely to determine floodplain sedimentation rates (e.g., Ormerod, 1998; Owens and Walling, 2002). The method generally involves determining the amount of deposition since an identifiable activity level in 137Cs in a soil profile. For a detailed description of the use of 137Cs to determine floodplain sedimentation rates see Walling and He (1997).

The floodplain core extracted from the lower reaches of Theresa Creek (Figure 2.1) was sectioned into 2 cm increments. The sediment from each increment was oven-dried and sieved to remove material > 2 mm. High resolution gamma spectrometry was used to determine 137Cs concentrations for each increment. The sedimentation rate since ca.

32 Chapter 2 - Sediment source changes over the last 250 years

1954 was predicted from the resulting 137Cs depth profile using the advection-diffusion model approach described in Chapter 3 (Section 3.3.2).

2.4. Results

2.4.1. Geochemical differentiation of geological source areas

The results from the stepwise LDA (Table 2.1) show that MgO, ZrO2, K2O, V2O5, and

NiO were able to correctly differentiate all of the geological source area samples.

However, the inclusion of further tracer properties into such an analysis improves the reliability of the resulting model (Collins et al., 1998). Accordingly, a further four geochemical properties (CeO2, TiO2, P2O5, SrO) were added by the stepwise process and, as indicated by the decrease in the Wilks’ lambda value, the model improved as a result (Table 2.1). The nine geochemical properties selected by the stepwise LDA provide good differentiation between sediments derived from the different source area rock types (Figure 2.4). This provides confidence that the geochemical signature of the

< 10 µm fraction is distinct enough to distinguish all of the sources. Basic descriptive statistics for the geochemistry of each source area are presented in Appendix 2. As is commonly observed, Mg, Ti, Sr, P, and Ni are present in high concentrations in basalt while K and Ce are more abundant in granite (Faure, 1998). The element concentration ranges of the source samples encompass those of the river bed and floodplain core deposits (Figure 2.4), which is consistent with all major source areas being sampled.

33 Chapter 2 - Sediment source changes over the last 250 years

Table 2.1. Geochemical properties that provide the best differentiation of geological source areas by stepwise linear discriminant analysis.

Geochemical property % of source samples Wilks’ lambda correctly classified MgO 82.05 0.1524 ZrO2 87.18 0.0324 K2O 92.31 0.0106 V2O5 97.44 0.0057 NiO 100.00 0.0028 CeO2 100.00 0.0016 P2O5 100.00 0.0010 TiO2 100.00 0.0006 SrO 100.00 0.0004

The nine geochemical properties selected by the stepwise LDA are used in the mixing model to estimate the relative contribution of the four geological source areas (see below).

2.4.2. Fallout radionuclide differentiation of sediment sources

137 210 Concentrations of both Cs and Pbex are highest in the uncultivated, pasture-derived sediment, lowest in the sediment derived from channel sources, and of low/intermediate concentration from cultivation sources (Table 2.2). The radionuclide concentrations of the cultivated and channel sources are both low; however, they are significantly

210 137 different from each other (p < 0.001 for Pbex and p < 0.05 for Cs). The concentrations of 137Cs are low compared to studies using similar methodologies (e.g.,

He and Owens, 1995; Wallbrink et al., 1998). This can be attributed to the relatively small amount of atmospheric nuclear testing that occurred in the southern hemisphere with the consequent levels of soil 137Cs being an order of magnitude lower than the northern hemisphere (Wallbrink et al., 1998). In addition, the catchment’s relatively low latitudinal position (23˚S) means that it was subjected to a lower level of fallout

34 Chapter 2 - Sediment source changes over the last 250 years

6.0 250

200 4.0

(ppm) 150 2

MgO (%) 2.0 CeO 100

0.0 50 45 50 55 60 65 0 20 40 60 80 6.0 400

300 4.0 200 O (%) 2 K

2.0 NiO(ppm) 100

0.0 0 45 50 55 60 65 0 20 40 60 80 0.6 650

550 0.4 (%) 5 (ppm) 450 5 O 2 O 2 P 0.2 V 350

0.0 250 45 50 55 60 65 0 20 40 60 80 3.0 350

2.0 250 (%) (ppm) 2 2

TiO 1.0 150 ZrO

0.0 50 45 50 55 60 65 0 20 40 60 80 400 SiO2 (%)

300 Granite Metasediment 200 Basalt Clastic SrO (ppm) 100 Channel Floodplain core 0 0 20 40 60 80 ThO2 (ppm)

Figure 2.4. Concentrations of nine elements (as selected by stepwise linear discriminant analysis) in the sediment samples collected from the four principal source areas, floodplain core, and river bed. Major elements are plotted against SiO2. Trace elements are plotted against ThO2.

35 Chapter 2 - Sediment source changes over the last 250 years

than that experienced in the mid-latitudes (UNSCEAR, 2000). The negative mean

210 226 Pbex value in Table 2.2 is because the parent Ra concentration is greater than the

210 210 total Pb value. This, in effect, means Pbex concentrations are zero in channel- derived sediment.

137 210 Table 2.2. Mean Cs and Pbex concentrations of the < 10 µm fraction of sediment from uncultivated pasture, cultivated land, and channels (errors are two standard deviations).

137 -1 210 -1 Source type Cs (Bq kg ) Pbex (Bq kg ) Uncultivated (n = 11) 7.3 ± 6.6 126.8 ± 117.2 Cultivated (n = 10) 0.7 ± 0.6 3.5 ± 5.4 Channel (n = 10) 0.0 ± 0.8 -6.9 ± 10.0

Previous research has found that the concentration of fallout radionuclides at cultivated sites is ~50% less than at uncultivated sites (e.g., He and Owens, 1995; Wallbrink et al.,

137 210 1998). In this study, however, the concentration of both Cs and Pbex are at least one order of magnitude lower at cultivated sites (Table 2.2). Although extensive and deep (~ 50 cm) ploughing of the Vertisols has contributed to the lower fallout radionuclide concentrations, these cannot be explained by ploughing dilution alone.

The lower than expected concentrations may be due to the highly disturbed nature of these cultivated soils that have undergone extensive redistribution during the construction of erosion control measures, such as contour banks (Figure 2.5). These contour banks are extensive, with a 23 km2 landholding within the catchment reported to have over 120 km of contour banks (Burns and Brimblecombe, 2006).

36 Chapter 2 - Sediment source changes over the last 250 years

Figure 2.5. A contour bank typical of those found throughout the intensively cultivated basaltic-derived Vertisol areas.

137 210 137 To test the effectiveness of Cs and Pbex to differentiate sediment sources, Cs and

210 Pbex were analysed by LDA. The two radionuclides were able to correctly differentiate over 83% of the sediment source samples.

2.4.3. Sources of fine river bed sediment

Radionuclide data indicate that hillslope erosion from cultivated land (41 - 43%) and channel erosion (50 - 57%) dominate sediment sources on the main stem of Sandy

Creek (SC1 - SC3; Figure 2.6). Hillslope erosion-derived sediment from uncultivated land only makes up a small proportion (2 - 7%) of the total sediment delivered to the stream network. Basaltic, metasediment, and to a lesser extent, clastic sources make up the majority of sediment at all three Sandy Creek river bed sites (SC1, SC2, SC3). The

37 Chapter 2 - Sediment source changes over the last 250 years low level of granitic-derived sediment (0-7%) is indicative of the limited extent of granite in the contributing areas of the Sandy Creek sites. The fact that cultivation only occurs on the catchment’s basaltic soils suggests that the areas underlain by metasediment and clastic rock types contribute sediment mainly from channel sources.

Field observations confirm that upper Sandy Creek (upstream of the confluence with

Wolfang Creek) is an area of significant active gully erosion (Figure 2.7). The basaltic areas are also likely to contribute channel-derived sediment; however, because of land management practices, such as the construction of contour banks to prevent flow convergence (Figure 2.5), the development of gullies on cultivated land is uncommon

(G. Bourne, Queensland Department of Natural Resources and Water, personal communication, 2006).

The sediment at site TC1 (upstream of the confluence of Sandy and Theresa Creeks) is largely comprised of granitic sources (69%), which is consistent with the dominance of granite in the contributing catchment. Clastic-derived sediment comprises the rest of the fine river bed material. The mixing model predicted no contribution from metasediment-derived material. Metasediment is present upstream of site TC1, but it is located entirely upstream of the Theresa Creek dam. This suggests that the dam is effectively trapping most, if not all, sediment from this part of the catchment.

38 Chapter 2 - Sediment source changes over the last 250 years

Figure 2.6. Sediment source contributions to each river bed sampling site. Each sampling site has two associated pie graphs, the left graph illustrates the predicted relative contribution (%) from each rock type source area, and the right graph indicates the predicted relative contribution (%) from each erosion type.

39 Chapter 2 - Sediment source changes over the last 250 years

Figure 2.7. An example of severe gully erosion from the metasediment-underlain upper Sandy Creek area.

Channel-derived sediment dominates (82%) the sediment sources contributing to site

TC1, while the remainder is derived from uncultivated hillslope sources. This is strong evidence that the granitic terrain is an important source of channel-derived sediment.

This is supported by field observations and previous gully density modelling that predicted severe gully erosion in the granitic-based soils of Theresa Creek (NLWRA,

2001). Uncultivated sources upstream of TC1 contribute a relatively high proportion of sediment (18%) compared to other sites.

Basaltic-derived sediment dominates (74%) downstream of where all the major tributaries join Theresa Creek (i.e., site TC2). Notably, Capella Creek joins the main branch of Theresa Creek upstream of this site (Figure 2.1). Capella Creek drains a large

40 Chapter 2 - Sediment source changes over the last 250 years proportion (~ 55%) of the eastern side of the catchment and is almost completely underlain by Tertiary basalts. Although sediment is contributed from the largely granitic-based tributary upstream of site TC1, this is largely masked at site TC2 by the basaltic input from the eastern catchment. The proportion of sediment derived from cultivated land (64%) closely mirrors the proportion of basaltic-derived sources (74%) as determined by geochemical tracing, again pointing toward the dominance of cultivation sources from this geology type.

2.4.4. Floodplain sediment ages and accretion rates

Four burial ages were determined by optical dating of floodplain sediment (Table 2.3).

The two deepest samples obtained from 0.95 m (LTC-F-95) and 0.50 m (LTC-F-50) produced pre-European settlement burial ages of 791 ± 69 years BP and 254 ± 24 years

BP, respectively. Post-European burial ages of 118 ± 18 years BP and 64 ± 6 years BP were determined at 0.35 m (LTC-F-35) and 0.22 m (LTC-F-22), respectively.

Table 2.3. Dating results and accretion rates for the floodplain sediment core dated from lower Theresa Creek (LTC-F).

Sample ID Dating Depth Age Age Accretion rate method (m) (years BP) (calendar years) (mm y-1) 137Cs profile 137Cs n/a 52 1954 1.4 ± 0.1 LTC-F-22 OSL 0.22 64 ± 6 1936 - 1948 6.3 ± 2.7 LTC-F-35 OSL 0.35 118 ± 18 1870 - 1906 2.4 ± 1.3 LTC-F-50 OSL 0.50 254 ± 24 1728 - 1776 1.1 ± 0.4 LTC-F-95 OSL 0.95 791 ± 69 1146 - 1284 0.8 ± 0.1

Analysis of gauging station data (Appendix 3) indicates that the floodplain core site has been inundated at least six times since the commissioning of gauging station 130210A

(Figure 2.1) in 1972. This, together with the fact that the OSL burial ages and the 137Cs depth profile date are stratigraphically consistent, provides confidence that the sampled

41 Chapter 2 - Sediment source changes over the last 250 years floodplain sediment represents numerous overbank depositional events and not just a few large events.

The accretion rates calculated from the above sediment ages show a significant increase in floodplain sedimentation since European settlement (Table 2.3). Prior to European settlement accretion, rates were low, with ~ 1 mm y-1 being deposited from 791 ± 69 years BP to 118 ± 18 years BP. Between 118 ± 18 years BP and 64 ± 6 years BP, the accretion rate doubled to 2.4 ± 1.3 mm y-1, presumably in response to initial catchment disturbance. Between 118 ± 18 years BP and ~ 48 years BP, the accretion increased again to 6.3 ± 2.7 mm y-1. The results of the 137Cs depth profile advection-diffusion model (see Chapter 3) suggest a recent reduction in accretion rates (1.4 ± 0.1 mm y-1), although they remain higher than the pre-European period.

2.4.5. The spatial provenance of floodplain deposits over the last 250 years

The lower section of floodplain core between 60 and 34 cm is characterised by relatively consistent contribution (with the exception of the 40 - 42 cm increment) of all four main geological sources (Figure 2.8). Based on the optical dating results, this increment range represents a period prior to the eighteenth century to ca. 1888.

European settlement in this part of Australia began ca. 1855; therefore, this section of the core encompasses the pre-European to early settlement period within the catchment.

Between the depths of ca. 34 - 26 cm (ca. 1888 to early 1900s), the relative contributions of both granitic and clastic sediment increase with a corresponding decrease in the contribution of basalt and metasediment sources. The mean contribution of granite and clastic for this section of core are ~ 40% and ~ 32%, respectively. Below this section of core (between 60 - 34 cm), granite and clastic sediment only contributed

~ 25% each.

42 Chapter 2 - Sediment source changes over the last 250 years

Source contribution (%) 0 20 40 60 80 100

0-2

4-6

8-10 AD 1954

12-14

16-18

20-22 AD 1943 ± 7

24-26

28-30

32-34

Depth (cm) Depth AD 1888 ± 18 36-38

40-42

44-46

48-50 AD 1752 ± 24

52-54

56-58

Granite Clastic Metasediment Basalt

Figure 2.8. The relative contributions from each of the major geological source areas to the floodplain core (0 - 60 cm) from lower Theresa Creek. The 1954 date was determined by analysis of a 137Cs depth profile for the site. All other dates were obtained by OSL dating of single grains of quartz.

Between the core depths ca. 26 - 16 cm, which is dated to ca. 1940s - 1950s, the relative contribution of basaltic sediment increases. The mean contribution of basalt in this section of core is 56%, while it is ~ 20% for the 34 - 26 cm section and ~ 30% for the 60

- 34 cm section. This increase in basaltic sources is mainly at the expense of clastic sources, which decrease to the lowest proportions in the core, although granitic sources also contributed proportionately less sediment.

43 Chapter 2 - Sediment source changes over the last 250 years

Between 10 - 4 cm (ca. 1954 - 1980s), the relative importance of basaltic-derived sediment decreases and granitic- and clastic-derived sources again dominate. Also of note, in the top 10 cm of the core is the increased importance of metasediment sources, perhaps indicating some recent land disturbance from the steeper metasediment land over the last few decades.

In the top couple of increments, the relative contribution of basaltic sources increased.

This, together with the large contribution (74%) of basalt-derived sediment at the nearby river bed site TC2, may suggest a recent increase in the relative importance of basaltic sources.

2.5. Discussion

This study indicates that most fine river sediment in this catchment is derived from hillslope erosion of cultivated land and channel erosion. Hillslope erosion of uncultivated land is a comparatively minor contributor of sediment to the river network.

The large contribution of sediment from cultivated sources is attributed to the relatively large proportion (22%) of the catchment with highly erodible basalt-based soils used for intensive cultivation. High rates of soil loss from Vertisol-based soils have been reported from northeastern Australia (e.g., Freebairn and Wockner, 1986; Carroll et al.,

1997). Carroll et al. (1997) found that cultivated Vertisol-based contour bay catchments within Theresa Creek can yield over 4 t ha-1 y-1.

Perhaps of more significance, given that cultivation throughout the Fitzroy River basin only accounts for 7% of total land use, is the contribution of sediment from channel erosion. Previous research has attributed increased catchment sediment yield from dry- tropical GBR catchments to increased hillslope erosion rates (resulting from grazing pressure and intensive cultivation), with limited consideration of channel sources. This

44 Chapter 2 - Sediment source changes over the last 250 years study shows that channel erosion is a significant contributor of sediment to Theresa

Creek. Site TC1 is the only river bed sampling site without a significant area of cultivation in its catchment, and channel sources account for over 80% of the fine river bed sediment. This level of channel erosion contribution is similar to that found in parts of southeastern Australia (e.g., Wallbrink et al., 1998; Wasson et al., 1998). The methods used in this study are not able to distinguish between channel sediment derived from river banks and that derived from gully erosion. However, based on field observations and previous findings in the region (e.g., Bartley et al., 2007), gully sidewalls and headcuts probably contribute significantly more sediment than river banks. Extensive sediment slugs or waves found throughout the river network also suggest a large flux of sediment from gully sources (cf. Bartley et al., 2007). The sand and gravel in these sediment slugs are similar to sediment found in gully sidewalls and beds while it is completely absent from the finer floodplain deposits found in most river banks. Hillslope erosion processes are unlikely to contribute large volumes of coarse sediment to the river network (Bartley et al., 2006).

As summarised in Table 2.4, the river bed sediment tracing results suggest that the majority of basaltic sediment is derived from cultivated land, while the granitic, clastic, and metasediment areas contribute mostly channel-derived sediment. Importantly, this information enables interpretation of the changes in relative contributions of each geological source over the last ~ 250 years, as indicated by the floodplain core extracted from the lower reaches of Theresa Creek.

45 Chapter 2 - Sediment source changes over the last 250 years

Table 2.4. Qualitative assessment of the relative importance of each of the source types by geological source area as determined by the geochemical and radionuclide tracing results (*** = high, ** = moderate, * = low).

Source type Rock type source Channel Cultivated Uncultivated Granite *** n/a ** Basalt * *** * Metasediment *** n/a * Clastic *** n/a *

Geochemical analysis of the floodplain core obtained from the lower reaches of Theresa

Creek has shown that sediment sources have changed since the area was settled by

Europeans over 150 years ago. Two principle post-European settlement-related erosion/deposition periods have been identified. The first of these is an increased mobilisation of granitic- and clastic-derived sediment beginning in the late nineteenth century. Based on the river bed sediment tracing results, the increase of granitic- and clastic-derived material is inferred to be a period of increased sediment flux from channel sources. This is interpreted as the start of a phase of gully/channel incision as a result of increased grazing intensity within the catchment, as has been noted from many parts of southeastern Australia (e.g., Prosser and Winchester, 1996; Gale and Haworth,

2005). Initial settlement from the 1850s saw the introduction of sheep farming to the region, but toward the end of the nineteenth century cattle numbers also increased rapidly (Lewis et al., 2007). The grazing pressure of increased sheep and cattle numbers, in combination with the greater impact of larger grazing animals, probably resulted in a threshold being reached whereby the land became susceptible to gully initiation because of factors, such as reduced ground cover, soil compaction, and tree clearance. An increase in the number and/or size of erosive events with the beginning

46 Chapter 2 - Sediment source changes over the last 250 years of a wetter period around 1890 (Lough, 2007; Figure 2.2) may also have provided a trigger for the onset of widespread gully incision.

The second period of increased erosion and sediment deposition is a result of changes in land use on the low-lying fertile basaltic soils in the mid-twentieth century. During this period, sheep farming on the basaltic-based areas was replaced with dryland cropping

(Burns and Brimblecombe, 2006). The change from relatively low density sheep farming to intensive cultivation is likely to have increased sediment yield due to increased erosion rates from seasonally tilled soils. From the late 1940s through the

1950s, the Queensland-British Food Corporation, then later individual landholders, developed the area for food crops, such as sorghum (Johnson, 1982). This period of significant land use change, in particular the soil disturbance associated with initial vegetation clearance and tillage, coincides with increased sediment yield from the basaltic-based soils. This change in land use also occurred during a period (1950s) of above-average rainfall (Figure 2.2) that may have exacerbated the effect of land use intensification.

A decrease in the relative contribution of basaltic sources after ca. 1960s can be attributed to a decrease in erosion rates after the initial flux associated with land clearance. However, recognition of high rates of soil loss and the resulting improvement in land management practices may have also played a role. Such improvements include the construction of contour banks to reduce flow convergence during large rainfall events and reduced levels of tillage.

The floodplain core mixing model results show that, above the basaltic-dominated section of core (post ~ ca. 1954), granite- and clastic-derived sediment still contributed a relatively large proportion of sediment to the river’s load, with no significant

47 Chapter 2 - Sediment source changes over the last 250 years reduction in sediment yield, as shown by consistently elevated post-European floodplain accretion rates (Table 2.3). This suggests that, although the initial phase of channel/gully incision may have been triggered in the late nineteenth century, gully erosion continues to contribute a significant volume of sediment to the river network.

Field observations throughout the catchment of steep-walled gullies undergoing active head cut suggest that the gully network in Theresa Creek may not yet be stabilised, particularly in the granitic and metasediment areas. Even where the gully network has stabilised, sediment yield may remain high because the vast bulk of sediment yielded from gullies is derived from the sidewalls rather than the gully head (Blong et al., 1982).

This pattern of continued high sediment yield from gullies following gully network stabilisation has been described in other parts of Australia (e.g., Prosser and Winchester,

1996; Wasson et al., 1998).

The finding that the late nineteenth century is a phase of increased sediment yield is in close agreement with previous work that used Ba/Ca ratios in offshore corals to identify an increased sediment yield from the neighbouring Burdekin River after ca. 1870 (see

McCulloch et al., 2003; Lewis et al., 2007). Lewis et al. (2007) also attributed an approximately tenfold increase in manganese (Mn) concentrations in corals to erosion initiated by sheep grazing on basaltic-based soils in the Burdekin River in the 1850s.

Although the Burdekin River is a neighbouring catchment, the area of basaltic-based soils that Lewis et al. (2007) attributed this flux of Mn to, extends into the Theresa

Creek catchment. Lewis et al. (2007) used sheep and cattle statistics from the Clermont and Emerald districts to support their case of grazing-initiated release of Mn-enriched basalt sediment. Interestingly, no analogous increase in Mn concentration was detected in the Theresa Creek core, which would be expected if the dramatic increase in Mn

48 Chapter 2 - Sediment source changes over the last 250 years concentrations found in the corals is related to land use change. Furthermore, Mn concentrations in samples obtained from river beds draining only basaltic soils suggest that basaltic-derived Vertisols in Theresa Creek are not enriched with Mn. An analysis of variance (ANOVA) test failed to detect any significant difference in the mean concentrations of Mn in basalt and the three other geology types. While the parent

Tertiary basalt rock may have high levels of Mn, Vertisols may be low in Mn because of oxidation during the weathering process (Faure, 1998).

Theresa Creek is a relatively small part of the Fitzroy River basin, which at 143 000 km2 is the largest catchment discharging to the GBR. Despite its limited extent, Theresa

Creek is an important subcatchment supplying a significant amount of sediment to the river system for potential export to the GBR. The Nogoa River subcatchment, which

Theresa Creek is part of, is a significant contributor of sediment to the river system.

Using sediment-discharge rating curves, Joo et al. (2005) estimated sediment loads and specific yields for the six main Fitzroy River basin subcatchments and found that the

Nogoa River subcatchment had the highest sediment load and specific yield for the

1974-2003 period. Notably, almost 60% of the Nogoa River’s catchment area discharges into Lake Maraboon (Figure 2.1), a water storage reservoir commissioned in

1973. The reservoir only releases water during large flood events, with the dam wall having only overflowed once between 1985 and 2006 (Anon, 2006). It is, therefore, likely that a large proportion of the sediment load is derived from parts of the Nogoa

River system that do not flow into the reservoir (e.g., Theresa Creek and Retreat Creek).

Basaltic-derived material is clearly an important source of sediment within Theresa

Creek. This is evident from the results for site TC2 that indicate a high proportion of cultivation-derived basaltic sediment in the lower reaches of the river, despite a large

49 Chapter 2 - Sediment source changes over the last 250 years input of channel-derived granitic sediment from the main branch of Theresa Creek.

This indicates that the fine sediment yield from the extensively cropped basaltic areas may be significantly greater than that of the largely grazed granitic-based areas. The clay-rich composition of the basaltic soils suggests this is likely, although it may also be indicative of a localised flow event generated in the basalt-based part of the catchment.

The importance of basaltic sources at a basin-wide scale is, however, uncertain.

Previous studies in large Australian catchments have predicted that, despite extensive headwater contributions, basalt-derived sediment may only contribute a small proportion of the sediment found in the lower reaches of a catchment (e.g., Olley and

Caitcheon, 2000). While the reasons for this are unclear, the highly aggregated nature of the sediment derived from the Vertisols (Freebairn and Wockner, 1986) may result in a high proportion being transported as bed material, which may be stored for long periods of time within river beds. The transport of large volume of aggregated material has also been noted in other environments (e.g., Droppo and Ongley, 1994; Droppo and

Stone, 1994). The view that basaltic sources comprise a limited contribution at the basin-scale is supported by the results of Douglas et al. (2006), who found that sediment deposited in catchment impoundments is mainly derived from non-basaltic sources.

2.6. Conclusions

This study uses geochemical and radionuclide properties of river bed and floodplain sediments to determine the spatial provenance and relative contribution of different source types to fluvial deposits within Theresa Creek, a subcatchment of the Fitzroy

River basin. The data indicate that the river sediments are derived primarily from cultivated basaltic-based land and channel erosion from non-basaltic parts of the catchment. While cultivated areas of the catchment are identified as a major source of

50 Chapter 2 - Sediment source changes over the last 250 years sediment within Theresa Creek, cultivation only accounts for 7% of the total land use in the Fitzroy River basin, with 80% of the basin grazed with beef cattle.

Evidence suggests that the dominant form of channel erosion is gully headcut and sidewall erosion. Hillslope erosion from uncultivated land (i.e., grazed pasture/woodland) is a comparatively minor contributor of sediment to the river network. These findings are in contrast to previous research that has assessed hillslope erosion from grazing land to be the principle contributor of sediment from the dry- tropical river systems draining to the Great Barrier Reef.

Sediment sources in the catchment have varied in response to land use changes over the last 150 years since European settlement. A phase of channel erosion began in the late nineteenth century in response to grazing pressure and appears to still be contributing a large proportion of the catchment’s sediment yield. During the mid-twentieth century, in response to the introduction of intensive cultivation, the catchment’s sediment load became dominated by material derived from basaltic-based cultivated land.

Improvements in land management practices have probably resulted in a decrease in sediment yield from cultivated areas in the latter half of the twentieth century.

Ongoing improvements in cultivation practices are likely to result in reduced sediment yields; however, catchment managers also need to focus erosion prevention and rehabilitation on channel sources. Because of the region’s climate, rehabilitation of gully banks by revegetation may have limited effect. An alternative approach may be to reassess stocking densities within the catchment so as to retain as much groundcover as practicable and, in particular, avoid grazing areas of high gully erosion risk.

51 Chapter 2 - Sediment source changes over the last 250 years

This study provides field-based information on patterns of erosion in northern Australia and supports previous claims that channel erosion is the largest contributor of sediment to many Australian rivers. Further work to determine the relative contribution of the different forms of channel (gully and riverbank) erosion is warranted.

52 CHAPTER 3

Determining floodplain sedimentation rates using 137Cs in a low fallout environment dominated by channel- and cultivation-derived sediment inputs, central Queensland, Australia

Cultivated land in the Theresa Creek catchment

A version of this chapter has been published in the Journal of Environmental

Radioactivity as: Hughes, A.O., Olley, J.M., Croke, J.C., Webster, I.T., 2009.

Determining floodplain sedimentation rates using 137Cs in a low fallout environment dominated by channel- and cultivation-derived sediment inputs, central Queensland,

Australia, doi 10.1016/j.jenvrad.2009.06.011. Chapter 3 – Determining floodplain sedimentation rates using 137Cs

3.1. Introduction

Caesium-137 (half-life 30.2 years) is a fallout radionuclide produced by atmospheric testing of nuclear weapons. Fallout in the northern hemisphere was up to ten times higher than in the southern hemisphere (Figure 3.1). In both hemispheres most fallout occurred during the testing period between 1954 and 1972. Since the early 1980s fallout rates have been very low (Walling and He, 1997). As well as varying between hemispheres, the distribution of fallout 137Cs varies with latitude and rainfall, tending to be higher in the mid-latitudes (UNSCEAR, 2000) and in higher rainfall zones (Basher,

2000; Schuller et al., 2002).

160

140

120

100 Northern hemisphere Southern hemisphere 80

60 Cs deposition (PBq) 40 137

20

0 1952 1954 1956 1958 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 Year

Figure 3.1. Deposition of nuclear weapon testing-derived 137Cs in the northern and southern hemispheres from 1954 - 1989 (source: UNSCEAR, 2000).

Caesium-137 has been widely used to determine floodplain sedimentation rates in temperate environments, particularly in the northern hemisphere (e.g., Walling and He

1997; Goodbred and Kuehl, 1998; Owens and Walling, 2002; Pavlovic et al., 2005).

These are regions where fallout rates are relatively high and the rainfall generally more

54 Chapter 3 – Determining floodplain sedimentation rates using 137Cs spatially uniform than in the dry, low-latitude southern hemisphere area examined in this study. This study examines the applicability of 137Cs for determining sedimentation rates in Theresa Creek, a subcatchment of the Fitzroy River basin, central Queensland,

Australia (Figure 3.2). The two most commonly used methods for determining floodplain sedimentation rates using 137Cs, depth profiles and total inventories, are tested.

Previous floodplain studies based in the southern hemisphere have almost exclusively reported rates of sedimentation using the 137Cs depth profile method (e.g., Ormerod,

1998; Cisternas et al., 2001). There appears to be only one published study that reports the use of the total 137Cs inventory method to assess floodplain sedimentation rates from a southern hemisphere catchment (i.e., Amos et al., in press). Amos et al. (in press) used the technique to quantify floodplain deposition using bulked core samples to a depth of 50 cm. The resulting core inventories had large measurement errors and it was not possible to differentiate the inventories of reference sites from those of floodplain sites. They suggested that it may be possible to obtain useful total 137Cs inventory data when they are calculated from detailed depth profile data.

3.2. Study Site

Theresa Creek is located within the Nogoa River subcatchment in the western Fitzroy

River basin (Figure 3.2). The main population centre in the catchment is Clermont

(population ~ 2000) which lies 300 km west of Rockhampton (population ~ 65 000), the largest centre in the Fitzroy River basin. The climate is dry-tropical (Köppen climate classification: BSh). The mean annual rainfall at Clermont is 649 mm (sd = 240 mm) with most rainfall occurring between November and March. Mean annual potential evaporation at Clermont is ~ 2080 mm (BoM, 2007). Most large runoff events are a

55 Chapter 3 – Determining floodplain sedimentation rates using 137Cs result of tropical low pressure systems that periodically make landfall. For a more detailed physiographic description of the study catchment see Chapter 2.

Figure 3.2. Theresa Creek catchment and tributaries showing the floodplain and reference site sampling sites, local climate stations and gauging station 130210A.

56 Chapter 3 – Determining floodplain sedimentation rates using 137Cs

Sediment sources in Theresa Creek are dominated by channel (principally gullies) and cultivation sources; sheetwash and rill erosion from uncultivated land contributes little sediment to the river system (Chapter 2). Channel sources have probably been the major source of sediment since a phase of gully incision began in the late nineteenth century. Cultivation sources became important from the 1950s and 1960s when the basalt-derived black clay-based (Vertisols) parts of the catchment were rapidly developed for crops such as sorghum, sunflowers and wheat (Chapter 2).

Channel sources and cultivated land deliver sediment with low concentrations of 137Cs to the river system. Fine sediment (< 10 µm) derived from channel sources within

Theresa Creek has near zero 137Cs concentrations (0.0 ± 0.4 Bq kg-1). The concentration of 137Cs from cultivation-derived fine sediment is also very low (0.7 ± 0.3 Bq kg-1), which is an order of magnitude lower than from uncultivated land (7.3 ± 3.3 Bq kg-1).

Lower concentrations of 137Cs from cultivation-derived sources are typically the result of dilution caused by mixing of soil layers during tillage (Wallbrink and Murray, 1993;

He and Owens, 1995). Although extensive and deep (~ 50 cm) ploughing of the

Vertisols has contributed to the lower fallout radionuclide concentrations, these cannot be explained by tillage dilution alone. The lower than expected concentrations may be due to the highly disturbed nature of these cultivated soils that have undergone extensive redistribution during the construction of erosion control measures, such as contour banks.

3.3. Using 137Cs to determine rates of floodplain sedimentation

Accumulation of 137Cs on a floodplain is associated with the deposition of fallout 137Cs from the atmosphere and sediment containing 137Cs derived from the upstream catchment during overbank flood events (Walling and He, 1997). A clear understanding

57 Chapter 3 – Determining floodplain sedimentation rates using 137Cs of the processes controlling the vertical distribution of 137Cs in floodplain sediments is crucial for using the 137Cs profile to date the sediments (Walling and He, 1997). The two most commonly used methods for determining floodplain sedimentation rates using

137Cs are sediment depth profiles and total 137Cs inventories.

3.3.1. Total 137Cs inventory

The total 137Cs inventory approach is based on the premise that the amount of deposition, or erosion at a site, can be determined by comparing the total 137Cs content of a sediment core to that from “reference” site (i.e., undisturbed, non-erosional, non- depositional sites; Owens and Walling, 1996; Walling and He, 1997). An inherent assumption in the total 137Cs inventory approach is that any sediment deposited on the floodplain since fallout began will carry some measurable quantity of 137Cs. Reference sites are assumed to have only received direct fallout of 137Cs from the atmosphere. A floodplain sediment core with a total 137Cs inventory greater than that at the reference site is indicative of deposition. A total inventory less than the reference site indicates erosion. Models have been developed to estimate sedimentation rates using total inventory data (e.g., Walling and He, 1997; Chappell, 1999), although core inventories can simply be used to provide a qualitative assessment of whether a site has eroded or accreted. This approach has the advantage of potentially only requiring one measurement of 137Cs concentration per floodplain site.

3.3.2. Caesium-137 depth profiles

The 137Cs depth profile method entails detailed measurements of 137Cs concentrations down the sediment core, typically at 1 – 2 cm increments, and then assigning an age to particular concentration levels. In the northern hemisphere a 137Cs activity peak in sediment core profiles is typically assigned to 1963, the year of maximum 137Cs fallout

58 Chapter 3 – Determining floodplain sedimentation rates using 137Cs

(e.g., Ritchie and McHenry, 1990; Walling and He, 1997). The first appearance of 137Cs in core profiles has also been used to represent 1954, the year of first detectable levels of fallout (e.g., Allison et al., 1998; Ritchie et al., 2004). In the southern hemisphere, because of the small number of atmospheric nuclear tests, a definitive activity peak is not usually detectable in sediment core profiles (Pfitzner et al., 2004; Leslie and

Hancock, 2007). Consequently, the first detectable concentration of 137Cs in core profiles is considered to be the more reliable marker. Leslie and Hancock (2007) calculated that for latitude 30 – 40° S the year of first detection for 137Cs in a sediment core is 1955 (assuming a minimum detectable activity of 0.4 Bq kg-1). This also assumes that there has been no significant post depositional migration of the radionuclide down the sediment profile.

Downward migration of 137Cs in soil profiles, by processes such as, diffusion, leaching and bioturbation has been noted by previous studies (e.g., Owens et al., 1999; Szerbin et al., 1999; Tyler et al., 2001; Doering et al., 2006) and should be accounted for if accurate sedimentation rates are to be calculated. Caesium-137 migration rates at deposition/erosion sites can be estimated by determining the rate, or total amount, of migration at nearby reference sites. A number of approaches have been used by previous researchers to account for 137Cs migration rates (e.g., Owens et al., 1996;

Walling and He, 1997; Ormerod, 1998; Ritchie et al., 2004). Migration rates have been determined here using a modification of the method used by Walling and He (1997).

In this study it is assumed that the distribution of 137Cs in both soil and floodplain cores can be described by the advection-diffusion equation of Walling and He (1997). Where variation of 137Cs concentration tzC ),( (Bq cm-3) with depth z (cm) and time t (y) in the sediment/soil profile is represented by:

59 Chapter 3 – Determining floodplain sedimentation rates using 137Cs

∂C(,)(,)(,) z t ∂ ∂ C z t  ∂ VC z t =D  − − λC(,) z t (3.1) ∂t ∂ z ∂ z  ∂ z

Where λ (y-1) is the decay constant of 137Cs, D (cm2 y-1) is the effective diffusion coefficient, and V (cm y-1) is the effective convection rate of 137Cs in sediment/soil. V

is defined as the sum of the sediment deposition rate ( vd ) and the downward migration

137 rate ( vs ) of Cs in the sediment/soil profile due to bioturbation etc. such that:

+= vvV sd (3.2)

V is assumed to be uniform down the profile, but varies with time depending on deposition rate.

The incorporation of the migration rate ( vs ) into V is necessary to take account of the

137 downward migration of Cs in undisturbed soils. The value of D and vs has been assumed to be equal to those estimated for 137Cs profiles in undisturbed catchment soils.

Equation 1 was discretised on a regular grid using central differences for the diffusion terms and upwind differencing for the advection term (Roache, 1982). The required boundary conditions are the specified flux of 137Cs to the sediment surface and zero concentration of 137Cs at depth within the profile. The system was solved using a

Microsoft Excel spreadsheet with a vertical grid size of 0.5 cm and a time increment of

0.25 years.

3.4. Sample collection and laboratory analysis

Floodplain cores for 137Cs analysis were extracted from three catchment locations; upper Theresa Creek (UTC-F), lower Theresa Creek (LTC-F) and Capella Creek (CC-F)

60 Chapter 3 – Determining floodplain sedimentation rates using 137Cs

(Figure 3.2). Cores were obtained using a vehicle-mounted hydraulic soil coring rig which extracted 80 cm long cores with a diameter of 76 mm.

Much of the catchment has been subject to land disturbance such as vegetation clearance (often by bulldozer raking) and/or cultivation, therefore locating suitable reference core locations was problematic. Only one suitable reference location was identified within the Theresa Creek catchment. Two other cores were obtained from undisturbed sites within 15 km of the catchment boundary (Figure 3.2). It is acknowledged that more reference cores would be required to provide increased statistical confidence to the calculated reference core inventories (cf. Sutherland, 1996).

Caesium-137 analysis is time consuming therefore to permit detailed incremental analysis of each floodplain and reference site core, it was only possible to analyse a limited number of cores.

Floodplain and reference cores were sectioned into 2 cm increments to a depth of 30 cm. The sediment from each increment was oven-dried and then sieved to remove material greater than 2 mm before being ground in a tungsten ring mill. Caesium-137 is preferentially bound to smaller particle sizes (Ritchie and McHenry, 1990; Spezzano,

2005). To minimise the effect that varying increment/core grain size distributions have on 137Cs concentrations some previous studies have only analysed the < 63 µm fraction

(e.g., Walling and He, 1997; Owens et al., 1999). As detection of the deepest point of

137Cs penetration was the objective in this study, fractionation was not required.

Retention of all sediment < 2 mm in size also allowed the determination of true total

137Cs inventories.

Caesium-137 analysis using high resolution gamma spectrometry, as described by

Murray et al. (1987), was carried out at the University of New South Wales gamma

61 Chapter 3 – Determining floodplain sedimentation rates using 137Cs spectrometry laboratory in Canberra, Australia. The 137Cs activity of each sample was determined using the 661.7 keV peak after subtraction of the 214Bi peak. The detector was calibrated with International Atomic Energy Agency (IAEA) reference materials to enable calculation of 137Cs concentrations. Uncertainties in concentration were calculated using a two standard deviation counting error (95% confidence interval).

Sample masses ranged between 50 and 80 g and count times were typically 48 h.

3.5. Results

3.5.1. Expected total 137Cs inventories

To determine the expected total 137Cs inventories in the region the fallout history and fractional deposition for individual latitudinal bands of the Earth were determined

(following Leslie and Hancock, 2007). For each year the total fallout (Table 10 in

UNSCEAR, 2000) was partitioned according to the fractional deposition of radionuclides based on strontium-90 (90Sr) fallout records (Table 8 in UNSCEAR,

2000) and then normalised to PBq m-2 for each of the Earth’s latitudinal bands based on

-2 each band’s surface area. The cumulative radioactivity fallout (Ft), Bq m deposited at a point in time (t) in the past corrected for radioactive decay to 2006 was then calculated by:

t eFF −λ −t)2006( t = ∑ i )( (3.3) i=1

-2 where Fi is the fallout activity (PBq m ) in the ith year. The first year of fallout (i = 1) is taken as 1951 (fallout prior to this year is assumed to be negligible). Figure 3.3 shows the expected fallout accumulation curves for the 10 - 20° and 20 - 30° south latitudinal bands. The study catchment is at ~ 23° S, close to the boundary between

62 Chapter 3 – Determining floodplain sedimentation rates using 137Cs these latitudinal bands, the total 137Cs inventories at undisturbed sites in this region should therefore lie between ~ 234 and 390 Bq m-2.

500

) o o

-2 20 - 30 s 400

300 23o s

200 10o - 20o s

Cs Cumulative Cumulative falloutCs m (Bq 100 137

0 1950 1960 1970 1980 1990 2000 Year

Figure 3.3. The expected 137Cs fallout accumulation curves for the 10 - 20° and 20 - 30° south latitudinal bands based on UNSCEAR (2000) data, and that estimated for Theresa Creek catchment (23° S) using the UNSCEAR (2000) and measured soil core inventories.

3.5.2. Reference cores

The 137Cs depth profiles for the three reference sites indicate that most 137Cs is concentrated in the top 6 cm of the soil (Figure 3.4; Appendix 4). The profiles are similar to those from reference sites in other studies (e.g., Wallbrink and Murray, 1993;

Owens and Walling, 1996). The presence of 137Cs below the top few centimetres of soil can be attributed to migration through the soil profile. For the reference cores, 137Cs is present in the top 8, 4, and 6 cm for reference cores REF1, REF2 and REF3, respectively. REF1 was collected from an area that is dominated by clay-rich Vertisols,

63 Chapter 3 – Determining floodplain sedimentation rates using 137Cs

(A) 137Cs concentration (Bq kg-1) (B) 137Cs concentration (Bq kg-1) (C) 137Cs concentration (Bq kg-1) 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10

0-2 0-2 0-2

2-4 2-4 2-4

4-6 4-6

Depth (cm) 4-6 Depth (cm)Depth Depth (cm)Depth

6-8 6-8 6-8

8-10 8-10 8-10

Figure 3.4. Caesium-137 depth profiles for the three reference core sites: A) REF1, B) REF2 and C) REF3. The error bars represent measurement precision (95% confidence limits).

64 Chapter 3 – Determining floodplain sedimentation rates using 137Cs while REF2 and REF3 were extracted from sandy loam soils. The deeper penetration of

137Cs in REF1 is probably due to the self-mulching nature of the Vertisols found within

Theresa Creek. They are susceptible to extensive cracking during dry conditions and swelling when wet (Freebairn and Boughton, 1981; Carroll et al., 1997). Accordingly,

REF1 was used to determine the D and vs (Equations 1 and 2) at the floodplain site in the Vertisol-based part of the catchment (CC-F). REF2 and REF3 were used to determine D and vs for the other two floodplain cores (UTC-F and LTC-F). The effective diffusion coefficient (D) for REF1, obtained as the best fit between measured

2 -1 and modelled profiles, was 6.0 ± 1.0 mm y , and vs , the downward migration rate of

137Cs in the soil profile due to bioturbation, was estimated to be 0.63 ± 0.03 mm y-1. For

2 -1 REF2 and REF3 the averaged D and vs were estimated to be 2.9 ± 0.4 mm y , and

0.40 ± 0.01 mm y-1 respectively. The uncertainties were estimated by fitting the advection-diffusion model to the upper and lower uncertainties (at 2 σ) of the 137Cs profile data.

The total 137Cs inventories for the three reference cores are within analytical uncertainties of one another (at 2 σ; Figure 3.5) suggesting that fallout of 137Cs was uniformly distributed throughout the catchment and local area. Near uniform fallout may be attributed to the even distribution of rainfall throughout the catchment and surrounding area (Figure 3.6). The average total inventory measured in the sediment cores is significantly (36%) lower than the calculated total 137Cs inventory of ~ 390 Bq m-2 based on UNSCEAR (2000) data for the 20 - 30° S latitudinal band and close to that calculated for the 10 - 20° S latitudinal band (~ 234 Bq m-2). This disparity probably reflects a difference in regional fallout distribution and consequently the fallout deposition rates for this site have been scaled accordingly (Figure 3.3).

65 Chapter 3 – Determining floodplain sedimentation rates using 137Cs

REF1

REF2

REF3

REF average Core ID

UTC-F

CC-F

LTC-F

0 50 100 150 200 250 300 350 400 450 137Cs inventory (Bq m-2)

Figure 3.5. Total 137Cs inventories for the reference (REF1, REF2, REF3) and floodplain cores (CC-F, UTC-F, LTC-F) collected from within and around Theresa Creek. REF average is the mean average of the reference site total 137Cs inventories. The error bars for each core represent measurement precision (95% confidence limits). The error bars for the mean reference core value represent one standard error.

1600 Capella PO Emerald PO 1400 Clermont PO Blair Athol Old Banchory Station 1200

1000

800

Rainfall (mm) Rainfall 600

400

200

0 1889 1899 1909 1919 1929 1939 1949 1959 1969 1979 1989 1999 Year

Figure 3.6. Annual rainfall (1889 – 2006) for climate stations within, and local to, the Theresa Creek catchment. Locations of climate stations are indicated on Figure 3.2.

66 Chapter 3 – Determining floodplain sedimentation rates using 137Cs

3.5.3. Floodplain cores

The 137Cs depth profiles for all three floodplain cores are indicative of accretion, with

137Cs detectable to depths below that observed in the reference site cores; 18, 8 and 16 cm for floodplain cores CC-F, UTC-F and LTC-F, respectively (Figure 3.7; Appendix

4). The 137Cs inventories of the CC-F, UTC-F and LTC-F cores are 262 ± 32 Bq m-2,

214 ± 35 Bq m-2 and 367 ± 43 Bq m-2, respectively. Comparison with the mean total

137Cs reference inventory of 246 ± 12 Bq m-2 indicates that only core LTC-F has received additional 137Cs (Table 3.1), above that expected from direct fallout, presumably associated with the deposition of sediments derived from hillslope erosion.

The 137Cs inventories of the other two cores (Table 3.1) are consistent with those expected from direct fallout with no contributions from elsewhere. Consequently if the total 137Cs inventory approach was utilised to determine sedimentation rates for these core sites only LTC-F would be unambiguously identified as a depositional site.

Table 3.1. Results of the total 137Cs inventory and 137Cs depth profile methods for determining accretion/erosion for the three sampled floodplain cores. The mean reference site inventory is 246 ± 12 Bq m-2 (n = 3; uncertainty is equivalent to one standard error).

Core ID Total 137Cs inventory method 137Cs depth profile method CC-F stable/marginal accretion detected (262 ± 32 Bq m-2) accretion UTC-F stable/marginal erosion detected (214 ± 35 Bq m-2) accretion LTC-F accretion detected (367 ± 43 Bq m-2) accretion

67 Chapter 3 – Determining floodplain sedimentation rates using 137Cs

(A) 137Cs concentration (Bq kg-1) (B) 137Cs concentration (Bq kg-1) (C) 137Cs concentration (Bq kg-1) -1 0 1 2 3 4 5 6 -1 0 1 2 3 4 5 6 -1 0 1 2 3 4 5 6

0-2 0-2 0-2

2-4 2-4 2-4 4-6 4-6 4-6 6-8 6-8 6-8 8-10

10-12 8-10 8-10

12-14 10-12 10-12 Depth (cm) Depth Depth (cm) Depth (cm) Depth 14-16 12-14 12-14 16-18 14-16 14-16 18-20 16-18 16-18 20-22

22-24 18-20 18-20

Figure 3.7. Caesium-137 depth profiles and total inventories for A) Capella Creek (CC-F), B) upper Theresa Creek (UTC-F) and C) lower Theresa Creek (LTC-F) floodplain sampling locations. The error bars represent measurement precision (95% confidence limits). The smoothed curves indicate 137Cs concentrations as determined by the advection-diffusion model.

68 Chapter 3 – Determining floodplain sedimentation rates using 137Cs

The depth profiles illustrate the low levels of 137Cs input from both direct fallout and from catchment sources, with maximum increment concentrations of < 5 Bq kg-1. At depths of 12 - 14 cm, 4 - 6 cm and 6 - 8 cm for CC-F, UTC-F and LTC-F, respectively, there appear to be peaks in 137Cs activity. For UTC-F and LTC-F, these “peaks” are within measurement error of the increment concentrations either side of it. In the case of CC-F, the rapid decline in 137Cs concentrations above the 12 - 14 cm increment

(Figure 3.7A) is likely to result from the input of sediment from cultivation sources (cf.

Walling and He, 1992). At the same time as fallout became detectable (1950s) the catchment of site CC-F was converted from mainly sheep grazing to intensive cultivation. As a result, the site has been dominated by inputs of sediment with low

137Cs concentrations.

In most northern hemisphere studies, and some from the southern hemisphere, a peak in

137Cs activity is attributed to the peak in atmospheric fallout in (1963 for the northern hemisphere and 1964 for the southern hemisphere). It is argued here that because of the potential effect of catchment inputs as well as the lack of a pronounced peak in southern hemisphere fallout rates (Figure 3.1) it is inappropriate to assign the year 1964 to peaks in 137Cs activity in floodplain cores in this environment. Furthermore, the 1960s was a particularly dry decade with 8 of the 10 years (including 1964) receiving below average rainfall (Clermont PO Climate station, 1889 - 2006; BoM, 2007). Because most 137Cs fallout was associated with rainfall, this may have reduced the amount of fallout received in Theresa Creek during the peak fallout period.

To estimate sedimentation rates for each of the floodplain sites the 137Cs profile data was fitted using the advection-diffusion model described above (Equation 1). Annual fallout data for latitude 23° S was used as the direct fallout input for all of the cores.

69 Chapter 3 – Determining floodplain sedimentation rates using 137Cs

For core LTC-F, the additional 137Cs above that expected from direct fallout (assumed to be from hillslope erosion), was estimated to be 0.06 of the annual fallout in those years in which sediment deposition occurred. This value was obtained by minimising the difference between the modelled and measured total inventory for the core. The

best estimate of the sediment deposition rate vd for each core was determined by minimising the sum of squares of the deviations of the measured and modelled concentrations (Bq kg-1) for each core segment. For the purpose of this modelling, sedimentation was only allowed to occur in years in which overbank flooding occurred.

Data regarding the occurrence of overbank flood events was acquired from gauging station 130210A (1972 – 2006; Figure 3.2). Bankfull discharge at 130210A is ~ 660 m3 s-1. This was exceeded on seven occasions between 1972 and 2006 (1973, 1976, 1978,

1983, 1994 and 1999). There is no river gauging record for Theresa Creek catchment prior to 1965; however, data from a gauging station on the nearby Nogoa River

(gauging station 130202A) extends back to 1949 and indicates significant discharge events occurred in 1954, 1955 and 1956. In the absence of any other data regarding the timing of overbank events it is therefore assumed that the three floodplain sites have been inundated on these nine occasions since 137Cs fallout commenced in 1951. The

137Cs depth distributions determined by the advection-diffusion model are illustrated in

Figure 3.7. Factoring in the downward migration determined at the reference core sites, the CC-F, UTC-F and LTC-F sites have accreted at an average rates of 2.1 ± 0.1 mm y-

1, 0.7 ± 0.2 mm y-1, and 1.4 ± 0.1 mm y-1, respectively (Table 3.2). These sedimentation rates are for the period between the year of the first known overbank discharge event

(1954) after the first year of significant 137Cs fallout (1951) to 2006, the year of core extraction.

70 Chapter 3 – Determining floodplain sedimentation rates using 137Cs

Table 3.2. Total accretion and accretion rates as determined by the advection-diffusion model for the three floodplain cores. The accretion rates are for the period between the year of the first known overbank discharge event (1954) after the first year of significant fallout (1951) to 2006, the year of core extraction.

Core ID Depth of 137Cs Depth of 137Cs Total accretion Accretion rate detection (cm) detection at reference since ca. 1954 (mm y-1) sites (cm) (mm) CC-F 18 10 (REF1) 109.2 2.1 ± 0.1 UTC-F 8 4 - 6 (REF2 & REF3) 36.4 0.7 ± 0.2 LTC-F 16 4 – 6 (REF2 & REF3) 72.8 1.4 ± 0.1

3.6. Discussion

The results presented here illustrate the limitation of using the total inventory method for inferring deposition rates. Interpretation of the 137Cs depth profile indicates that accretion has occurred at all three sites. Unequivocal accretion, however, is only apparent at one site (LTC-F) using the total inventory method (Table 3.2). It is hypothesised that the disparity between the two methods for inferring deposition rates arises from most of the sediment deposited at each of the sites being sourced primarily from channel erosion or other sediment sources which have low fallout radionuclide concentrations (e.g., cultivation-derived sediment). These results demonstrate that the inventory approach is inappropriate for determining sedimentation rates in environments in which sediments are primarily derived from areas with low fallout radionuclide concentrations.

In contrast this chapter has demonstrated that sedimentation rates can be predicted from the 137Cs depth profile method using the advection-diffusion model described here, which is a modification of the approach of Walling and He (1997). The model requires four key inputs: (i) the annual 137Cs fallout (Bq m-2 y-1) – determined from the

UNSCEAR (2000) data calibrated to the site using the total 137Cs inventory of reference

71 Chapter 3 – Determining floodplain sedimentation rates using 137Cs site cores, (ii) reference site core 137Cs profile(s) (resolved to 2 cm or better), (iii) sediment core 137Cs profile(s) (resolved to 2 cm or better), and (iv) the timing of overbank flood events. Diffusion and advection as a result of soil processes are estimated from the fallout data and the 137Cs profiles from reference site cores. These are then used as inputs, in combination with the timing of overbank deposition to enable the sedimentation rates to be determined from the sediment core 137Cs profile. One of the major advantages of the technique is that it does not require the year of first detection for 137Cs to be known.

3.7. Conclusions

Caesium-137 depth profiles and total 137Cs core inventories have been assessed for their suitability as tools for assessing medium term (last ~ 50 years) floodplain sedimentation rates in a dry-tropical, low fallout environment that is dominated by inputs from channel erosion and cultivated land. Advection-diffusion modelling of the 137Cs depth profiles in this environment has been demonstrated to be an effective way of determining sedimentation rates. In contrast low fallout levels, together with the dominance of sediment sources with low 137Cs concentrations, result in depositional sites having total inventories that are indistinguishable from undisturbed reference sites. Although, the total 137Cs inventory approach cannot be unequivocally discounted as a viable approach in low fallout environments per se, the compounding effect of sediment sources that are low in 137Cs make its application unsuitable in this setting.

72 CHAPTER 4

Changes in the rates of floodplain and in-channel bench accretion in response to catchment disturbance, central Queensland, Australia

Sediment core extraction from an in-channel bench, lower Theresa Creek.

A version of this chapter has been accepted for publication in Geomorphology as:

Hughes, A.O., Croke, J.C., Pietsch, T.J., Olley, J.M. Changes in the rates of floodplain and in-channel bench accretion in response to catchment disturbance, central

Queensland, Australia. Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

4.1. Introduction

“New world” locations, where large-scale catchment disturbance occurred relatively recently, provide ideal conditions to assess the effects of catchment disturbance on sediment loads (e.g., Leahy et al., 2005; Knox, 2006). Ideally, assessing the impact of human activity in catchments involves direct measurements of river sediment loads by catchment monitoring (e.g., Yang et al., 2002; Walling and Fang, 2003). However, when no long-term catchment monitoring data is available the sedimentary record contained within catchment sediment sinks, such as floodplains, lakes, and reservoirs can be an important source of information on changes in catchment sediment flux as a result of land use disturbance (Owens et al., 1999). Floodplains in particular, are one the most important catchment sediment sinks, storing up to 50% of a catchment’s annual sediment load (Trimble, 1983; Phillips, 1991).

Despite having ideal conditions (i.e., known settlement timing and extensive historical land use records) for the investigation of the effects of catchment disturbance on sediment flux, few such studies have been carried out in Australia and most of these have been undertaken within the temperate catchments of southeastern Australia (e.g.,

Olley and Wasson, 2003; Leahy et al., 2005). Little attention has been paid to northern dry-tropical catchments. Given the increased interest in quantifying sediment yields from tropical rivers in Australia, particularly those draining to the Great Barrier Reef

(GBR), there is a need to improve our understanding of the role of floodplains, and other sediment stores, in the sediment budgets of these catchments.

Model-based predictions of sediment yield from GBR catchments suggest increases of four to ten times since European settlement (e.g., Neil et al., 2002; McCulloch et al.,

2003; McKergow et al., 2005). The resulting predicted increase in suspended sediment

74 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion entering the GBR may be degrading water quality and adversely affecting marine ecosystems (Bramley and Roth, 2002; Neil et al., 2002). Many of these predictions are derived from catchment or regional scale modelling studies and make critical assumptions with regards to pre-disturbance erosion rates and/or sediment delivery ratios. In addition they have generally not been validated by field-based data.

Other attempts to assess the effect of European land use activities within GBR catchments include plot-based erosion studies from agricultural areas (e.g., McIvor et al., 1995; Scanlan et al., 1996). These studies show that areas affected by European land use activities generally have high rates of erosion. However, these results are of limited use in assessing the impact of land use changes on catchment sediment flux as they do not consider sediment delivery factors, such as hillslope or floodplain and channel storage (Walling, 1983).

This study examines the floodplain sedimentation rates within Theresa Creek, a ~ 6000 km2 subcatchment of the Fitzroy River basin in central Queensland, Australia (Figure

4.1). Single-grain optically stimulated luminescence (OSL) dating and caesium-137

(137Cs) soil depth profiles are used to determine sedimentation rates for three floodplain and two in-channel bench sites. The principal objectives of this chapter are to: (i) report field-derived alluvial sedimentation rates from a dry-tropical catchment - an environment that is poorly represented in the international literature; (2) assess the importance of floodplains and in-channel deposits as sinks within such an environment;

(iii) determine to what extent catchment disturbance has impacted alluvial sedimentation rates; and (iv) determine whether measured floodplain sedimentation rates support previously published estimates of large increases in post-disturbance sediment yield.

75 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

Figure 4.1. Theresa Creek catchment showing the locations of the floodplain core sites and gauging stations.

4.2. Study Site

Theresa Creek is located within the Nogoa River subcatchment in the western Fitzroy

River basin (Figure 4.1). The main population centre in the catchment is Clermont

(population ~ 2000) which lies 300 km west of Rockhampton (population ~ 65 000), the largest centre in the Fitzroy River basin. The climate is dry-tropical (Köppen climate classification: BSh). The mean annual rainfall at Clermont is 649 mm (sd = 240 mm) with most rainfall occurring between November and March. Mean annual potential evaporation at Clermont is ~ 2080 mm (BoM, 2007). Most large runoff events are a result of tropical low pressure systems that periodically make landfall. For a more detailed physiographic description of the study catchment see Chapter 2.

76 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

Significant floodplain landforms typical of lateral migration have been described in the

Fitzroy River basin (e.g., Amos et al., 2008); however, little evidence exists of such features in the Theresa Creek subcatchment. This observation, together with the widespread occurrence of deep channels with well defined levees suggests that vertical accretion is the dominant mechanism of floodplain development within the study catchment.

Table 4.1. Characteristics of the floodplain/in-channel bench sampling sites.

Site % of catchment % of catchment Catchment Channel cross- Bench cross- basalt/ cultivated/ area sectional area sectional non-basalt Grazed (km2) (m2) area (m2) UTC 0/100 0/98 560 1100 60 × 2 LTC 25/75 20/74 6000 420 30 CC 85/15 53/45 420 54 0

Three sites from upper Theresa Creek (UTC), lower Theresa Creek (LTC) and Capella

Creek (CC) were selected for detailed analysis of sedimentation rates (Figure 4.1; Table

4.1). The sample sites were selected so as to provide a range of the of predominant physiographic/land use conditions. The sites are also representative of the major floodplain morphological units and include near channel (levee) and distal locations.

While the collection of multiple cores from each site would have been ideal, the time and cost of core processing prohibited this. Accordingly, it is assumed that the information obtained from each core is representative of conditions at each site.

The channel beds at both UTC and LTC are composed of quartz dominant sand to gravel sized sediment which is transported in waves composed of quartz dominant sand to gravel sized sediment which is transported in waves or mini-sand slugs (cf. Bartley et al., 2007), which is typical of the channels draining the non-basaltic terrain of the

Theresa Creek system. Interestingly, the channel at LTC has a significantly smaller

77 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion cross-sectional area than UTC despite draining an order of magnitude greater catchment area (Table 4.1; Appendix 5). Both UTC and LTC have in- channel benches that extend

~ 3.5 - 4 m above the channel bed and are between 10 - 20 m wide (Figure 4.2;

Appendix 5). The benches are composed mainly of silty-sand and are well vegetated with grass and/or large Eucalyptus sp. trees. The channel at the Capella Creek site is considerably smaller than the Theresa Creek sites and is composed almost entirely of dark-brown silt to clay sized sediment that has eroded from the Vertisols that dominate the eastern side of the catchment (Table 4.1). Although the surrounding land is intensively cultivated, the riparian areas and immediate floodplains are generally well vegetated with large Eucalyptus sp. trees.

Figure 4.2. Example of an in-channel bench on the main tributary of Theresa Creek.

78 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

4.3. Methods

Alluvial sedimentation rates were determined by both single-grain OSL dating and floodplain depth profiles of the fallout radionuclide 137Cs. OSL dating has proved to be a reliable method for dating fluvial sediments (e.g., Zhang et al., 2003; Olley et al.,

2004). Recent advances in analysis techniques have improved the confidence in dating very young (< 200 y) deposits (e.g., Galbraith et al., 1999; Olley et al., 2004) and as a result more researchers are using the technique (e.g., Madsen et al., 2005; Page et al.,

2007; Rustomji and Pietsch, 2007; Thoms et al., 2007). In many of these studies, alternative dating techniques have been used to provide some validation of OSL ages

(e.g., Madsen et al., 2005; Rowland et al., 2005; Rustomji and Pietsch, 2007). In this study, 137Cs depth profiles and the presence/absence of 137Cs in samples taken for OSL provide some verification of the general accuracy of the youngest OSL ages.

4.3.1. Caesium-137 depth profiles

Caesium-137 (half-life 30.2 y) is a fallout radionuclide that was a product of atmospheric nuclear weapons testing that occurred from the 1950s to 1970s. It has been used widely to determine floodplain sedimentation rates in different parts of the world

(e.g., Goodbred and Kuehl, 1998; Owens and Walling, 2002; Pavlovic et al., 2005). On deposition, usually by rainfall, 137Cs binds closely with fine sediment particles.

Caesium-137 present in floodplain material is a result of both direct fallout from the atmosphere and from sediment deposited during overbank flood events. Sedimentation rates are usually calculated by determining the amount of deposition since an identifiable activity level in 137Cs in a soil profile. For a detailed description of the use of 137Cs to determine floodplain sedimentation rates see Walling and He (1997).

79 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

Cores for 137Cs depth profile analysis were obtained using a vehicle-mounted hydraulic soil coring rig which extracted 80 cm long cores with a diameter of 76 mm. Cores were extracted from each of the floodplain/levee sites (UTC-F, LTC-F and CC-F).

In the laboratory each core was sectioned into 2 cm increments. The sediment from each increment was oven-dried and sieved to remove material greater than 2 mm. High resolution gamma spectrometry (Murray et al., 1987) was used to determine 137Cs concentrations for each increment. The sedimentation rate for each core since 1954 was predicted from the resulting 137Cs depth profiles using the advection-diffusion model described in Chapter 3 (Section 3.3.2).

4.3.2. Optical dating

The premise of optical dating is that, when buried, quartz grains begin to accumulate a trapped-charge population that increases in a measurable and predictable way in response to the ionising radiation dose to which the grains are exposed. Exposure to sunlight releases the light-sensitive trapped charge, thereby resetting the OSL signal; a process commonly referred to as ‘bleaching’. The time elapsed since sediment grains were last exposed to sunlight may be estimated by measuring the OSL signal from a sample of sediment, determining the equivalent dose (De) that this represents (for which the SI unit is the gray, Gy), and estimating the rate of exposure of the grains to ionising radiation averaged over the period of burial. The latter parameter of interest is termed the dose rate (Dr). The burial age of well-bleached grains may then be obtained from the following equation:

Burial age = De / Dr (4.1)

80 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

The OSL signal of clean quartz grains exposed directly to sunlight is reduced to a negligible level within a few seconds (Wintle, 1997; Aitken, 1998). However, incomplete or non-uniform bleaching is commonplace in many depositional environments (Murray and Olley, 2002), due to surface coatings on the grains and/or insufficient exposure to sunlight during sediment transport. This results in grains being deposited with a heterogeneous distribution of residual trapped charge, and a correspondingly wide range of measured De values. For such sediments, the population of grains with the lowest measured De values provides the most accurate estimate of Db: the burial dose to which those grains that were well bleached at deposition have been exposed since the most recent transport event (Olley et al., 2004).

At each site a one metre deep pit was excavated and sediment sampled for OSL dating by driving a stainless steel tube into the sidewall. Samples were extracted from various depths and sealed to remain lightproof. At the upper and lower Theresa Creek sites, material for dating was acquired from floodplain levees (UTC-F and LTC-F) and in- channel benches (UTC-B and LTC-B). The Capella Creek site did not have an in- channel bench, as a result, dateable material was only obtained from the floodplain (CC-

F).

Sand-sized grains of quartz (180 - 212 µm) were extracted and prepared using the methods described by Olley et al. (2004). A modified version of the single-aliquot regenerative dose protocol (SAR), as developed by Olley et al. (2004), was used to obtain the De distributions for each sample. Sample preparation and analysis was carried out at the CSIRO Land and Water Optically Stimulated Luminescence

Laboratory in Canberra. For samples with burial doses > ~ 0.5 Gy and no evidence of partial bleaching, sample burial doses were estimated using the central age model

81 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

(Galbraith et al., 1999). All other sample burial doses were estimated using a probability density summation method described by Rustomji (2006).

The contribution of lithogenic radionuclides to the Dr was obtained by high-resolution gamma spectrometry analysis of the sediment immediately surrounding the sampled quartz grains. Cosmic dose rates were determined following Prescott and Hutton

(1994). The dry dose rates were adjusted, following Aitken (1985), assuming water content of 7.5 ± 5% for all samples. See Appendix 6 for radial plots for all OSL samples and Appendix 7 for OSL sample site data.

4.4. Results

4.4.1. Floodplain and in-channel bench sediment dating

4.4.1.1. Upper Theresa Creek

At UTC-F 137Cs was detected to a depth of 8 cm (Figure 4.3A). The advection- diffusion model estimated that ~ 36 mm of sediment has been deposited since ca. 1954

(see Chapter 3). Three sediment samples were extracted for OSL dating from 15, 60 and 100 cm. The sample summary data, including calculated Dr and De are presented in

Table 4.2. Burial ages of 110 ± 15, 1010 ± 90 and 1640 ± 150 y BP were determined for 15, 60 and 100 cm, respectively.

The in-channel bench site (UTC-B) was composed of homogenous fine to coarse sized sand. Burial ages of 30 ± 5 and 65 ± 70 y BP were obtained from depths of 40 and 90 cm, respectively (Table 4.2). Despite analysing ~ 1000 grains of quartz, the luminescence response from the 90 cm sample (UTC-B-90) was poor, therefore, the resulting burial age has a large measurement error (i.e., 65 ± 70 y BP). Caesium-137 was not detectable at 90 cm (Table 4.2) which suggests that the sediment was deposited

82 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion before fallout levels became detectable (i.e., before ca. 1954). This information

+70 improves the error margin of UTC-B-90 to 65 −20 y BP.

4.4.1.2. Lower Theresa Creek

Caesium-137 at LTC-F was detected to a depth 16 cm (Figure 4.3B). The advection- diffusion model estimated that ~ 73 mm of sediment has been deposited since ca. 1954

(see Chapter 3). Four samples, from 22, 35, 50 and 95 cm, produced burial ages of 65 ±

5, 120 ± 20, 255 ± 25 and 790 ± 70 y BP, respectively (Table 4.2).

The in-channel bench at this site (LTC-B) was similar in composition to the bench at the

UTC site and was principally composed of sand-sized material. However, unlike site

UTC-B, dark brown to black clay was interspersed within the sand. No clear facies boundaries were apparent. Burial ages of 20 ± 5 and 40 ± 15 y BP were obtained from depths of 38 and 95 cm, respectively (Table 4.2).

4.4.1.3. Capella Creek

Caesium-137 at CC-F was detected to a depth of 18 cm (Figure 4.3C). The advection- diffusion model estimated that ~ 109 mm of sediment has been deposited since ca. 1954

(see Chapter 3). Burial ages of 120 ± 20, 540 ± 145 y BP and 1050 ± 140 y BP were obtained from 20, 40 and 60 cm, respectively (Table 4.2).

83 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

(A) 137Cs concentration (Bq kg-1) (B) 137Cs concentration (Bq kg-1) (C) 137Cs concentration (Bq kg-1) -2-10 1 2 3 4 5 6 -1 0 1 2 3 4 5 6 -1 0 1 2 3 4 5 6

0-2 0-2 0-2

2-4 2-4 2-4

4-6 4-6 4-6

6-8 6-8 6-8

8-10 8-10 8-10

10-12 10-12 10-12 Depth (cm) Depth Depth (cm) Depth (cm) Depth

12-14 12-14 12-14

14-16 14-16 14-16

16-18 16-18 16-18

18-20 18-20 18-20

Figure 4.3. Caesium-137 depth profiles for (A) upper Theresa Creek (UTC-F), (B) lower Theresa Creek (LTC-F) and (C) Capella Creek (CC-F) floodplain sampling locations. The error bars represent measurement precision (95% confidence limits).

84 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

Table 4.2. Burial ages for the sediment obtained from floodplain and bench locations at the three sampling locations. Also indicated is whether or not 137Cs was present in the material surrounding the dated quartz grains. The detection of 137Cs indicates that sediment was likely to have been deposited since ca. 1954.

-1 137 OSL sample Depth Dr (Gy ky ) De (Gy) No. of Burial age Statistical Cs (cm) grains (y BP) model* detectable Floodplain UTC-F-15 15 4.20 ± 0.27 0.46 ± 0.05 45 110 ± 15 PDS no UTC-F-60 60 3.78 ± 0.29 3.82 ± 0.10 66 1010 ± 90 CE no UTC-F-100 100 3.71 ± 0.29 6.08 ± 0.24 55 1640 ± 150 CE no

LTC-F-22 22 3.05 ± 0.24 0.19 ± 0.01 61 65 ± 5 PDS no LTC-F-35 35 3.32 ± 0.25 0.39 ± 0.05 18 120 ± 20 PDS no LTC-F-50 50 3.71 ± 0.27 0.94 ± 0.05 39 255 ± 25 CE no LTC-F-95 95 3.54 ± 0.26 2.80 ± 0.10 98 790 ± 70 CE no

CC-F-20 20 0.81 ± 0.07 0.10 ± 0.01 50 120 ± 20 PDS no CC-F-40 40 0.75 ± 0.06 0.40 ± 0.10 118 540 ± 145 PDS no CC-F-60 60 0.74 ± 0.06 0.77 ± 0.08 38 1050 ± 140 CE no

In-channel bench UTC-B-40 40 3.65 ± 0.26 0.12 ± 0.01 40 35 ± 5 PDS yes UTC-B-90 90 2.85 ± 0.19 0.19 ± 0.04 40 65 ± 70 PDS no

LTC-B-38 38 2.95 ± 0.21 0.06 ± 0.01 23 20 ± 5 PDS yes LTC-B-95 95 3.01 ± 0.22 0.12 ± 0.04 25 40 ± 15 PDS yes

* PDS = probability density summation method, CE = central age model

85 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

4.4.2. Estimated floodplain accretion rates

The 137Cs dates from the floodplain cores and OSL samples suggest that the youngest

OSL burial ages are reliable. For all OSL burial ages younger than 1954, 137Cs was detected in the sediment surrounding the OSL samples while the radionuclide was undetectable in all of the pre-1954 burial ages (Table 4.2). In addition, the 137Cs depth profile determined sediment ages do not overlap with the OSL burial ages, providing further confidence that the OSL burial ages are reliable. Furthermore, at the precision levels achievable within the present study, and where the methods cover the same time interval, accretion rates determined using OSL are equivalent to those determined using

137Cs depth profiles. This general agreement between two independent methods provides encouragement for accepting the general veracity of the results.

The data indicate three depositional periods based on changes in accretion rates: (i) late

Holocene/pre-European settlement (pre- ca. 1850); (ii) early post-European settlement

(ca. 1850 – mid-twentieth century); and (iii) late post-European settlement (mid- twentieth century – present). Due to the physiographic differences between core sites and their positions relative to the river channels these periods were determined on the basis of changes in rates of accretion rather than direct comparison of absolute accretion between sites.

4.4.2.1. Late Holocene/pre-European settlement period (pre- ca. 1850)

The accretion rates at all three sites during the late Holocene/pre-European settlement period were low (≤ 0.9 mm y-1). At the upper Theresa Creek floodplain levee site

(UTC-F) the three deepest burial ages produce a linear accretion rate of 0.6 ± 0.1 mm y-

1 (Figure 4.4A). Similarly, for the lower Theresa Creek levee site (LTC-F) a linear accretion rate of 0.9 ± 0.1 mm y-1 is defined by the deepest three burial ages (Figure

86 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

4.4B). The higher accretion rate at the lower catchment site (LTC-F) may, in part, be attributed to the frequency of overbank events. Overbank events have an approximate return period of 5.7 y (660 m3 s-1) at the gauging station upstream of site LTC (gauging station 130210A, 1972-2005; Figure 4.1). No gauging data are available to calculate a return period for overbank events at site UTC. However, UTC is likely to experience less frequent overbank events as it has a significantly larger channel capacity than LTC, but a much smaller catchment area (Table 4.1).

The oldest three sediment ages from the Capella Creek floodplain site (CC-F) indicate a linear pre-disturbance accretion rate of 0.4 ± 0.1 mm y-1 (Figure 4.4C). The lower accretion rate at CC-F, compared to the Theresa Creek floodplain sites (UTC-F and

LTC-F) may be related to the greater sampling distance from the channel (~ 40 m) and/or the nature of the material transported by from the Vertisol-based catchment area which is dominated by clay material.

4.4.2.2. Early post-European settlement period (ca. 1850 – mid-twentieth century)

The first Europeans explored the Fitzroy River basin in the mid-1840s and settlement began soon after. Sheep grazing established by 1855 (Lewis et al., 2007). All three floodplain sites show increases in accretion rates during this initial post-disturbance period (Figure 4.4A-C). The accretion rates at both UTC-F and CC-F increased from the pre-disturbance rates by about three times to 1.8 ± 0.5 mm y-1 and 1.4 ± 0.5 mm y-1, respectively.

87 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

0.00 -1 0.00 (A) 0.7 ± 0.2 mm y (B)(B) 1.4 ± 0.1 mm y-1 -1 0.10 1.8 ± 0.5 mm y 0.10

0.20 -1 0.20 4.0 ± 1.2 mm y 0.30 0.30 0.40 0.40 -1 0.50 0.6 ± 0.1 mm y 0.50 0.60 -1 Depth (m) Depth

Depth (m) 0.9 ± 0.1 mm y 0.60 0.70 0.70 0.80

0.90 0.80

1.00 0.90

1.10 1.00 0 200 400 600 800 1000 1200 1400 1600 1800 0 100 200 300 400 500 600 700 800 900 Age (years BP) Age (years BP) (C) 0.00 2.1 ± 0.1 mm y-1 0.10 1.4 ± 0.5 mm y-1

0.20

0.30

0.4 ± 0.1 mm y-1

Depth (m) Depth 0.40

0.50

0.60

0.70 0 200 400 600 800 1000 1200 1400 Age (years BP)

Figure 4.4. Age-depth relationships for (A) upper Theresa Creek (UTC-F), (B) lower Theresa Creek (LTC-F), and (C) Capella Creek (CC-F) floodplain sampling locations. The vertical dashed line represents ~ 150 y BP; the time of first European settlement within the catchment.

At LTC-F an overall accretion rate of 4.0 ± 1.2 mm y-1 was calculated for the period between 118 ± 18 y BP and ca. 52 y BP. This is an approximate four times increase in accretion rates from the pre-disturbance period. Within this period, two distinct phases of accretion can be identified: (i) 120 ± 20 y BP to 64 ± 15 y BP at 3.1 ± 2.2 mm y-1, and (ii) 65 ± 15 y BP to ca. 52 y BP at 6.3 ± 2.7 mm y-1. This latter rate is affected; however, by its calculation over a relatively short period during the 1940s and 1950s.

88 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

The 1950s was a particularly wet decade, with four years receiving 50% more rainfall than the mean annual rate (BoM, 2007). There is no river gauging record for Theresa

Creek catchment prior to 1965; however, data from a gauging station on the nearby

Nogoa River (gauging station 130202A) extends back to 1949 and indicates significant overbank events in 1950, 1954, 1955 and 1956.

4.4.2.3. Late post-European settlement period (mid-twentieth century – present)

Evidence presented here indicates that from the mid-twentieth century, depending on catchment location, there were both increases and decreases in rates of floodplain accretion. At UTC-F (Figure 4.4A) and LTC-F (Figure 4.4B), the floodplain accretion rates have decreased to 0.7 ± 0.2 mm y-1 and 1.4 ± 0.1 mm y-1, respectively. Despite the decreases, both sites are still experiencing sedimentation rates higher than pre- disturbance rates. In contrast, the rate of accretion at CC-F increased by a further ~

30% from the initial post-disturbance period, with the 137Cs depth profile data indicating a post-1954 accretion rate of 2.1 ± 0.1 mm y-1.

4.4.3. In-channel bench accretion rates

Burial ages for the in-channel benches (UTC-B and LTC-B; Table 4.2) suggest that they are rapidly aggrading modern features. Because of the modern nature of the sediments, accretion rates have only been determined from the deepest date to the surface for both locations. For UTC-B, 90 cm of sediment has aggraded at a rate of 13.6 ± 6.0 mm y-1

+70 since the early twentieth century (65 −20 y BP; Table 4.2). At site LTC-B, 95 cm of sediment was deposited in the last 40 ± 15 y BP (Table 4.2) at a rate of 23.8 ± 10.4 mm y-1.

89 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

Sedimentation on the in-channel benches is an order of a magnitude greater than the floodplain sites. This can be attributed to the fact that the benches are regularly inundated by flow events. The rapid rate of accretion is also indicative of the large amount of coarse sediment (> 63 µm) that is transported during flow events. The widespread occurrence of sand to gravel sized sediment transported as “waves” or mini- sand slugs at both sites provides confirmation of the amount of coarse sediment transported in the channel during flow events (Figure 4.5).

Figure 4.5. Sand wave at the upper Theresa Creek (UTC) site. Note the fence post in the centre of the photograph for scale. Photograph was taken looking upstream.

4.5. Discussion

4.5.1. Patterns of floodplain sedimentation

Data presented in this study indicates temporal variability in rates of alluvial sedimentation within Theresa Creek. Single-grain OSL dates and 137Cs depth profiles

90 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion appear reliable and while the limitations of using depth-based estimates to determine sedimentation rates are acknowledged (see Rumsby, 2000), notable differences in sedimentation rates are apparent. The linear accretion rates at all three floodplain locations indicate that the sediment flux within Theresa Creek during the late Holocene

(up to ca. 1600 y BP) was relatively stable. This is likely to reflect the reported stable climate, and hence discharge regime, in northeastern Australia over the late Holocene

(Dimitriadis and Cranston, 2001; Lough, 2007).

In contrast, increased rates of floodplain sedimentation appear to have occurred throughout the catchment shortly after European settlement ca. 1850. These increases are attributed to the increased delivery of sediment to the river system associated with land use changes. In particular, the introduction of grazing animals would have reduced groundcover, therefore, increasing the potential for sheetwash and rill erosion (Bartley et al., 2006) and increased landscape susceptibility to gully initiation (Prosser and

Winchester, 1996). The findings reported in Chapter 2 suggest that a period of gully incision, beginning in the late nineteenth century, would have contributed large amounts of sediment to the river system.

The basaltic and non-basaltic parts of the catchment exhibit differing behaviours since the mid-twentieth century. Accretion rates at the two Theresa Creek sites (UTC-F and

LTC-F) have slowed, while increased accretion is evident at Capella Creek (CC-F).

The contrasting floodplain depositional pattern may be because of the contrasting physiography and differing land use histories. The Theresa Creek sites are sourced from a mixture of rock types. Seventy-five to 100% of the catchment area contributing to the Theresa Creek floodplain sites is composed of granite, metasediment and clastic

91 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion rocks (Table 4.1; non-basaltic). These rock types have been found to contribute a high proportion of channel-derived sediment (see Chapter 2) and these landscapes have been continuously grazed since European settlement. The Capella Creek site (CC-F) catchment is almost exclusively underlain by highly weathered Tertiary basalts which have been intensively cultivated since the mid-twentieth century (Table 4.1).

The reduction in floodplain accretion rates at the two Theresa Creek floodplain sites

(UTC-F and LTC-F) in recent times is consistent with findings reported elsewhere (e.g.,

Ritchie et al., 2004; Knox, 2006). However, many of these studies have been from agricultural catchments dominated by sheetwash and rill erosion and reduced rates of floodplain accretion were attributed to improvements in land management practices. In southeastern Australia, reductions in river sediment fluxes have been attributed to the exhaustion of gully-derived sediment by around the mid-twentieth century (Wasson et al., 1998; Olley and Wasson, 2003). Theresa Creek may be experiencing a similar response with most of the catchment upstream of the upper Theresa Creek site, and a large proportion of the catchment contributing to the lower Theresa Creek site, containing non-basaltic terrain that has experienced a phase of gully incision (see

Chapter 2). Central Queensland was settled later than southeastern Australia, therefore, gully incision is likely to have begun later and gully networks are yet to stabilise. Field observations and anecdotal reports from catchment landowners confirm that the gully network continues to expand.

Floodplain accretion rates determined at the Capella Creek site suggest the sediment flux at sites supplied by the cultivated basaltic terrain continued to increase throughout the post-European period. Up until the mid-twentieth century much of the low-lying

92 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion basaltic terrain was principally used for sheep grazing. During the 1950s there was a relatively abrupt change in land use to dryland cropping of food crops, such as sorghum

(Johnson, 1982). The findings reported in Chapter 2 attributed an increase in the proportion of basaltic-derived sediment transported by the river system since the mid- twentieth century to this change in land use. The increased rate of floodplain accretion detected in this study suggests that the land use intensification that took place resulted in increases in river sediment flux. The resolution of the sediment dating does not allow any definitive comment to be made on changes in accretion rates over the last 50 - 60 years. However, it is likely that sediment loads may have been highest shortly after the introduction of cultivation then decreased after the 1970s. This is because the 1950s and 1970s were above-average wet periods (BoM, 2007) with a number of significant flood events occurring. Furthermore, improvements in erosion control and land management practices since the 1970s are likely to have reduced the sediment yield from cultivated land.

4.5.2. Increased rates of floodplain accretion as an indicator of the impact of catchment disturbance on river sediment flux

Theresa Creek is a headwater catchment identified as having high rates of both gully and sheetwash erosion (e.g., Carroll et al., 1997; NLWRA, 2001). The granite underlain parts of the catchment have some of the highest gully densities in Australia

(Hughes et al., 2001). Despite being identified as an erosion “hotspot”, the three to four times increase in floodplain accretion rates reported here are low compared to previous studies. Most previously reported post-disturbance accretion rates are at least an order of magnitude greater than the pre-disturbance rates (Table 4.3). For example, increases of up to three orders of magnitude have been reported from southeastern Australia (e.g.,

93 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

Rustomji and Pietsch, 2007). European settlement in central Queensland took place several decades later than in southeastern Australia; however, the land uses introduced were very similar. The question remains, therefore, of why the changes observed in this study are so modest in comparison to those observed elsewhere (e.g., Brooks and

Brierley, 1997; Knox, 2006; Rustomji and Pietsch, 2007).

A major difference between the catchments of southeastern Australia and northern

Australia is that prior to European settlement many southeastern Australian rivers were characterised by deep pools with limited connecting channels referred to as “chains of ponds” (Eyles, 1977). These discontinuous channels, together with low rates of erosion, resulted in naturally very low sediment yields (Olley and Wasson, 2003). Because of the disturbance of valley floors by vegetation clearance and the introduction of grazing animals in the post-European settlement period, large scale scour occurred that obliterated most chains of ponds and resulted in a massive influx of sediment (Prosser and Winchester, 1996; Brierley and Murn, 1997; Olley and Wasson, 2003). The resulting increased efficiency of sediment delivery probably accounts for the large relative increases in floodplain sedimentation rates observed in southeastern Australia.

No historical accounts of such chains of ponds exist from the dry-tropical catchments of northern Australia. Indeed, the deep, stable channels with well-formed natural levees, which occur throughout much of the Fitzroy River basin, suggest that the river network has been highly connected over the long-term (Figure 4.6).

94 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

Table 4.3. Rates of pre- and post-disturbance alluvial accretion from previous selected studies.

Reference Location Alluvial site Pre-disturbance Post-disturbance accretion rate (mm y-1) accretion rate (mm y-1) Theresa Creek NE Australia floodplain levee 0.6 – 0.9 0.8 – 4.0 Theresa Creek NE Australia floodplain 0.4 1.4 - 2.1 Theresa Creek NE Australia in-channel bench 0* 13.6 - 23.8 Rustomji and Pietsch (2007) SE Australia floodplain levee 0.06 - 0.15 5.9 - 16.4 Rustomji and Pietsch (2007) SE Australia in-channel bench 0* 9.2 - 16.2 Brooks and Brierley (1997) SE Australia floodplain 0.75 12.5 Leahy et al. (2005) SE Australia floodplain lake 0.3 9.5 Gell et al. (2005) SE Australia floodplain lake 0.6 1 - 10 Knox (1987) USA floodplain 0.2 3 - 50 Knox (2006) USA floodplain 0.2 – 0.9 2 - 20 Phillips (1997) USA floodplain 0.05 3 - 9 Florsheim and Mount (2003) USA floodplain 1.15 - 3.0 25.0

* In-channel benches are modern features that are believed to be a response to the influx of sediment from post-settlement catchment disturbance.

95 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

Low increases in rates of floodplain sedimentation may also reflect the naturally high sediment loads characteristic of dry-tropical river systems (e.g., Jansson, 1988; Ludwig and Probst, 1996). Naturally high flood sediment concentrations in the dry-tropical catchments of northeastern Australia may be related to the cyclical nature of the regional climate that results in extended periods of drought followed by above average wet periods (Lough, 2007). During droughts, groundcover is reduced which increases the potential for erosion in any subsequent rainfall events (McIvor et al., 1995; Scanlan et al., 1996). Drought-breaking floods in particular have elevated suspended sediment loads (McCulloch et al., 2003). The Fitzroy River basin has been subjected to numerous cycles of drought followed by above average rainfall periods over the late

Holocene (Lough, 2007), hence it is likely to have experienced brief but repeated periods of greatly elevated suspended sediment load. The cyclical nature of the climate of northeastern Australia may be related may be related to the El Niño Southern

Oscillation (ENSO) and/or the Pacific Decadal Oscillation (PDO; Power et al., 1999;

Mantua and Hare, 2002).

The argument presented here, that low level increases in post-European sediment yield are related to naturally high sediment loads, is also supported by the findings of Bostock et al. (2007) from the lower parts of the Fitzroy River. Bostock et al. (2007) estimated the mean annual sediment accumulation in the Fitzroy River estuary over the Holocene and concluded that the volumes of pre-disturbance sediment yield predicted by some previous modelling studies (e.g., Furnas, 2003; McKergow et al., 2005) were insufficient to account for the sediment deposited in the estuary. In Bostock et al.

(2007), like in this study, relatively high pre-disturbance suspended sediment yields are required to explain the observed rates of sediment storage.

96 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

Figure 4.6. Photograph looking downstream at the upper Theresa Creek (UTC) site. Photograph was taken standing on the in-channel bench. Note the well-established vegetation indicating channel stability.

The modest increase in the rate of floodplain accretion is also surprising when the apparent increased storage of sediment within the channels is considered. The formation of in-channel benches and anecdotal reports of increased volumes of bed material suggest that the channel capacity within Theresa Creek may have reduced since

European settlement (cf. Page and Carden, 1998). This does not appear to have been compensated by changes in channel width, as most channels are laterally stable with well formed natural levees. Unfortunately, the limited coverage of river gauging data within the catchment can not be used to quantify precise increases in overbank event frequency.

97 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

The findings presented here, together with our understanding of the ability of large catchments to store sediment and buffer the effects of land use changes (cf. Walling,

1999), suggest that previous predictions of large increases in sediment yields (e.g., Neil et al., 2002; McKergow et al., 2005) may be overestimates within the dry-tropical catchments draining to the GBR.

4.5.3. The significance of in-channel benches as stores of post-disturbance generated sediment

While the importance of floodplains as sediment sinks within the large Fitzroy River basin has been previously noted (e.g., Amos et al., 2008), the significance of in-channel benches is poorly appreciated. In-channel benches, deposited entirely since European settlement, have been reported from catchments in southeastern Australia (e.g., Carthew and Drysdale, 2003; Sheldon and Thoms, 2006; Rustomji and Pietsch, 2007). Rustomji and Pietsch (2007) attributed the occurrence of in-channel benches to the flux of gully- derived sediment that has been delivered to the stream network since European settlement.

The bench sediment dates and accretion rates in this study support the hypothesis that they are rapidly aggrading modern depositional features. Unlike the mobile coarse sand to gravel-sized sediment transported in the channels proper (cf. Alexander and Fielding,

1997), the sediment within the benches is not readily reworked, as demonstrated by the

OSL burial age sequence and the detectabilty of 137Cs (Table 4.2).

Given the dominance of gully erosion as a sediment source (see Chapter 2), the benches in the Theresa Creek catchment are likely to have similar origins to those previously reported in southeastern Australia. Benches are largely composed of sand-sized material which also suggests channel origins. Sheetwash and rill erosion do not

98 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion contribute much coarse sediment to the river systems in such environments (Bartley et al., 2006). Grain-size analysis of gully sidewall and headcut sediment from throughout the catchment, showed that about half (mean = 54%, sd = 17%, n = 28) of the sediment was sand or gravel sized (> 63 µm). Gullies have the potential, therefore, to deliver a large volume of coarse sediment to the river system that will be transported as bed (or near-bed) material. Because of the coarse nature of the sediment delivered to the river network and the deep channels that occur throughout the Theresa Creek catchment, much of this sediment is unable to be deposited overbank and as a result, the sediment is deposited as lateral channel deposits.

Flow events (gauging station 130210A, 1972 - 2007; Figure 4.1) that exceed the top of the benches at the LTC site (> 3.5 m above the channel bed or > 73 m3 s-1) have a return period of 0.43 y. This inundation frequency may help explain the rapid accretion rates of the in-channel benches. The sediment burial ages determined for the benches also suggest that they are stable features and, therefore, may be effectively storing large volumes of sediment. Additional factors that are likely to contribute to bench stability are vegetation growth and the absence of grazing animals within the channels.

High rates of channel sediment storage may mean that catchment disturbance related increases in erosion rates have not necessarily translated to commensurately large increases in catchment sediment yields. The gully erosion dominated non-basaltic parts of the catchment may have delivered as much as 1.2 x 106 t of sediment to the stream network since European settlement (unpublished data from McKergow et al. (2005)).

However, assuming an average bench cross-sectional area of 50 m2 and a sediment density of 1.5 t m-3, just 20% of the ~ 400 km of the non-basaltic channels would need

99 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion to have in-channel benches on both sides of the river to store this sediment. At present spatial data on the distribution of in-channel benches within Theresa Creek is not available; however, field observations suggest that they are common. Further field investigation of the distribution of in-channel benches is required to confirm this assessment.

4.6. Conclusions

This study used single-grain OSL dating and 137Cs depth profiles to determine the pre- and post-disturbance rates of floodplain and in-channel bench accretion in a dry-tropical catchment in northeastern Australia. A 3 - 4 times increase in floodplain accretion rates in the last ~ 110 - 120 years shows that catchment disturbance related to European settlement has significantly increased the sediment flux from Theresa Creek. However, since the mid-twentieth century floodplains downstream of areas where sediment sources are dominated by gully erosion show signs of decreasing accretion rates. This is probably the result of a reduction in the amount of sediment delivered from an increasingly stabilising gully network. In contrast, continued increases in the rate of floodplain sedimentation from basaltic-based areas suggest that intensive cultivation continues to supply relatively large amounts of sediment to the river system.

The 3 - 4 times increase in floodplain sedimentation rates that have occurred since

European settlement are low compared to previously reported increases from mainly temperate areas. This is attributed to naturally high sediment loads experienced in dry- tropical catchments. Also, unlike many of the catchments of temperate southeastern

Australia, the dry-tropical catchments of northeastern Australia are also likely to have always had a high degree of channel connectivity and, therefore, sediment conveyance through the river network.

100 Chapter 4 – Changes in rates of floodplain and in-channel bench accretion

A large amount of post-disturbance generated sediment may be stored within in-channel benches. These features are relatively common and appear to be stable at the decadal scale. Although the in-channel benches are dominated by sand-sized sediment their common occurrence throughout the catchment may mean that they are also storing large volumes of fine sediment.

The findings here are important for improving our understanding of the impact of anthropogenic activities on sediment yield from dry-tropical catchments draining to the

GBR. Previous large-scale modelling studies suggest sediment yield increases of up to an order of magnitude. Naturally high sediment yields together with the opportunity for storage in catchment sediment sinks, such as floodplains and in-channel benches, mean that these previous estimates may be exaggerated. Further work is required to quantify the role of sediment sinks in the downstream, low-lying parts of the Fitzroy River basin and indeed other large dry-tropical catchments that drain to the GBR.

101

CHAPTER 5

Validation of a spatially distributed erosion/sediment yield model (SedNet) with empirically-derived data

Channel erosion in upper Theresa Creek

Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

5.1. Introduction

The use of erosion/sediment yield models has become a common method to assess the impacts of land use changes within catchments and determine appropriate management options (e.g., Wicks and Bathurst, 1996; Van Rompaey et al., 2001; de Vente et al.,

2008; Pandey et al., 2008). Traditionally, many erosion/sediment yield models were developed to make predictions on a spatially lumped basis (Merritt et al., 2003). The increased availability of digital spatial data, utility of geographic information system

(GIS) tools, and the rapid development in computational power over the last couple of decades (Finlayson and Montgomery, 2003; Jetten et al., 2003; Merritt et al., 2003) has led to a proliferation in the development and use of spatially distributed approaches

(e.g., Wicks and Bathurst, 1996; Takken et al., 1999; Prosser et al., 2001a; Van

Rompaey et al., 2001; de Vente et al., 2008).

Spatially distributed erosion/sediment yield models do not necessarily perform any better than spatially lumped models in the prediction of catchment sediment yields (De

Roo, 1998; de Vente et al., 2008). They do, however, have the advantage of being able to account for the heterogeneity of catchment attributes, land use, rainfall and physiographic factors (Merritt et al., 2003; Walling et al., 2003). Furthermore, such models are also able to provide spatially distributed predictions of erosion, sediment transport and storage within catchments (Walling et al., 2003; de Vente et al., 2008), which is often the information that is of most benefit to catchment managers.

Lack of model validation is, however, increasingly recognised as an issue in the application of spatially distributed erosion/sediment yield models (Jetten et al., 2003;

Walling et al., 2003; de Vente et al., 2008). Although there are many examples of field- based data being used to validate and/or calibrate erosion/sediment yield models (e.g.,

104 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

Van Rompaey et al., 2001; Stefano et al., 2005; Hessel et al., 2006; Polyakov et al.,

2007; Sarangi et al., 2007; de Vente et al., 2008), these have generally been performed at the hillslope or small catchment scale. Because of the absence, and difficulty in obtaining, detailed field-based data at the appropriate scale (Quine, 1999; Takken et al.,

1999; Jetten et al., 2003), validation is more problematic for models that are applied at the large-scale catchment or regional scale.

Nonetheless, important catchment management decisions are being made on the basis of model predictions and it is essential that more effort is made to validate their results. A common method of model validation is the comparison of modelled estimates of sediment yield with information obtained from historical suspended sediment sampling at gauging sites (Collins and Walling, 2004). While catchment outlet sediment yield comparisons are useful, it is also important to ascertain if the predicted spatial patterns are accurate (Grayson et al., 1992; Kirchner, 2006), or as highlighted by Kirchner

(2006), if the models are predicting the “right answers for the right reasons”.

In Australia, the spatially distributed sediment erosion/sediment yield model SedNet

(Sediment River Network Model) has recently been used to determine catchment sediment yields and sediment sources for a number of catchments (e.g., DeRose et al.,

2003; Dougall et al., 2005; McKergow et al., 2005; Hateley et al., 2007). Despite its widespread adoption as a catchment management tool, there have been very few attempts to validate the modelled outputs with empirical data (e.g., Rustomji et al.,

2008). Rustomji et al. (2008) found that utilising multiple independent data sets (such as gauging station and sediment tracing data) provided a detailed evaluation of SedNet’s performance and increased the likelihood that the model was correctly representing the

105 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model main components of a sediment budget for the Lake Burragong catchment in southeastern Australia.

The model has been applied mostly at the large catchment or regional-scale so validation of its outputs has been problematic. To date, most model validation has been limited to comparing modelled catchment outlet sediment yields with those calculated from available suspended sediment monitoring (e.g. Fentie et al., 2005; McKergow et al., 2005).

This study applies the SedNet model to Theresa Creek, a 6000 km2 subcatchment of the

Fitzroy River basin in central Queensland, Australia. SedNet’s performance, in terms of its predictions of sediment loads and the contribution of major sediment sources, is evaluated against sediment load data determined from suspended sediment monitoring and radionuclide tracing of sediment sources. An assessment is also made of model predictions of pre-European settlement (pre ca. 1850) sediment yield by comparing changes in measured rates of pre- and post-disturbance floodplain accretion with those predicted by SedNet. The findings may be used to improve model predictions made at the large catchment- to regional-scale particularly for the dry-tropical catchments of northeastern Australia. Suggestions for future model development are also made.

5.2. SedNet background

SedNet was originally developed for use at the regional scale (> 104 km2) during the

Australian National Land and Water Resources Audit (NLWRA; NLWRA, 2001;

Prosser et al., 2001a). It was designed as a broad-scale model to provide relative approximations of catchment sediment yield and to identify general spatial patterns of erosion. Since the completion of the NLWRA, SedNet has been further developed and applied at different scales throughout Australia (e.g., DeRose et al., 2003; Wilkinson et

106 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model al., 2006; Kinsey-Henderson et al., 2007). In particular, the catchments of northeastern

Australia that drain to the Great Barrier Reef (GBR) have been the focus of a number model applications (e.g., Bartley et al., 2003; Kinsey-Henderson et al., 2005;

McKergow et al., 2005; Dougall et al., 2007). Previous SedNet modelling predicted that sediment yields from GBR catchments have increased by up to ten times since

European settlement (NLWRA, 2001; Bartley et al., 2003; McKergow et al., 2005).

SedNet has been identified as an important tool for setting specific water quality targets by the Queensland Department of Natural Resources and Water (QDNRW), the state government department responsible for management of the catchments draining to the

GBR. Some work has taken place to improve model inputs, such as spatial data and catchment specific parameters (e.g., Dougall et al., 2005; Cogle et al., 2006; Dougall et al., 2006; Bartley et al., 2008; Trevithick et al., 2008).

SedNet has two key advantages over the previous sediment yield modelling approaches used within Australian catchments: (i) a complete sediment budget is produced where the contributions of all the major sediment inputs are considered and the significance of catchment storages (e.g., floodplains and reservoirs) are quantified, and (ii) because the model operates within a GIS framework, spatial information of the relative importance of different sediment sources and their contribution to the catchment outlet can be acquired. An additional benefit of SedNet is that pre-disturbance sediment budgets can be generated by scaling input parameters to represent conditions in the catchment prior to major landscape disturbance (commonly referred to as pre-disturbance catchment conditions). This allows users to assess relative changes in catchment conditions over a range of timescales.

107 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

5.3. Study area

Theresa Creek is located in the western Fitzroy River basin (Figure 5.1) in central

Queensland. The main population centre in the catchment is Clermont (population ~

2000) which lies 300 km west of Rockhampton (population ~ 65 000), the largest centre in the Fitzroy River basin. The climate is dry-tropical (Köppen climate classification:

BSh). The mean annual rainfall at Clermont is 649 mm (sd = 240 mm) with most rainfall occurring between November and March. Mean annual potential evaporation at

Clermont is ~ 2080 mm (BoM, 2007). Most large runoff events are a result of tropical low pressure systems that periodically make landfall. For a more detailed physiographic description of the study catchment see Chapter 2.

The first Europeans explored the hinterland of Queensland (including the Fitzroy River basin) in the mid-1840s. Settlement of the region began soon after initial exploration with sheep grazing being the dominant land use established by 1855 (Lewis et al.,

2007). The modelling in this study is carried out for two periods which are referred to as pre-disturbance and post-disturbance. For the purpose of this study the pre- disturbance period encompasses the late Holocene up until the settlement of the catchment by Europeans (ca. 1850). The post-disturbance period is from ca. 1850 to the present day. It is recognised that the Australian landscape has been subject to human disturbance over a much longer time frame, particularly through the fire management practices of the Aborigines (Russell-Smith et al., 1997). However, major catchment disturbance that significantly affected erosion rates and sediment yield did not occur until the introduction of European-style land uses (Prosser, 1990; Dodson and

Mooney, 2002).

108 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

Figure 5.1. Theresa Creek catchment, showing the location of river bed sediment sampling sites and gauging stations referred to in the text.

109 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

5.4. Methods

5.4.1. SedNet model

Only a brief description of the SedNet model is present here; a detailed model description and discussion of its conceptual basis are available in Prosser et al. (2001a) and Prosser et al. (2001b). SedNet constructs a drainage network from a digital elevation model (DEM). The basic unit of calculation is a river link, which is defined as a section of river between tributaries (Figure 5.2). Each link has an internal catchment area that represents the land that drains directly into that link. The catchment area of each link varies but the user sets a threshold area for first order links. A smaller threshold area results in more first order links and therefore smaller link internal catchment areas. Typical catchment areas range from 10 - 100 km2 depending on the size of the catchment being investigated.

Figure 5.2. A schematic river network showing links, Shreve magnitude of each link and the internal catchment area (shaded) of a magnitude one and magnitude four link (from Prosser et al., 2001a).

110 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

To reflect the separate transport processes involved, SedNet has separate modules for suspended load and bed load. The model routes all “fine” sediment into the suspended sediment budget and all “coarse” sediment into the bed load budget. Although the division of fine and coarse sediment can be defined by the user, fine sediment is usually assumed to be < 63 µm material (silt and clay) and coarse sediment is material > 63 µm.

Only the suspended sediment component is considered in this study.

For each river link (i, Figure 5.3) a mean annual mass balance (t y-1) is calculated:

SBGHTY − +++= iiiiii (5.4)

where Yi is the mean annual suspended sediment yield, Ti is the input of sediment from the upstream tributary; Hi is hillslope erosion; Gi is gully erosion, Bi is riverbank erosion and Si is net storage as deposition on the floodplain or lake/reservoir.

Figure 5.3. Conceptual diagram of sediment budget for a river link and its associated subcatchment (light shade). Solid black arrows represent link budget input terms while open arrows represent link budget loss terms. For the illustrated river link storage loss is in the form of floodplain (grey shade) deposition; however, this can also be reservoir deposition where a link ends in a reservoir (from McKergow et al., 2005).

111 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

Unlike many similar models, SedNet considers all major potential catchment sediment inputs (de Vente et al., 2008). In most Australian catchments these are sheetwash and rill erosion (hereafter referred to as hillslope erosion), gully erosion, and river bank erosion. The model requires spatial data of hillslope erosion and gully erosion in the landscape. River bank erosion rates are calculated by the model using an empirically- derived algorithm.

5.4.1.1. Hillslope erosion

Mean annual hillslope erosion is predicted using the Revised Universal Soil Loss

Equation (RUSLE) (Renard et al., 1997). The RUSLE calculates mean annual soil loss

(A, t ha-1 y-1) as a product of six factors: rainfall erosivity (R), soil erodibility (K), hillslope length (L), hillslope gradient (S), vegetation cover and land use (C), and supporting practice (P):

A = R K L S C P (5.5)

The proportion of hillslope erosion that reaches a river link is controlled by a hillslope delivery ratio (HSDR). Most previous model applications used HSDR values between 5

- 10% (see DeRose et al., 2003; McKergow et al., 2005; Dougall et al., 2006), as this range produced results consistent with measured loads and the general understanding that channel sources dominate in many Australian catchments (Prosser et al., 2001b).

5.4.1.2. Gully erosion

The sediment contribution from gullies to the suspended sediment load of a link (Gi) (t y-1) is determined by:

112 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

ρ αPDA fssii Gi = τ (5.6)

2 Where, Ai is the link catchment area (km ), D i is the gully density of the link

-2 -3 subcatchment (km km ), ρ s is the bulk density of eroded sediment (t m ), α is the

average cross-sectional area of a gully, Pfs is the proportion of sediment that contributes to the suspended sediment load, and τ is the gully age (y). Table 5.1 indicates the range of gully parameter values used by previous SedNet model applications in Australia. Also indicated in Table 5.1 are the default gully parameter settings used by most large-scale applications of the model (e.g., NLWRA, 2001;

McKergow et al., 2005).

Table 5.1. Range of gully parameter values used by previous applications of SedNet. See text for description of terms.

Gully SedNet default Range of Reference parameter values previously used values 1.5 t m-3 1.5 t m-3 All known published SedNet ρ s applications 2 2 α 10 m 10 - 23 m Rustomji et al. (2008) P 50 % 47 – 58 % Rustomji et al. (2008) fs τ 100 y 100 - 150 y DeRose et al. (2003); Rustomji et al. (2008)

A gully production factor can also be applied that scales the long-term gully sediment yield to account for temporal changes in gully network extension rate (Wilkinson et al.,

2004).

5.4.1.3. River bank erosion

River bank erosion is estimated from a conceptual relationship based on limited empirical data, that shows that potential bank erosion rate is proportional to stream

113 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model power (Rutherfurd, 2000). The actual rate of river bank erosion (BR; m y-1) for a river link is reduced by the presence of riparian vegetation and the proportion of non-alluvial material in the banks, mapped from floodplain width (Hughes and Prosser, 2003):

W BR = 00002.0 ρ −− ePRSgQ − 08.0 i )1)(1( i MAw i i i (5.7)

Where ρ w is the density of water, g is acceleration due to gravity. For each river link i,

-1 S is the channel slope, QMA is the mean annual flow (ML y ),R Pis the proportion of intact riparian vegetation, W is mean floodplain width (m).

For each river link, the amount of fine sediment eroded from the banks (B; t y-1) is determined by:

= ρ PLHBRB fsiiisi (5.8)

-3 where ρ s is the bulk density of the eroded sediment (t m ), H is the bank height (m)

and L is the length (m) of a river link i. Pfs is the proportion of eroded sediment that

contributes to the suspended sediment load. Based on limited available data, Pfs is assigned a value of 0.5 (Prosser et al., 2001a).

5.4.1.4. Floodplain area and width

Floodplain area for each link is calculated by overlaying each link’s internal catchment area with a spatial layer of floodplain extent. Mean floodplain width for each river link

(Wi) is determined by:

Af W = i i L i (5.9)

114 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

2 Where Afi is the floodplain area within a link’s internal catchment (m ), and Li is the length of a river link (m).

5.4.1.5. Suspended sediment load

A fundamental assumption behind SedNet is that rivers have the capacity to transport all fine sediment delivered to them, i.e., they are sediment supply limited. Therefore, the amount and pattern of mean annual suspended load is determined by the supply of erosion sources to rivers. This is consistent with previous measurements of river sediment loads and concentrations (e.g., Williams, 1989; Asselman, 1999; Hovius et al.,

2000).

All hillslope erosion, and a user defined proportion of gully and riverbank erosion, form the inputs to SedNet’s suspended sediment budget. SedNet’s default setting is to assign

50% of the sediment from gully and river bank erosion to the suspended sediment budget and the balance to the bedload budget. This partition is based upon observations from detailed field-based sediment budgets (Prosser et al., 2001a).

Sediment can be lost to further downstream transport by floodplain or reservoir deposition. Floodplain deposition for a link (F; t y-1) is predicted from settling theory and the residence time of water on floodplains using inputs of floodplain width and regionalised hydrological parameters (Prosser et al., 2001a):

 vAf   − i   Qf  Qf  F i  eI  i   i i 1−=  Qi   (5.10) where, for each link i, Q is total discharge (m3 s-1); Qf is floodplain discharge (m3 s-1); I is the incoming sediment (t y-1); v is the settling velocity of suspended sediment (m s-1);

115 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model and Af is the floodplain area (m2). The default settling velocities of sediment are based on silt-sized sediment. SedNet uses a modification of the Brune Rule (Brune, 1953) to deposit sediment in reservoirs (see Prosser et al., 2001a).

5.4.1.6. Hydrology

The SedNet suspended sediment budget module requires a number of hydrological parameters for each river link, these are mean annual flow (QMA), bankfull flow (QBF), and median flood flow (QMO). Mean annual flow is required to determine bank erosion, reservoir sediment trapping efficiency and is the basis for the prediction of bankfull and median overbank flow. Bankfull flow and median overbank flow are required to separate in-channel flow from the flood flow which enables the proportion of flood water (and hence suspended sediment) inundating the floodplain to be determined.

5.4.2. Model input data

The data sources for the various model inputs are summarised in Table 5.2. For the

RUSLE model of hillslope erosion, ground cover was derived from land use mapping of the Fitzroy River basin (Calvert et al., 2000) and C factors were assigned to land uses from data obtained from Brodie et al. (2003) and Dougall et al. (2006). No spatial data was available regarding land management practices (contour banks etc.) so the P factor was assigned a value of 1 everywhere. Hillslope erosion rates for pre-disturbance conditions were also estimated using the RUSLE. Natural vegetation cover was obtained from the Australian Natural Resources Data Library (ANRDL; BRS, 2008) and C factors were assigned in accordance with the measured erosion rates reviewed by

Lu et al. (2003b).

A floodplain layer was generated from a 25 m resolution DEM of the catchment using the multi-resolution index of valley bottom flatness (MRVBF) algorithm (Gallant and

116 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

Dowling, 2003). The MRVBF algorithm identifies valley floors on the basis of their topographic signature as flat low-lying areas (Gallant and Dowling, 2003). The algorithm does not differentiate floodplains from other flat areas, such as adjacent flat colluvial areas. However, it was found that areas with a MRVBF index value ≥ 2.5 matched well with floodplain extent shown on aerial photographs and satellite images.

Table 5.2. Sources of spatial and hydrological data used as inputs to the SedNet model for Theresa Creek.

Model input data Source Digital elevation model 25 m resolution DEM for the Fitzroy River basin (Queensland Department of Natural Resources and Water) RUSLE R factor Australian Natural Resources Data Library (BRS, 2008) K factor Australian Natural Resources Data Library (BRS, 2008) L factor Australian Natural Resources Data Library (BRS, 2008) S factor Australian Natural Resources Data Library (BRS, 2008) Land use Calvert et al. (2000) Natural vegetation Australian Natural Resources Data Library (BRS, 2008) C factor Brodie et al. (2003); Lu et al. (2003b); Dougall et al. (2003) Gully density data Gully density model for Fitzroy River basin (Trevithick et al., 2008) Bank erosion Riparian vegetation Australian Natural Resources Data Library (BRS, 2008) Floodplain extent MRVBF (Gallant and Dowling, 2003) model based on 25 m DEM Hydrological data Queensland Department of Natural Resources and Water

The hydrological parameters were calculated using daily flow data from catchment gauging stations (following Brodie et al. (2003)). Flow data for ungauged river links was extrapolated from multiple regression relationships with rainfall and catchment area. Only two gauging stations in Theresa Creek have adequate data (record length ≥

20 years; Wilkinson et al., 2004) for the hydrological regionalisation. Accordingly, the data from 13 additional gauging stations from the western Fitzroy River basin, with

117 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model similar rainfall ranges to Theresa Creek, were also included in the regionalisation calculations.

For each of the 15 selected gauging stations, a mean annual runoff coefficient (ROC) was calculated from the measured mean annual flow, catchment area and catchment mean annual rainfall. The ROC of a river link was determined as a function of the ratio of the potential evaporation (PET, mm) to the mean annual rainfall (RF, mm) for the catchment area upstream of each link:

bRFPETa +bRFPETa ))/(/(1   PET  + ))/((  PET ROC 1+=    −   RF   RF   (5.11) where, a and b are empirical values fitted by regression. Mean annual rainfall and potential evaporation data was obtained from the ANRDL (BRS, 2008)

Mean annual flow (Ml y-1) for a link is calculated from catchment area (A, km2) and mean annual rainfall:

ARFROCQ MA ××= (5.12)

Bankfull flow is determined by:

QcQ d BF = ( MA ) (5.13) where c and d are empirical values fitted by regression.

Median overbank flow is determined by:

118 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

(QeQ ) f MO = MA (5.14) where e, and f are empirical values fitted by regression.

The coefficient of efficiency, E, was used as a measure of hydrological parameter performance. The coefficient of efficiency represents a form of noise to signal ratio, comparing the average variability of model residuals to the variability of the target output (Schaefli and Gupta, 2007). Coefficient of efficiency values of between 0.0 and

1.0 are considered acceptable with higher positive values indicating superior model performance (Chiew and McMahon, 1993; Schaefli and Gupta, 2007). A negative E value indicates a poorly performing model (Schaefli and Gupta, 2007). The coefficient of efficiency is defined by:

n OBSEST 2 ∑( i − ) E i=1 = 1− n OBSOBS 2 ∑( i − ) i=1 (5.15)

where OBSi and ESTi are the observed and estimated parameter, respectively, and

OBS is the mean value of all the observations. Table 5.3 indicates the parameters fitted by regression analysis and the associated measures of E.

Table 5.3. Regionalised hydrological parameters calculated for use in SedNet.

Parameter Model R2 E ROC a = -1.0639; b = 5.9540 0.58 n/a 0.7027 QBF Q(2 years) BF2 = 2.7323QMA 0.87 0.71 0.7606 QBF Q(6 years) BF6 = 5.17QMA 0.91 0.67 0.7438 QMO QMO = 1.2131QMA 0.84 0.68

119 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

5.4.3. Model validation data

5.4.3.1. Estimation of suspended sediment loads

River gauging and suspended sediment sample data was obtained from the Queensland

Department of Natural Resources and Water. Only two gauging stations within the study catchment had both reasonably long records of flow and suspended sediment sampling; 130210A (Theresa Creek@Valeria) and 130207A (Sandy Creek@Clermont)

(Figure 5.1). Gauging stations 130210A and 130207A are hereafter referred to as

Valeria and Clermont, respectively. All suspended sediment samples were grab samples collected at irregular time intervals since the early 1970s (Figure 5.4). The data is of reasonably low quality for load predictions and the estimates must be considered as approximations only.

Suspended sediment loads were calculated using two methods. Method 1 is a simple averaging estimator (Preston et al., 1989):

n  C j  = QKLoad MA ∑ n   j  (5.16) where K is a conversion factor to account for differences in units and periods of time,

-1 -1 QMA is the mean annual flow (Ml y ), C j is the concentration (mg l ) for suspended sediment sample j, and n is the number of samples.

120 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

(A)

) 10000 -1

1000

100

10

y = 88.71x0.4394 R2 = 0.4395 Suspended sediment concentration (mg l concentration (mg sediment Suspended 1 0.001 0.01 0.1 1 10 100 1000 Discharge (m3 s-1) (B) 10000 ) -1

1000

100

10

y = 133.23x0.2455 R2 = 0.2756 Suspended sediment concentration (mg l concentration sediment Suspended 1 0.001 0.01 0.1 1 10 100 1000 Discharge (m3 s-1)

Figure 5.4. Suspended sediment rating curves for gauging stations (A) 130207A (Sandy Creek@Clermont) and (B) 130210A (Theresa Creek@Valeria). The data (47 samples) for 130207A were collected between 1974 and 2001. The data (38 samples) for 130210A were collected between 1973 and 2006.

To assess the representativeness of the suspended sediment samples, flow duration plots were constructed for the Valeria and Clermont gauging stations (Figure 5.5). The data showed that the range of discharges at which the in-stream sediment samples were collected was exceeded by ~ 20% and ~ 50% of all river flows at Valeria and Clermont,

121 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model respectively. Because larger flow events are underrepresented, the calculated loads for method 1 should be considered as minimum values.

Method 2 is a relationship derived between discharge and suspended sediment concentration at the time of sampling by a log-transformed least-squares regression:

aQC b = (5.17) where C is the sampled instantaneous suspended sediment concentration, Q is the instantaneous discharge associated with C, while a and b are the constant and exponent of the rating relationship (Figure 5.4). Retransformation bias was corrected using the non-parametric smearing estimate of Duan (1983). The coefficient of efficiency was used as a regression model performance measure (Table 5.4). Loads were then predicted at daily intervals.

Despite the limitations of the sediment load estimates (Table 5.4), they provide the only independent method available to validate the modelled sediment loads using field-based data.

122 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

10000

Clermont (130207A) Valeria (130210A) 1000 ) -1 s

3 100

10 Discharge (m

1

0.1 0 20 40 60 80 100 % equalled or exceeded

Figure 5.5. Flow duration plots for the Sandy Creek@Clermont (130207A; 1965 - 2008) and Theresa Creek@Valeria (130210A; 1971 – 2008) gauging stations. The horizontal dashed lines indicate the maximum flow for the QDNRW suspended sediment sampling datasets (47 m3 s-1 for 130207A and 517 m3 s-1 for 130210A).

Table 5.4. Mean suspended sediment load estimates for gauging stations 130207A (Sandy Creek@Clermont) and 130210A (Theresa Creek@Valeria).

Method Site Mean annual load Performance (kt y-1) Method 1 130207A 4 n/a 130210A 63 n/a

Method 2 130207A 16 E = 0.33, R2 = 0.45 130210A 151 E = 0.05, R2 = 0.28

5.4.3.2. Sediment source tracing

The contribution of major sediment sources was determined by measuring the properties of fallout radionuclides in sediments. The method entails characterising the major sediment sources (e.g., cultivated land, uncultivated land and channel sources) on the

123 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

137 210 basis of selected fallout radionuclide concentrations ( Cs and Pbex). These concentrations are then compared with the radionuclide concentrations of samples from downstream sediment deposits using a numerical mixing model to determine the relative contribution of each of the sources (e.g., Walling, 2005). This method is dependent on the documented observation that the concentration of fallout radionuclides varies significantly with soil depth (e.g., Wallbrink and Murray, 1993; He and Owens,

1995). For a more detailed description of the radionuclide tracing technique see He and

Owens (1995) and Wallbrink et al. (1998).

Source samples were obtained from throughout the catchment from each of the three major sediment sources. Because of the lack of river flow during the study period it was not possible obtain suspended sediment samples for analysis. River bed samples were, therefore, obtained from five locations (SC1, SC2, SC3, TC1, and TC2; Figure

5.1) between August and November 2006. To minimise the effect that fluvial sorting can have on the radionuclide signal the < 10 µm fraction of both source sites and channel-bed deposits was analysed for radionuclide concentrations. Radionuclide analysis was performed using high resolution gamma spectrometry, as described by

Murray et al. (1987). For a detailed description of the methods used to determine the relative contributions of each of the sediment sources to the river bed sites see Chapter 2

(Section 2.3.1.1).

Figure 5.6A illustrates the relative proportions for the major erosion sources determined by the radionuclide tracing. The Sandy Creek sites (SC1 - SC3) show little variation, with almost even contributions of channel and cultivation sources and very low input from uncultivated sources. The high contribution of cultivated sources at these sites is attributed to the relatively high proportion of the upslope contributing area in cultivation

124 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

(Figure 5.6B). In contrast, channel sources account for over 80% of the sediment at site

TC1 and this is attributed to the area upstream of TC1 being an area of significant gully erosion (Figure 5.6C).

Figure 5.6. (A) Sediment source contributions to each river bed sampling site as determined by radionuclide tracing, (B) land use (source: Calvert et al., 2000), and (C) gully density (source: Trevithick et al., 2008) for Theresa Creek.

The sediment at site TC2 is predicted to be dominated by cultivation sources, although the channel-derived sources are still important. The relative contributions predicted for

TC2 may, however, be anomalous as both upstream sites (TC1 and SC3) are dominated by channel inputs. The contribution of mainly cultivation-derived sediment from the unsampled Capella Creek (Figure 5.6B) may be partly responsible for this. However, even if it is assumed that Capella Creek contributed 100% cultivation-derived sediment,

125 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model and this is mixed with the contributions of TC1 and SC3 on an area-weighted basis, it would still not account for the dominance of cultivation sources at site TC2.

5.4.3.3. Floodplain sedimentation rates

Floodplain sediment cores were extracted from three catchment locations: upper

Theresa Creek (UTC-F), Lower Theresa Creek (LTC-F) and Capella Creek (CC-F)

(Figure 5.1). Floodplain sedimentation rates were determined using a combination of

137Cs depth profiles (see Chapter 3) and single-grain optically stimulated luminescence

(OSL; see Chapter 4) dating (Table 5.5).

Table 5.5. Floodplain accretion rates and increases since catchment disturbance commenced (ca. 1850), as determined by OSL dating and 137Cs depth profiles for the three floodplain sampling sites.

Sample ID Depth Sediment pre-/post- Average Increase in (cm) age (y) catchment accretion rate accretion disturbance (mm y-1) rate UTC-F 137Cs profile - - post- 1.4 ± 0.2 ~ 2 × UTC-F-15 15 110 ± 15 post- UTC-F-60 60 1010 ± 90 pre- 0.6 ± 0.1 UTC-F-100 100 1640 ± 150 pre- LTC-F 137Cs profile - - post- 3.0 ± 0.4 ~ 3 × LTC-F-22 22 65 ± 5 post- LTC-F-35 35 120 ± 20 post- LTC-F-50 50 255 ± 25 pre- 0.9 ± 0.1 LTC-F-95 95 790 ± 70 pre- CC-F 137Cs profile - - post- 1.7 ± 0.3 ~ 4 × CC-F-20 20 120 ± 20 post- CC-F-40 40 540 ± 145 pre- 0.4 ± 0.1 CC-F-60 60 1050 ± 140 pre-

126 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

5.5. Model application

5.5.1. Post-disturbance models

To evaluate the ability of SedNet to produce reasonable predictions of post-disturbance sediment yield and patterns of erosion, a number of model parameterisations were investigated. Model parameters were modified principally on the basis of observational data obtained for Theresa Creek. To assess the outcome of varying model input parameters, a baseline model parameterisation was also applied using the settings of previous GBR catchment applications of the model (e.g. NLWRA, 2001; McKergow et al., 2005). Table 5.6 illustrates the parameterisations tested here, with model A representing the “default” GBR catchment parameters used in the previous applications.

Table 5.6. Summary of the post-disturbance SedNet model parameterisations.

Model Input parameter A B C D HSDR (%) 10 10 10 5

QBF recurrence interval (y) 2 6 6 6 River bank height (m) 3 3 variable variable Gully production factor 1 1 0.5 0.5 Gully age (y) 100 100 130 130 Mean gully cross-section area (m2) 10 10 5 5 Gully fine sediment contribution (%) 50 50 46 46

Model B only modified the recurrence interval of bankfull events. Based on the analysis of cross-sectional and discharge data from gauging stations throughout the

Fitzroy River basin (Dougall et al., 2006) the recurrence interval of bankfull events was modified to 6 y.

Model C modified the channel (gully and river bank) erosion input parameters only. A uniform catchment-wide bank height is normally used to calculate river bank erosion

127 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

(Equation 5) in the absence of any other information. However, in Theresa Creek this is likely to result in an underestimation of the volume of sediment contributed from bank erosion as bank heights can exceed 10 m on the main tributaries. A bank height relationship determined by Dougall et al. (2006) for the Fitzroy River was, therefore, used to calculate river bank height:

2805.0 H = a777.0 C (5.18) where Ca is the upslope contributing area.

The other parameters modified in model C relate directly to gully erosion inputs. A gully production factor of 0.5 was applied because recent reductions in floodplain sedimentation rates, from areas dominated by gully erosion in Theresa Creek, suggest sediment yield from gullies has recently declined (see Chapter 2). The reported findings in Chapter 2 also suggest that a significant phase of gully incision associated with land use disturbance began ca. 1880. Therefore, the period over which the post-disturbance model was applied was increased from 100 to 130 years. Grain-size analysis of gully sidewall and head cut sediment from throughout Theresa Creek determined that 46 %

(sd = 17%, n = 28) of the sediment was silt or clay sized (< 63 µm). Although not considerably different from the default setting of 50% this minor change may still be important, because of the significance of gully erosion in the catchment.

Gullies are widespread throughout Theresa Creek (Hughes et al., 2001; Trevithick et al.,

2008), however, by Australian standards the gullies tend to be relatively small.

Previous gully measurements from southeastern Australia suggest average cross- sectional areas of 10 - 23 m2 are appropriate (Prosser and Winchester, 1996; Rustomji,

128 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

2006). Field observations within Theresa Creek, however, suggest that an average cross-sectional area of 5 m2 is more appropriate.

Model D uses the same parameter settings as model C except that the HSDR is reduced to 5%. Previous applications of SedNet have used an HSDR of 5% in the catchments of southeastern Australia and 10% in the GBR catchments (e.g., DeRose et al., 2003;

McKergow et al., 2005; Rustomji et al., 2008). Measured data on actual HSDR rates is sparse and the values of 5% and 10% were based on qualitative analysis of hillslope plot loads compared to small catchment loads (pers comm. I. Prosser, CSIRO Land and

Water, October 2008). Hillslope sediment delivery ratios of between 5% and 10% are low compared to those reported from other countries (Edwards, 1993), however, because of the largely flat terrain, long hillslope lengths and high clay content of many soils these low values are considered appropriate for many Australian catchments (Lu et al., 2003a). A value of 10% was previously applied in the GBR catchments because of the large number of small, steep, wet coastal catchments. However, while a higher

HSDR may be appropriate for these coastal catchments there is little justification for applying the 10% rate to large dry-tropical catchments, such as the Fitzroy River basin.

In fact, the physiography of the low-lying dry-tropical catchments of northeastern

Australia is more comparable to the inland catchments of southeastern Australia, which suggests that an HSDR of 5% may be more appropriate (Lu et al., 2003a).

5.5.2. Pre-disturbance models

SedNet was also use to estimate sediment yield for pre-disturbance conditions. Some previous applications of SedNet in GBR catchments (e.g., NLWRA, 2001; McKergow et al., 2005) assume pre-disturbance conditions to be “pristine” with high levels of vegetative cover providing protection from both hillslope and channel erosion. The

129 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model resulting pre-disturbance sediment yields are, therefore, typically very low as a result.

While there may be some environments where sediment yields were very low under pre-disturbance conditions, it is unlikely to be the case in the dry-tropics where highly variable rainfall results in frequent drought conditions. In such conditions groundcover is significantly reduced and consequently the potential for erosion is high, particularly during drought-breaking rainfall events (McCulloch et al., 2003).

Directly assessing the reliability of estimates of pre-disturbance sediment yield is not possible; however, other modelled outputs can be examined to provide some assessment of model performance. Comparing the floodplain accretion rates predicted by the pre- and post-disturbance models to those determined from empirically-derived data provides such an evaluation method. SedNet determines an average sedimentation rate

(mm y-1) for the entire floodplain width for each river link; therefore, it is not appropriate to directly compare the modelled rates with the rates obtained from field- based measurements at specific points on the floodplain. Nonetheless, the magnitude of the increases in accretion rates between the pre- and post-disturbance periods can be compared to the magnitudes derived from field measurements to provide a measure of model performance.

For the pre-disturbance models a number of parameters were modified within reasonable limits and the impact on the model’s ability to accurately represent the correct magnitude increase in floodplain accretion rates was examined. This analysis assumes that if the model is able to accurately predict the magnitude of increase in floodplain accretions between the pre- and post-disturbance periods then the magnitude of increases in sediment yield should also be reliable.

130 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

Table 5.7 illustrates the pre-disturbance parameterisations tested here with model P1 representing the settings used in previous GBR applications (e.g., NLWRA, 2001;

McKergow et al., 2005). Model P1 sets gully erosion input to zero and river bank erosion is minimised by assigning a 95% cover of riparian vegetation to all river links.

Rates of natural hillslope erosion were determined from the pre-disturbance RUSLE calculations of Lu et al. (2003b).

In an attempt to more accurately represent pre-disturbance conditions, both the input of sediment from gully erosion and river bank erosion were increased in model P2. The assumption in previous SedNet modelling that gully erosion is entirely a response to anthropogenic activities is probably not justified given that naturally occurring catchment disturbance is likely to have resulted in localised gully incision, as has been noted in other parts of Australia (Prosser et al., 1994; Hancock and Evans, 2006). No known empirical data exists on the extent of pre-European gully extent; therefore, a figure of 10% of the post-disturbance rate was applied to model P2 as a best guess estimate. Similarly, sediment input from river bank erosion was also increased by reducing the riparian vegetation cover on all river links to 80%.

Table 5.7. Summary of pre-disturbance SedNet model parameterisations.

Model Parameter P1 P2 P3 Hillslope erosion Natural RUSLE Natural RUSLE RUSLE C factor (Lu et al., (Lu et al., based on 70% 2003b) 2003b) cover (Dougall et al., 2006) Riparian vegetation (%) 95 80 80 Gully erosion (% of post- 0 10 10 disturbance rate)

Model P3 retains the gully and riparian vegetation settings of model P2 and modifies the input from hillslope erosion by assigning C factor values based on an average 70%

131 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model groundcover as determined by Dougall et al. (2006). This is believed to better represent the pre-disturbance conditions given the variable nature of the regional climate that results in periods of significantly reduced groundcover.

5.6. Results and discussion

5.6.1. Evaluation of modelled sediment loads

The suspended sediment sample-based load estimates for each gauging station together with the loads predicted by the four model parameterisations are illustrated in Figure

5.7. The suspended sediment sample-based estimates resulted in loads of between 63 and 151 kt y-1 for Valeria and 4 and 16 kt y-1 for Clermont. These loads are considerably less than those determined by the GBR default model parameterisation, model A, with 246 kt y-1 and 35 kt y-1 predicted for Valeria and Clermont, respectively.

The introduction of a 6 y recurrence interval for bankfull events (model B) increased sediment loads from model A by ca. 25 - 35% to 386 and 42 kt y-1 for Valeria and

Clermont, respectively. This increase can be attributed to more of the river’s flow remaining within the channel and therefore reducing the opportunity for floodplain deposition. Although this modification results in an even greater disparity between the results of the two methods, the increase in overbank full recurrence interval is justified as the deep channels found throughout the catchment, together with the highly variable flow regime, result in relatively infrequent over bankfull events.

Of the parameterisations tested, model C (modified channel erosion input parameters) had the greatest impact on sediment loads, reducing sediment loads to 125 kt y-1 for

Valeria and 11 kt y-1 for Clermont. These loads are much closer to those determined from the in-stream sediment sampling data, particularly for Valeria (Figure 5.7). In addition to the introduction of a variable bank height, the changes introduced by model

132 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

C had the effect of reducing sediment input from gully erosion. The large reduction in loads can be attributed to the importance of channel sources (gully erosion in particular) as a source of sediment within Theresa Creek (see Chapter 2). This illustrates the importance of using catchment-specific information to modify SedNet’s gully input parameters. Because of the observed dominance of gully erosion in many Australian catchments (e.g., Olley et al., 1993; Wallbrink et al., 1998; Wasson et al., 1998), careful assessment should be made of the gully parameters in future model applications.

450 Theresa Creek@Valeria (130210A) 400 386 Sandy Creek@Clermont (130207A) 350 ) -1 300 246 250

200 151 150 125

Sediment yield (kt y 98 100 63 42 50 35 4 16 11 9 0 Load estimate - Load estimate - ABCD method 1 method 2 Model

Figure 5.7. Post-disturbance sediment yields for gauging stations 130210A (Theresa Creek@Valeria) and 130207A (Sandy Creek@Clermont) calculated from historical suspended sediment sampling records and predicted by parameterisations (A - D) of the SedNet model.

Model D, which reduced the HSDR to 5%, only resulted in a small decrease in loads, with estimates of 98 kt y-1 and 9 kt y-1 for Valeria and Clermont, respectively. This indicates that, at least in this environment, the model is not very sensitive to changes in

HSDR.

133 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

5.6.2. Evaluation of modelled estimates of sediment source contributions

Because of the uncertainty associated with both the model outputs and sediment tracing results (see Chapter 2), it is not considered appropriate to compare estimated source contribution values for each method. A more valid approach is to assess the broad scale patterns and whether the different sources are being ranked correctly. It must also be noted that the SedNet model determines average conditions over the period of investigation (last 100 - 130 y for Theresa Creek). The sediment tracing results are based on samples obtained from river beds over a short time period and therefore may not necessarily represent the long term conditions. As identified by Rustomji et al.

(2008) the collection of more samples over a range of hydrologic conditions would enable this issue to be assessed. In the case of this study, no significant flow events occurred during the study period and such a detailed analysis was not possible.

The sediment source contributions estimated by both the radionuclide tracing and each model parameterisation are illustrated in Figure 5.8A-C. Although five channel sites were sampled and analysed for sediment source contributions only three (TC1, TC2 and

SC1) are illustrated here as the tracing and modelling results for the three Sandy Creek sites (SC1, SC2 and SC3) were indistinguishable (Figure 5.6). The similarity of the results from the Sandy Creek sites may be related to the absence of any major tributaries joining Sandy Creek between sites SC1 and SC3 (Figure 5.1).

Comparison of the radionuclide tracing data with the model predictions shows that

SedNet’s performance in predicting the relative contribution of the major sources of sediment is variable. Models A and B performed well at site TC1, by correctly predicting the high contribution of channel sources and no contribution from cultivated sources. However, they overestimate the contribution of channel sources at all the other

134 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model sites. Also note that because no erosion input parameters were modified by model B the relative contribution of each of the sediment sources is identical to model A for all channel sampling sites (Figure 5.8).

Overall, models C and D produce sediment source contribution estimates most similar to those of the radionuclide tracing. In particular model D performs well at site TC1

(Figure 5.8B) and the Sandy Creek sites (Figure 5.8A). The performance of model D at

TC2 is poorer (Figure 5.8C); however, as outlined earlier, the radionuclide tracing results for site TC2 may be anomalous. The inconsistent results for TC2 may be related to the occurrence of discharge events from the cultivation-dominated eastern side of the catchment resulting in little or no input from other sources. Therefore, the model may in fact be producing a reasonable representation of the long-term average conditions while the tracing results may be more representative of a recent localised flow event.

All models fail to predict the large contribution of cultivated sources and small contribution from uncultivated sources. Despite this, models C and D do predict reasonable contributions of hillslope erosion in general (i.e., both cultivated and uncultivated sources). This suggests that there may be an imbalance in the rates of erosion predicted by the RUSLE input data. Erosion rates from cultivated land appear to be underestimated while they are overestimated from uncultivated land. The RUSLE input data for this study was based primarily on research carried out at the continent- scale (Lu et al., 2003b). Further work to improve this input data for catchments of particular interest is warranted.

135 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

(A) 100

80

60

40

% contribution 20

0 Tracer A B C D estimate (B) 100

80

60

40

% contribution 20

0 Tracer A B C D estimate (C) 100

80

60

40

% contribution 20

0 Tracer A B C D estimate Model

Channel Cultivated Uncultivated

Figure 5.8. Relative contribution of channel, cultivation and non-cultivations sources to channel sampling sites (A) SC1 (= SC2 and SC3), (B) TC1, and (C) TC2 as estimated by radionuclide tracing and four parameterisations (A - D) of the SedNet model.

136 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

5.6.3. Floodplain accretion rates and prediction of pre-disturbance catchment sediment yield

The pre-and post-disturbance sedimentation rates measured by field-based data at the three catchment floodplain sampling sites are presented in Table 5.5. Despite being extracted from different parts of the catchment, floodplain accretion rates increased by similar levels (2 – 4 times) at all three sites. These increases are relatively low in comparison to increases observed from other regions suggesting that there has only been a modest increase in river sediment loads since European settlement (see Chapter 4).

Table 5.8 shows the increase in average floodplain accretion rates from pre-disturbance conditions to those determined by post-disturbance model D. Model D was used as the post-disturbance benchmark as it produced sediment yields most similar to those determined from the gauging station data (Figure 5.7). Model P1 (default GBR pre- disturbance parameters) results in an average increase in the rate of floodplain accretion of 22 times, which is an order of magnitude greater than the measured rates. Model P2, which included elevated pre-disturbance channel erosion inputs, significantly reduces the difference between the accretion rates of the pre- and post-disturbance periods.

Model P3, however, which increased natural hillslope erosion rates (in addition to increased channel inputs) produced floodplain accretion to rates similar to the measured rates. This is a relatively coarse measure of model performance and assumes that the measured increases in accretion rates occur uniformly across the entire floodplain.

Nonetheless, the low floodplain accretion rates produced by model P1 suggest that sediment loads are under-predicted when the default GBR parameters are used.

137 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

Table 5.8. Increases in floodplain accretion rates between each of the pre-disturbance model parameterisations and post-disturbance model D. Also indicated is the catchment sediment yield predicted by each of the pre-disturbance models.

Pre-disturbance Average increase in Catchment sediment Model floodplain accretion rate yield (kt y-1) P1 22 × 6 P2 8 × 19 P3 4 × 33

Model P1 produced a catchment sediment yield of 6 kt y-1 (Table 5.8) which is ~ 50 times lower than the yield estimated by the post-disturbance model A (319 kt y-1).

Increases of such magnitude since European settlement have been estimated for some catchments in southeastern Australia; and these were attributed to large scale river channel incision that occurred in that region (Olley and Wasson, 2003). There are no reports of such channel incision in the dry-tropical catchments of northeastern Australia.

In contrast, if post-disturbance model D and pre-disturbance model P3 are considered to be better representations based on closer agreement with empirical data, then catchment sediment yield would have only increased by ca. 4 times (33 kt y-1 to 133 kt y-1). This is a similar magnitude increase to that reported at the mouth of the Fitzroy River by previous modelling studies (e.g., Neil et al., 2002; McKergow et al., 2005). However, a four times increase in the sediment yield of Theresa Creek, which is a highly impacted headwater catchment, is unlikely to translate to such a large increase at the mouth of the

Fitzroy River because of the potential for sediment storage in downstream catchment sediment sinks (cf. Walling, 1988).

5.6.4. Future model development

This study has demonstrated that the predictive ability of SedNet can be improved with relatively simple, observation-based, changes to input parameters. However, all of the

138 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model model’s sediment inputs are currently based on the results of other modelling procedures (i.e., RUSLE, gully density models and a bank erosion relationship).

Further improvements in model predictions will result from refinements to these. This is primarily because the model assumes sediment transport is supply limited and therefore sediment loads are directly related to the amount of sediment that is available from hillslopes, gullies and riverbanks. Of equal importance to improvements in input data is the need to progress our understanding of the actual delivery of sediment from hillslopes and gullies. For example, assigning a uniform HSDR across an entire catchment is a very broad-brush approach. A better approach may be to define a spatially variable HSDR on the basis of such factors as rainfall and catchment topography (Lu et al., 2003a). Delivery of sediment from gullies also requires further investigation, as little information exists on current gully erosion rates. Furthermore, because of the high degree of connectivity with the river network, it is often assumed that gullies deliver most, if not all, of the sediment eroded from their banks to streams.

However, this is not always the case and in some environments gullies are effectively functioning as sinks of sediment (Zierholz et al., 2001).

Further information is also required on the relative contribution of gullies to the suspended sediment budget. For example, Rustomji (2006) found that the proportion of fine sediment contributed from gullies varied significantly with underlying rock type.

Given that gully erosion is an important contributor of sediment to many Australian rivers, easily obtained data on the lithological base of gullies may improve spatial predictions and allow identification of the most important sources.

Recent research in the wet topics of northeastern Australia also suggests that SedNet’s bank erosion algorithm produces poor results compared to field-derived measurements

139 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

(Bartley et al., 2008). Bartley et al. (2008) demonstrated that improved representations of channel slope and discharge can, however, produce significant improvements in bank erosion predictions. SedNet determines channel slope directly from a DEM, and the use of more accurate and finer resolution DEMs is likely to result in improved representations of channel slope (Montgomery and Foufoula-Georgiou, 1993; Finlayson and Montgomery, 2003). Improved representations of channel slope may also be achieved by determining channel slope directly from source data, such as contour intervals. The use of such data would avoid errors associated with using slopes interpolated from DEMs. An advanced version of SedNet has obtained channel slopes directly from contour intervals (see Wilkinson et al., 2006), however, this functionality has yet to be included in the existing version of the model.

This study assumes that modifying the input parameters is the most effective means of obtaining more accurate model estimates. However, some consideration must also be made of the accuracy of the model in quantifying catchment sediment sinks. The

SedNet model was developed for use at the regional-scale and some of its components are deliberately broad-scale. For example, floodplain sedimentation is assumed to occur at a uniform rate across the entire floodplain surface for each river link. Modelling the spatial variability of sediment deposition across floodplains is beyond the requirements of such a large-scale model (cf. Asselman and van Wijngaarden, 2002). However, it may be appropriate to consider the potential for variability in floodplain sedimentation at the large-scale in more detail. For example, within the Fitzroy River basin large-scale geologically controlled constrictions (or “bottlenecks”) in floodplains have been identified at almost 50 locations (Amos et al., 2008). The results of hydraulic modelling through one of these constrictions has shown that large backwater areas develop

140 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model upstream during flood events (Purvis-Smith, 2007). This has been confirmed by analysis of remotely sensed images of a recent large flood event (Croke et al., 2008).

These backwater areas are extensive and can remain flooded for many weeks after the flood peak has passed. As a result the floodplains upstream of the constrictions become areas of preferential floodplain sedimentation (Purvis-Smith, 2007). Given the large- scale and relative abundance of these features within the Fitzroy River basin, their incorporation into the floodplain deposition module of SedNet may be warranted.

Spatial information on floodplain constrictions can be relatively easily acquired from

DEMs (see Amos et al., 2008) and preliminary quantification of their significance in terms of floodplain sedimentation is currently available (Purvis-Smith, 2007).

In addition to incorporating areas of preferential floodplain deposition, it may also be necessary to reconsider the model’s approach to storage of fine sediment within river channels. The model currently assumes no long-term storage of fine sediment within river channels. However, within many Australian rivers large volumes of sediment are stored within in-channel benches, and in some environments these may be functioning as the primary floodplain (Nanson and Croke, 2002). It has been suggested that these features may be storing a large amount of the sediment that has been delivered to rivers since European settlement (Nanson and Croke, 2002). Currently, SedNet does not consider these features as potential sinks of sediment. Further research is required to quantify the role of in-channel benches as stores of fine sediment as well as to provide some conceptual or empirical basis to their spatial distribution.

5.7. Conclusions

Spatially distributed erosion/sediment yield models are valuable tools for identifying areas of significant erosion and sediment delivery at the large catchment- or regional-

141 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model scale. However, given the emphasis that is being placed on the results of such studies it is important that more attempts are made to validate the modelling results with independent data sources.

This study applies the spatially distributed erosion/sediment yield model SedNet to

Theresa Creek, a sub-catchment within the GBR catchment area. Substantial changes in model predictions are made when locally-derived data is used to modify model input parameters (e.g., bank geometry, gully history and morphology). Model predictions are validated against empirically-derived sediment load and sediment source data. Results indicate that significant improvement can be made to model predictions when catchment specific observations are used to refine model input parameters. However, there is residual error, especially in the prediction of pre-disturbance sediment loads.

No independent evidence of actual pre-disturbance erosion rates and sediment yield is available, but data on changes in floodplain accretion rates provide a means to verify the model outputs.

The use of generic input parameters used by previous SedNet applications within the

GBR catchment area results in overestimates of post-disturbance and underestimates of pre-disturbance sediment yields. This indicates that previous model applications may have overestimated the significance of post-European catchment disturbance on the sediment yields of the dry-tropical catchments draining to the GBR. Important catchment management decisions and end-of-catchment water quality targets are being set within GBR catchments on the basis of SedNet outputs. It is, therefore, essential that further catchment-specific data be collected that can be used to improve model inputs.

142 Chapter 5 - Validation of a spatially distributed erosion/sediment yield model

Improved representations of channel erosion (bank and gully) and deposition processes are likely to have greatest benefit in terms of model advancement. These processes remain poorly understand at the large scale. There are few catchment erosion/sediment yield models that represent depositional processes well. The absence of large-scale spatial patterns of floodplain deposition limits the potential to incorporate this data into an improved model.

143 CHAPTER 6

Summary and conclusions

Chapter 6 - Summary and conclusions

6.1. Summary and conclusions

The primary objective of this study was to improve both our understanding, and data availability, on spatial and temporal patterns of erosion and deposition in a dry-tropical catchment in northeastern Australia. This was achieved through a detailed investigation of changes in catchment sediment sources and sinks.

To summarise the major contributions of this thesis, this chapter returns to the original aims of the study as outlined in Chapter 1.

1. Determine the relative importance of major catchment sediment sources, both in

terms of spatial provenance and erosion process, and identify how sediment

sources have changed as a result of post-European land use changes.

Detailed knowledge of catchment sediment sources is essential if well-targeted catchment management decisions are to be made. Despite this, very little empirically- derived data exists on sediment sources from the vast dry-tropical catchments draining to the GBR in northeastern Australia. To gain a better understanding of the sediment sources within this region, a sediment tracing approach was used to determine major sediment sources from a representative source area subcatchment. Geochemical and fallout radionuclide properties of river bed sediment were used to determine the relative contribution of recent sediment sources, both in terms of spatial provenance and erosion type. This is the first study to apply such an approach within this region of Australia.

It was found that sheetwash and rill erosion from cultivated land and channel erosion from grazing land are the dominant sources of sediment to the river system. Evidence suggests that the dominant form of channel erosion is gully headcut and sidewall erosion. Cultivation is a minor land use within the two largest dry-tropical catchments

146 Chapter 6 - Summary and conclusions in the region (Fitzroy River, 140 000 km2; Burdekin River, 130 000 km2), therefore, channel sources are likely to be the largest contributor of sediment to these rivers. This finding is contrary to the widely held understanding that hillslope (sheetwash and rill) erosion is the most significant source of sediment to these river systems. This has important implications for future erosion management and rehabilitation within these catchments.

Based on the river bed sediment tracing results and information on catchment lithology and land use, the relationship between spatial provenance and erosion type was also explored. Channel erosion is mainly derived from granite, metasediment and clastic based areas, while basalt-based areas contribute a large proportion of cultivation- derived sediment. This association was used to interpret temporal changes in sediment sources as indicated by the geochemical record within a floodplain core. The floodplain core was dated primarily by OSL dating of single grains of quartz. This is the first known study to determine sediment sources through time using an OSL-dated core. The results showed that sediment sources in the catchment have changed in direct response to land use changes since European settlement (ca. 1850 AD). A phase of channel incision began in the late nineteenth century. This channel erosion, which is likely to have been dominated by gully network expansion, was probably the result of reduced ground cover, soil compaction, and tree clearance associated with increased grazing pressure. During the mid-twentieth century, in response to the rapid establishment of intensive cultivation, the dominant sediment source shifted to cultivated sources.

Although the contribution of sediment from cultivated land has declined more recently, as indicated by the river bed sediment tracing, it remains a significant source of sediment.

147 Chapter 6 - Summary and conclusions

2. To investigate the application of 137Cs dating techniques for determining rates of

recent floodplain sedimentation in a low fallout, southern hemisphere

environment.

The dating of floodplain sediment was an important component of this study. Optically stimulated luminescence dating provided the majority of floodplain sediment age data.

Caesium-137 dating was used to provide both additional sediment age data and verification of the general accuracy of the youngest OSL burial ages. Caesium-137 has been previously widely used to determine medium-term (ca. last 50 years) rates of floodplain, lake and reservoir deposition. There are few studies, however, from dry, low-latitude southern hemisphere locations where fallout levels were low. The need to acquire data on recent floodplain sedimentation rates in this study provided the opportunity to test the applicability of 137Cs dating within such an environment.

The two most commonly used methods for determining floodplain sedimentation rates using 137Cs, depth profiles and total inventories, were examined. Sediment and soil cores were obtained from three floodplain sites and three undisturbed reference sites.

The 137Cs depth profiles of the floodplain sites exhibited forms consistent with being depositional sites. Information on the rates of 137Cs migration through local soils was obtained from the reference site soil cores. This data was used in an advection-diffusion model to determine rates of floodplain accretion for the period between the year of the first known overbank discharge event (1954) after the first year of significant fallout

(1951) and 2006, the year of core extraction. The model determined that the three floodplain sites had accreted between ~ 36 mm and ~ 109 mm since ca. 1954.

In contrast, the total 137Cs inventory approach resulted in two of the three depositional sites having total 137Cs inventories that were not significantly different from undisturbed

148 Chapter 6 - Summary and conclusions reference sites. This was attributed to low fallout levels and the dominance of catchment inputs from sediment sources low in concentrations of 137Cs (i.e., channels and cultivated sources). Consequently, it is unlikely that the total 137Cs inventory approach in such an environment would be an effective means of determining rates of floodplain sedimentation.

3. To compare and contrast pre- and post-disturbance rates of alluvial sedimentation

and interpret how changes in rates of sedimentation provide evidence of changes

in catchment sediment flux.

Detailed information on floodplain sedimentation rates is lacking from many parts of the world. Optically stimulated luminescence dating of single grains of quartz and 137Cs depth profiles were used to determine the pre- and post-disturbance rates of floodplain sedimentation. Floodplain sedimentation rates for the late-Holocene pre-disturbance period (ca. 1600 y BP – ca. 1850 AD) were low and ranged between 0.4 – 0.9 mm y-1.

In the post-settlement period floodplain sedimentation rates increased by 3 – 4 times.

These increases are attributed to increased sediment loads as a result of human-induced catchment disturbance. Two distinct post-settlement periods were identified based on sedimentation rates: (i) an early post-European settlement period (ca. 1850 AD – mid- twentieth century), when sedimentation rates were at their highest (1.4 – 4.0 mm y-1), and (ii) a late post-settlement period when sedimentation rates reduced as a result of decreased catchment sediment loads (0.8 – 2.5 mm y-1). Reductions in catchment sediment loads are likely to be related to declining sediment generation from a stabilising gully network and improvements in land management practices.

The estimated 3 - 4 times increase in post-disturbance floodplain sedimentation rates is low compared to rates reported elsewhere. It is proposed here that low magnitude

149 Chapter 6 - Summary and conclusions

increases in sedimentation rates are the result of modest increases in post-European river sediment loads. This is attributed to naturally high sediment loads that are likely during large events in dry-tropical catchments. Naturally high flood sediment concentrations may be related to the cyclical nature of the climate within the dry-tropics that results in extended periods of drought followed by above-average wet periods.

During droughts, reduced groundcover increases the potential for erosion in any subsequent rainfall events. Furthermore, unlike many catchments in temperate southeastern Australia, the dry-tropical catchments of northeastern Australia are characterised by naturally well-connected river channel networks. Well established channel connectivity would have contributed to efficient transport of sediment through the river system prior to catchment disturbance.

In-channel benches are alluvial deposits that have been described in a number of environments; however, they appear to be relatively common features in Australian river systems. Studies from other regions of Australia suggest that bench deposition is a response to the influx of post-European generated sediment. This study presents one of the first attempts to date and quantify the sediment storage importance of in-channel benches. Optically stimulated luminescence and 137Cs dating demonstrated that, in the study catchment, they are modern, rapidly accreting (13.6 - 23.8 mm y-1) permanent features. The young nature of the dated bench sediment (< 100 years) suggests that in- channel benches in this region are also a response to an influx of post-disturbance sediment. Although the in-channel benches are composed primarily of sand-sized sediment, because of their widespread occurrence, they may still be significant stores of fine sediment.

150 Chapter 6 - Summary and conclusions

4. Use empirically-derived data to validate a spatially distributed erosion/sediment

yield model with a view to providing improved estimates of pre- and post-

disturbance catchment sediment yield.

The spatially distributed erosion/sediment yield model, SedNet, has been used to determine catchment sediment yields and sediment sources for a number of Australian catchments. Model outputs are being used to guide catchment management decisions, often without verification with empirical data. This study presented one of the first attempts to validate the results of SedNet with field-based data.

For the study catchment, models representing post-disturbance catchment conditions were generated using both previously used “generic” catchment input parameters and input parameters modified on the basis of catchment-specific observations (e.g., bank geometry, gully history and morphology). Model predictions of sediment loads were evaluated against suspended sediment sample-based load estimations. The model’s spatial representations of major sediment sources were assessed against sediment source data acquired from fallout radionuclide tracing of river bed sediment. Pre-disturbance models were also produced using previously used GBR catchment input parameters and modifications of these based on reasonable representations of pre-disturbance catchment processes. No independent evidence of actual pre-disturbance erosion rates and sediment yield was available. However, changes in rates of floodplain sedimentation determined from OSL and 137Cs profile dating did provide surrogate measurements of changes in sediment loads. These were used to evaluate the model’s performance.

The results show that the use of catchment-specific data to modify model input parameters improves post-disturbance estimates of sediment loads. The generic GBR parameters produced end-of-catchment sediment yields of 6 kt y-1 (pre-disturbance) and

151 Chapter 6 - Summary and conclusions

319 kt y-1 (post-disturbance). The improved parameterisations resulted in sediment yields of 33 kt y-1 (pre-disturbance) and 133 kt y-1 (post-disturbance). This suggests that the use of generic input parameters, as used by previous large-scale applications of

SedNet within the GBR catchment area, overestimate post-disturbance sediment yields and underestimate of pre-disturbance sediment yields. This indicates that previous model applications may, therefore, have overestimated the significance of post-

European catchment disturbance on the sediment yields of the dry-tropical catchments draining to the GBR.

6.2. Future research

On the basis of the results presented in this thesis, a number of suggestions for future research can be made:

• A significant contribution of this research has been the demonstration of the

importance of channel sources within a dry-tropical catchment in northeastern

Australia. Although the subcatchment examined in this study is believed to be

representative of the upper catchment areas within the region, there are likely to be

local differences that may affect the importance of different sources. To confirm the

importance of channel sources within the region, further field-based research should

be conducted within other regional catchments. Future work should also attempt to

decouple the relative contribution of the different forms of channel erosion (gully

and river bank). Current fallout radionuclide tracing techniques are unable to

provide this resolution. Further development of sediment tracing techniques,

whereby sediment derived from river banks and gullies is differentiated on the basis

of physical and/or chemical properties, is required.

152 Chapter 6 - Summary and conclusions

• Based on the relatively low increases in floodplain sedimentation rates, this study

proposed that human-induced increases in the sediment yield from the large dry-

tropical GBR catchments may not be as large as previously suggested. This study

deliberately targeted a subcatchment within the upper reaches of the Fitzroy River

basin with known high rates of erosion. Due to the proximity of the floodplains to

source areas, elevated sediment loads and hence floodplain sedimentation, would be

expected. Further work should be carried out to investigate floodplain accretion

rates in other areas, particularly the vast, low-lying parts of the large dry-tropical

catchments that drain to the GBR.

• Detailed analysis of the significance of in-channel benches as long-term stores of

sediment is required. Such research should include analysis of the spatial

distribution of these features with the view to quantifying their role in sediment

storage. In addition, because fine sediment is generally of most interest with regards

to downstream effects, further work is required to determine the significance of in-

channel benches as stores of fine sediment.

• The model validation carried out in this study identified a number of areas where

improvements are required in our understanding of catchment-scale processes. Two

such areas are the representation of delivery of sediment from hillslopes and

sediment deposition processes. The use of a universal hillslope delivery ratio, as

used by the SedNet model, is a broad brush approach and future work should focus

on improving our understanding of the factors that influence the variability in

delivery rates. In addition, few spatially-distributed erosion/sediment yield models

represent depositional processes well at the catchment scale. Future modelling

153 Chapter 6 - Summary and conclusions

should ensure that this important catchment sediment budget component is

adequately represented.

154 Chapter 6 - Summary and conclusions

REFERENCES

Ahearn, D.S., Sheibley, R.W., Dahlgren, R.A., Anderson, M., Johnson, J., Tate, K.W., 2005. Land use and land cover influence on water quality in the last free-flowing river draining the western Sierra Nevada, California. Journal of Hydrology 313, 234-247.

Aitken, M.J., 1985. Thermoluminescence Dating. Academic Press, London, 359 pp.

Aitken, M.J., 1998. An Introduction to Optical Dating: The Dating of Quaternary Sediments by the Use of Photon-stimulated Luminescence. Oxford University Press, Oxford, UK, 282 pp.

Alexander, J., Fielding, C., 1997. Gravel antidunes in the tropical Burdekin River, Queensland, Australia. Sedimentology 44, 327-337.

Allison, M.A., Kuehl, S.A., Martin, T.C., Hassan, A., 1998. Importance of flood-plain sedimentation for river sediment budgets and terrigenous input to the oceans: insights from the Brahmaputra-Jamuna River. Geology 26(2), 175-178.

Amos, K.J., Croke, J.C., Hughes, A.O., Chapman, J., Takken, I., Lymburner, L., 2008. A catchment-scale assessment of anabranching in the 143 000 km2 Fitzroy River catchment, north-eastern Australia. Earth Surface Processes and Landforms 33(8), 1222-1241.

Amos, K.J., Croke, J.C., Timmers, H., Owens, P.N., Thompson, C., 2009. The application of caesium-137 measurements to investigate floodplain deposition in a large semi-arid catchment in Queensland, Australia. Earth Surface Processes and Landforms 34(4), 515-529.

Amos, K.J., Croke, J.C., Timmers, H., Owens, P.N., Thompson, C., in press. The application of caesium-137 measurements to investigate floodplain deposition in a large semi-arid catchment in Queensland, Australia. Earth Surface Processes and Landforms.

155 References

Anon, 2006. Ensham Central Project Environmental Impact Statement, unpublished report prepared for Ensham Resource Pty. Ltd. by Hansen Consulting, Brisbane, Australia. http://www.ensham.com.au/updated/eis.asp.

Asselman, N.E., 1999. Suspended sediment dynamics in a large drainage basin: the River Rhine. Hydrological Processes 13, 1437-1450.

Asselman, N.E., Middelkoop, H., van Dijk, P.M., 2003. The impact of changes in climate and land use on soil erosion, transport and deposition of suspended sediment in the River Rhine. Hydrological Processes 17, 3225-3244.

Asselman, N.E.M., van Wijngaarden, M., 2002. Development and application of a 1D floodplain sedimentation model for the River Rhine in The Netherlands. Journal of Hydrology 268, 127-142.

Bartley, R., Henderson, A., Prosser, I.P., Hughes, A.O., McKergow, L., Lu, H., Brodie, J., Bainbridge, Z., Roth, C.H., 2003. Patterns of erosion and sediment and nutrient transport in the Herbert River catchment, Queensland, CSIRO Land and Water Consultancy Report. CSIRO Land and Water.

Bartley, R., Olley, J., Henderson, A., 2004. A sediment budget for the Herbert River catchment, , Australia. In: Golosov, V., Belyaev, V., Walling, D.E. (Eds.), Sediment Transfer through the Fluvial System. IAHS, Moscow, pp. 147-154.

Bartley, R., Roth, C.H., Ludwig, J., McJannet, D., Liedloff, A., Corfield, J., Hawdon, A., Abbott, B., 2006. Runoff and erosion from Australia's tropical semi-arid rangelands: influence of ground cover for differing space and time scales. Hydrological Processes 20, 3317-3333.

Bartley, R., Hawdon, A., Post, D.A., Roth, C.H., 2007. A sediment budget for a grazed semi-arid catchment in the Burdekin basin, Australia. Geomorphology 87, 302- 321.

156 Chapter 6 - Summary and conclusions

Bartley, R., Keen, R.J., Hawdon, A.A., Hairsine, P.B., Disher, M.G., Kinsey- Henderson, A.E., 2008. Bank erosion and channel width change in a tropical catchment. Earth Surface Processes and Landforms DOI: 10.1002/esp.1678.

Basher, L.R., 2000. Surface erosion assessment using 137Cs: examples from New Zealand. Acta Geologica Hispanica 35(3-4), 219-228.

Belperio, A.P., 1983. Terrigenous sedimentation in the central Great Barrier Reef lagoon: a model from the Burdekin region. BMR Journal of Australian Geology and Geophysics 8(3), 179-190.

Bishop, P., Campbell, B., McFadden, C., 1991. Absence of caesium-137 from recent sediments in eastern Australia - indications of catchment processes? Catena 18, 61-69.

Blong, R.J., Graham, O.P., Veness, J.A., 1982. The role of sidewall processes in gully development. Earth Surface Processes and Landforms 7, 381-385.

BoM, 2007. Statistics for Clermont Post Office climate station, Bureau of Meteorology, http://www.bom.gov.au/climate/averages/tables/cw_035019.shtml. Date accessed May 2007.

Bostock, H.C., Brooke, B.P., Ryan, D.A., Hancock, G., Pietsch, T., Packett, B., Harle, K., 2007. Holocene and modern sediment storage in the subtropical macrotidal river estuary, southeast Queensland, Australia. Sedimentary Geology 201, 321- 340.

Bramley, R.G.V., Roth, C.H., 2002. Land-use effects on water quality in an intensively managed catchment in the Australian humid tropics. Marine and Freshwater Research 53(5), 931-940.

Brierley, G.J., Murn, C.P., 1997. European impacts on downstream sediment transfer and bank erosion in Cobargo catchment, New South Wales, Australia. Catena 31, 119-136.

157 References

Brierley, G.J., Cohen, T., Fryirs, K., Brooks, A., 1999. Post-European changes to the fluvial geomorphology of Bega catchment, Australia: implications for river ecology. Freshwater Biology 41, 839-848.

Brodie, J.E., McKergow, L.A., Prosser, I.P., Furnas, M., Hughes, A.O., Hunter, H., 2003. Sources of sediment and nutrient exports to the Great Barrier Reef World Heritage Area. ACTFR Report No. 03/11, Australian Centre for Tropical Freshwater Research, James Cook University, Townsville.

Brodie, J.E., Mitchell, A.W., 2005. Nutrients in Australian tropical rivers: changes with agricultural development and implications for receiving environments. Marine and Freshwater Research 56, 279-302.

Brooks, A., Spencer, J., Knight, J., 2007. Alluvial gully erosion in Australia's tropical rivers: A conceptual model as a basis for a remote sensing mapping procedure. In: Wilson, A., Dehaan, R., Watts, R., Page, K., Bowmer, K., Curtis, A. (Eds.), 5th Australian Stream Management Conference, Albury 21-25 May, 2007. pp. 43-48.

Brooks, A.P., Brierley, G.J., 1997. Geomorphic responses of lower Bega River to catchment disturbance, 1851-1926. Geomorphology 18, 291-304.

Brooks, A.P., Brierley, G.J., 2004. Framing realistic river rehabilitation targets in light of altered sediment supply and transport relationships: lessons from East Gippsland, Australia. Geomorphology 58, 107-123.

BRS, 2008. Australian Natural Resources Data Library, http://adl.brs.gov.au/anrdl, Accessed July 2008.

Brune, G.M., 1953. Trap efficiency of reservoirs. Transactions of the American Geophysical Union 34, 407-418.

Burns, M., Brimblecombe, J. (Eds.), 2006. Worth the Gamble. The First Fifty Years. Personal Recollections From People Who Shaped the Peak Downs/Retro Holdings 1956 to 2006. Capella Pioneer Village, Capella, Queensland, Australia, 133 pp.

158 Chapter 6 - Summary and conclusions

Calvert, M., Simpson, J., Adsett, K., 2000. Land use mapping of the Fitzroy Catchment. Fitzroy Catchment Implementation Project Final Report, The State of Queensland, Department of Natural Resources, Coorparoo.

Carroll, C., Halpin, M., Burger, P., Bell, K., Sallaway, M.M., Yule, D.F., 1997. The effect of crop type, crop rotation, and tillage practice on runoff and soil loss on a Vertisol in central Queensland. Australian Journal of Soil Research 35, 925-939.

Carson, E.C., 2006. Hydrologic modeling of flood conveyance and impacts of historic overbank sedimentation on West Fork Black's Fork, Uinta Mountains, northeastern Utah, USA. Geomorphology 75, 368-383.

Carthew, K.D., Drysdale, R.N., 2003. Late Holocene fluvial change in a tufa-depositing stream: Davys Creek, New South Wales, Australia. Australian Geographer 34(1), 123-139.

Chaplot, V., Le Bissonnais, Y., 2000. Field measurements of interrill erosion under different slopes and plot sizes. Earth Surface Processes and Landforms 25(2), 145-153.

Chappell, A., 1999. The limitations of using 137Cs for estimating soil distribution in semi-arid environments. Geomorphology 29, 135-152.

Chiew, F.H.S., McMahon, T.A., 1993. Assessing the adequacy of catchment streamflow yield estimates. Australian Journal of Soil Research 31, 665-680.

Cisternas, M., Araneda, A., Martínez, P., Pérez, S., 2001. Effects of historical land use on sediment yield from a lacustrine watershed in central Chile. Earth Surface Processes and Landforms 26, 63-76.

Cogle, A.L., Carroll, C., Sherman, B.S., 2006. The use of SedNet and ANNEX models to guide GBR catchment sediment and nutrient target setting, Department of Natural Resources, Mines and Water.

Collins, A.L., Walling, D.E., Leeks, G.J.L., 1997a. Use of the geochemical record preserved in floodplain depositions to reconstruct recent changes in river basin sediment sources. Geomorphology 19, 151-167.

159 References

Collins, A.L., Walling, D.E., Leeks, G.J.L., 1997b. Fingerprinting the origin of fluvial suspended sediment in larger river basins: combining assessment of spatial provenance and source type. Geografiska Annaler 79A(4), 239-254.

Collins, A.L., Walling, D.E., Leeks, G.J.L., 1997c. Source type ascription for fluvial suspended sediment based on a quantitative composite fingerprinting technique. Catena 29, 1-27.

Collins, A.L., Walling, D.E., Leeks, G.J.L., 1998. Use of composite fingerprints to determine the provenance of the contemporary suspended sediment loads transported by rivers. Earth Surface Processes and Landforms 23, 31-52.

Collins, A.L., Walling, D.E., 2002. Selecting fingerprint properties for discriminating potential suspended sediment sources in river basins. Journal of Hydrology 261, 218-244.

Collins, A.L., Walling, D.E., 2004. Documenting catchment suspended sediment sources: problems, approaches and prospects. Progress in Physical Geography 28(2), 159-196.

Collins, A.L., Walling, D.E., 2007. Fine-grained bed sediment storage within the main channel systems of the Frome and Piddle catchments, Dorset, UK. Hydrological Processes 21, 1448-1459.

Cooper, C.M., 1993. Biological effects of agriculturally derived surface-water pollutants on aquatic systems. Journal of Environmental Quality 22(3), 402-408.

Croke, J.C., Purvis-Smith, D., Thompson, C.J., Lymbuner, L., 2008. The effect of local scale valley constrictions on flood inundation and catchment-scale sediment delivery in the Fitzroy River Basin, Australia, Sediment Dynamics in Changing Environments. IAHS Publ. 325, Christchurch, New Zealand.

De Roo, A.P.J., 1998. Modelling runoff and sediment transport in catchments using GIS. Hydrological Processes 12(6), 905-922.

160 Chapter 6 - Summary and conclusions de Vente, J., Poesen, J., Verstraeten, G., Van Rompaey, A., Govers, G., 2008. Spatially distributed modelling of soil erosion and sediment yield at regional scales in Spain. Global and Planetary Change 60, 393-415.

DeRose, R.C., Prosser, I.P., Weisse, M., Hughes, A.O., 2003. Patterns of Erosion and Sediment and Nutrient Transport in the Murray-Darling Basin. CSIRO Land and Water Technical Report 32/03, CSIRO Land and Water, Canberra.

Dimitriadis, S., Cranston, P.S., 2001. An Australian Holocene climate reconstruction using Chironomidae from a tropical volcanic maar lake. Palaeogeography, Palaeoclimatology, Palaeoecology 176(1-4), 109-131.

Dirszowsky, R.W., 2004. Bed sediment sources and mixing in the glacierized Upper Fraser River watershed, East-Central British Columbia. Earth Surface Processes and Landforms 29, 533-552.

Dodson, J.R., Mooney, S.D., 2002. An assessment of historic human impact on south- eastern Australian environmental systems, using late Holocene rates of environmental change. Australian Journal of Botany 50, 455-464.

Doering, C., Akber, R., Heijnis, H., 2006. Vertical distributions of 210Pb excess, 7Be and 137Cs in selected grass covered soils in Southeast Queensland, Australia. Journal of Environmental Radioactivity 87, 135-147.

Dougall, C., Packett, R., Caroll, C., 2005. Application of the SedNet model in partnership with the Fitzroy Basin community. In: Zerger, A., Argent, R.M. (Eds.), MODSIM 2005 International Congress on Modelling and Simulation, Melbourne December, 2005. Modelling and Simulation Society of Australia and New Zealand, pp. 1119-1125.

Dougall, C., Packett, R., Carroll, C., Sherman, B.S., Read, A., Chen, Y., Brodie, J., 2006. Sediment and nutrient modelling in the Fitzroy NRM region. In: Cogle, C., Carroll, C., Sherman, B.S. (Eds.), The use of SedNet and ANNEX models to guide GBR catchment sediment and nutrient target setting. Department of Natural Resource, Mines and Water, Rockhampton, Australia, pp. 27 pp.

161 References

Dougall, C., Carroll, C., Herring, M., Trevithick, R., 2007. Sednet modelling in the Fitzroy basin (2007); spatially variable ground cover and revised gully layers can potentially generate significant changes in erosion sources and patterns. In: Oxley, L., Kulasiri, D. (Eds.), MODSIM 2007 International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand, Christchurch, New Zealand, pp. pp. 74-80.

Douglas, G., Palmer, M., Caitcheon, G., Orr, P., 2007. Identification of sediment sources to Lake Wivenhoe, south-east Queensland, Australia. Marine and Freshwater Research 58, 793-810.

Douglas, G.B., Ford, P.W., Palmer, M., Noble, R.M., Packett, R., 2006. Fitzroy River Basin, Queensland, Australia. I. Identification of sediment sources in impoundments and flood events. Environmental Chemistry 3, 364-376.

Douglas, I., 1967. Man, vegetation and the sediment yield of rivers. Nature 215, 925- 928.

Droppo, I.G., Ongley, E.D., 1994. Flocculation of suspended sediment in rivers of southeastern Canada. Water Research 28(8), 1799-1809.

Droppo, I.G., Stone, M., 1994. In-channel surficial fine-grained sediment laminae. 1. Physical characteristics and formational processes. Hydrological Processes 8(2), 101-111.

Duan, N., 1983. Smearing estimate: a nonparametric retransformation method. Journal of the American Statistical Association 78, 605-610.

Edwards, K., 1993. Soil erosion and conservation in Australia. In: Pimentel, D. (Ed.), World soil erosion and conservation. Cambridge University Press, Cambridge, pp. 147-169.

Erskine, W.D., Livingstone, E.A., 1999. In-channel benches: the role of floods in their formation and destruction on bedrock confined rivers. In: Miller, A.J., Gupta, A. (Eds.), Varieties of Fluvial Form. Wiley, Chichester, pp. 445- 475.

162 Chapter 6 - Summary and conclusions

Eyles, R.J., 1977. Changes in drainage networks since 1820, Southern Tablelands, NSW. Australian Geographer 13, 377-387.

Fabricius, K.E., De'ath, G., 2004. Identifying ecological change and its causes: a case study on coral reefs. Ecological Applications 14(5), 1448-1465.

Fabricius, K.E., 2005. Effects of terrestrial runoff on the ecology of corals and coral reefs: review and synthesis. Marine Pollution Bulletin 50, 125-146.

Faure, G., 1998. Principles and Applications of Geochemistry: A Comprehensive Textbook for Geology Students. Prentice Hall, New Jersey.

Fentie, B., Joo, M., Hunter, H., Marsh, N., Carroll, C., Dougall, C., 2005. Comparison of mean annual suspended loads estimated by the SedNet model and rating curves in the Fitzroy Catchment, Australia. In: Zerger, A., Argent, R.M. (Eds.), MODSIM 2005 International Congress on Modelling and Simulation, Melbourne December, 2005. Modelling and Simulation Society of Australia and New Zealand, pp. 1133-1139.

Finlayson, B.L., Brizga, S.O., 1993. Anastomosing channels and arroyo development on the Nogoa River, Central Queensland, Australia. Sedimentary Geology 85, 179- 190.

Finlayson, D.P., Montgomery, D.R., 2003. Modeling large-scale fluvial erosion in geographic information systems. Geomorphology 53, 147-164.

Florsheim, J.L., Mount, J.F., 2003. Changes in lowland floodplain sedimentation processes: pre-disturbance to post-rehabilitation, Cosumnes River, CA. Geomorphology 56, 305-323.

Foster, I.D.L., Lees, J.A., Owens, P.N., Walling, D.E., 1998. Mineral magnetic characterization of sediment sources from an analysis of lake and floodplain sediments in the catchments of the Old mill Reservoir and Slaton Ley, South Devon, UK. Earth Surface Processes and Landforms 23, 685-703.

Freebairn, D.M., Boughton, W.C., 1981. Surface runoff experiments on the eastern . Australian Journal of Soil Research 19(2), 133-146.

163 References

Freebairn, D.M., Wockner, G.H., 1986. A study of soil erosion on vertisols of the eastern Darling Downs, Queensland. I Effects of surface conditions on soil movement within contour bay catchments. Australian Journal of Soil Research 24, 135-158.

Fryirs, K., Brierley, G.J., 1999. Slope-channel decoupling in Wolumla catchment, New South Wales, Australia: the changing nature of sediment sources following European settlement. Catena 35, 41-63.

Furnas, M., 2003. Catchments and Corals: Terrestrial Runoff to the Great Barrier Reef. Australian Institute of Marine Science and CRC Reef Research Centre, Townsville, 334 pp.

Galbraith, R.F., Roberts, R.G., Laslett, G.M., Yoshida, H., Olley, J.M., 1999. Optical dating of single and multiple grains of quartz from Jinmium rock shelter, northern Australia. Part I: Experimental design and statistical models. Archaeometry 41, 339-364.

Gale, S.J., Haworth, R.J., 2005. Catchment-wide soil loss from pre-agricultural times to the present: transport- and supply-limitation of erosion. Geomorphology 68, 314-333.

Gallant, J.C., Dowling, T.I., 2003. A multiresolution index of valley bottom flatness for mapping depositional areas. Water Resources Research 39(12), 1347, doi:10.1029/2002WR001426.

Gell, P.A., Bulpin, S., Wallbrink, P.J., Hancock, G., Bickford, S., 2005. Tareena Billabong - a palaeolimnological history of an ever-changing wetland, Chowilla Floodplain, lower Murray-Darling Basin, Australia. Marine and Freshwater Research 56, 441-456.

Gomez, B., Eden, D.N., Peacock, D.H., Pinkney, E.J., 1998. Floodplain construction by recent, rapid vertical accretion: Waipaoa River, New Zealand. Earth Surface Processes and Landforms 23, 405-413.

164 Chapter 6 - Summary and conclusions

Goodbred, S.L., Kuehl, S.A., 1998. Floodplain processes in the Bengal Basin and the storage of Ganges-Brahmaputra river sediment: an accretion study using 137Cs and 210Pb geochronology. Sedimentary Geology 121, 239-258.

Grayson, R.B., Moore, I.D., McMahon, T.A., 1992. Physically Based Hydrologic Modeling 1. A Terrain-Based Model for Investigative Purposes. Water Resources Research 28(10), 2639-2658.

Hancock, G.R., Evans, K.G., 2006. Gully position, characteristics and geomorphic thresholds in an undisturbed catchment in northern Australia. Hydrological Processes 20, 2935-2951.

Hateley, L., Armour, J., Pitt, G., Cogle, L., 2007. Use of SedNet model to establish sediment export targets for catchments of the wet tropics draining to the Great Barrier Reef. In: Oxley, L., Kulasiri, D. (Eds.), MODSIM 2007 International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand, Christchurch, New Zealand, pp. 74-80.

He, Q., Owens, P.N., 1995. Determination of suspended sediment provenance using Caesium-137, unsupported Lead-210 and Radium-226: a numerical mixing model approach. In: Foster, I.D.L., Gurnell, A.M., Webb, B.W. (Eds.), Sediment and Water Quality in River Catchments. John Wiley, New York, pp. 207-227.

Hessel, R., van den Bosch, R., Vigiak, O., 2006. Evaluation of the LISEM soil erosion model in two catchments in the East African Highlands. Earth Surface Processes and Landforms 31(4), 469-486.

Houben, P., Hoffmann, T., Zimmermann, A., Dikau, R., 2006. Land use and climatic impacts on the Rhine system (RheinLUCIFS): Quantifying sediment fluxes and human impact with available data. Catena 66(1-2), 42-52.

Hovius, N., Stark, C.P., Hao-Tsu, C., Jiun-Chuan, L., 2000. Supply and removal of sediment in a landslide-dominated mountain belt: Central Range, Taiwan. The Journal of Geology 108, 73-89.

165 References

Hudson, P.F., 2003. Floodplains: environment and process. Geomorphology 56, 223- 224.

Hughes, A.O., Prosser, I.P., Stevenson, J., Scott, A., Lu, H., Gallant, J., Moran, C.J., 2001. Gully density mapping for Australian river basins. In: Rutherfurd, I., Sheldon, F., Brierley, G., Kenyon, C. (Eds.), Third Australian Stream Management Conference, Brisbane, Australia 27-29 August 2001. Cooperative Research Centre for Catchment Hydrology, pp. 323-327.

Hughes, A.O., Prosser, I.P., 2003. Gully and riverbank erosion mapping for the Murray- Darling Basin. CSIRO Land and Water Technical Report 3/03, CSIRO Land and Water, Canberra.

Hunter, H., Walton, R.S., 2008. Land use effects on fluxes of suspended sediment, nitrogen and phosphorus from a river catchment of the Great Barrier Reef, Australia. Journal of Hydrology 356(1-2), 131-146.

Jansson, M.B., 1988. A global survey of sediment yield. Geografiska Annaler 70A(1-2), 81-98.

Jetten, V., Govers, G., Hessel, R., 2003. Erosion models: quality of spatial predictions. Hydrological Processes 17, 887-900.

Johnson, W.R., 1982. The Call of the Land: A History of Queensland to the Present Day. Jacaranda Press, Milton, Queensland, 171 pp.

Joo, M., Yu, B., Fentie, B., Carroll, C., 2005. Estimation of long-term sediment loads in the Fitzroy catchment, Queensland, Australia. In: Zerger, A., Argent, R.M. (Eds.), MODSIM 2005 International Congress on Modelling and Simulation, Melbourne December, 2005. Modelling and Simulation Society of Australia and New Zealand, pp. 1161-1167.

Keesstra, S.D., van Huissteden, J., Vandenberghe, J., Van Dam, O., de Gier, J., Pleizier, I.D., 2005. Evolution of the morphology of the river Dragonja (SW Slovenia) due to land-use changes. Geomorphology 69, 191-207.

166 Chapter 6 - Summary and conclusions

Kemp, J., 2004. Flood channel morphology of a quiet river, the Lachlan downstream from Cowra, southeastern Australia. Geomorphology 60, 171-190.

Kennish, M.J., 2002. Environmental threats and environmental future of estuaries. Environmental Conservation 29(1), 78-107.

Kinsey-Henderson, A.E., Post, D.A., Prosser, I.P., 2005. Modelling sources of sediment at sub-catchment scale: An example from the Burdekin catchment, North Queensland, Australia. Mathematics and Computers in Simulation 69, 90-102.

Kinsey-Henderson, A.E., Sherman, B., Bartley, R., 2007. Improved SedNet modelling of grazing lands in the Burdekin catchment. CSIRO Land and Water Science Report 63/07, CSIRO Land and Water, Townsville.

Kirchner, J.W., 2006. Getting the right answers for the right reasons: Linking measurements, analyses, and models to advance the science of hydrology. Water Resources Research 42, W03S04, doi:10.1029/2005WR004362.

Knox, J.C., 1977. Human impacts on Wisconsin stream channels. Annals of the Association of American Geographers 67(3), 323-342.

Knox, J.C., 1987. Historical Valley Floor Sedimentation in the Upper Mississippi Valley. Annals of the Association of American Geographers 77(2), 224-244.

Knox, J.C., 2001. Agricultural influence on landscape sensitivity in the Upper Mississippi River Valley. Catena 42, 193-224.

Knox, J.C., 2006. Floodplain sedimentation in the Upper Mississippi Valley: natural versus human accelerated. Geomorphology 79, 286-310.

Kurashige, Y., Fusejima, Y., 1997. Source identification of suspended sediment from grain-size distributions: I. Application of nonparametric statistical tests. Catena 31, 39-52.

Lal, R., 2003. Soil erosion and the global carbon budget. Environment International 29, 437-450.

167 References

Lane, L.J., Hernandez, M., Nichols, M., 1997. Processes controlling sediment yield from watersheds as functions of spatial scale. Environmental Modelling & Software 12(4), 355-369.

Langedal, M., 1997. The influence of a large anthropogenic sediment source on the fluvial geomorphology of the Knabeåna-Kvina rivers, Norway. Geomorphology 19, 117-132.

Leahy, P.J., Tibby, J., Kershaw, A.P., Heijnis, H., Kershaw, J.S., 2005. The impact of European settlement on Bolin Billabong, a Yarra River floodplain lake, Melbourne, Australia. River Research and Applications 21, 131-149.

Lecce, S.A., Pavlowsky, R.T., 2001. Use of mining-contaminated sediment tracers to investigate the timing and rates of historical flood plain sedimentation. Geomorphology 38, 85-108.

Leigh, D.S., Srivastava, P., Brook, G.A., 2004. Late Pleistocene braided rivers of the Atlantic Coastal Plain, USA. Quaternary Science Reviews 23, 65-84.

Lenat, D.R., Crawford, J.K., 1994. Effects of land-use on water quality and aquatic biota of 3 North-Carolina piedmont streams. Hydrobiologia 294(3), 185-199.

Leslie, C., Hancock, G.J., 2007. Estimating the date corresponding to the horizon of the first detection of 137Cs and 239+240Pu in sediment cores. Journal of Environmental Radioactivity doi:10.1016/j.jenvrad.2007.08.016.

Lewis, S.E., Shields, G.A., Kamber, B.S., Lough, J.M., 2007. A multi-trace element coral record of land-use changes in the Burdekin River catchment, NE Australia. Palaeogeography, Palaeoclimatology, Palaeoecology 246, 471-487.

Lian, O.B., Roberts, R.G., 2006. Dating the Quaternary: progress in luminescence dating of sediments. Quaternary Science Reviews 25, 2449-2468.

Lough, J.M., 2007. Tropical river flow and rainfall reconstructions from coral luminescence: Great Barrier Reef, Australia. Paleooceanography 22, doi:10.1029/2006PA001377.

168 Chapter 6 - Summary and conclusions

Loughran, R.J., Campbell, B.L., 1995. The identification of catchment sediment sources. In: Foster, I.D.L., Gurnell, A.M., Webb, B.W. (Eds.), Sediment and Water Quality in River Catchments. John Wiley, New York, pp. 189-205.

Lu, H., Moran, C., Prosser, I.P., Raupach, M.R., Olley, J., Petheram, C., 2003a. Sheet and rill erosion and sediment delivery to streams: a basin wide estimation at hillslope to medium catchment scale. CSIRO Land and Water Technical Report 15/03, CSIRO Land and Water, Canberra.

Lu, H., Prosser, I.P., Moran, C.J., Gallant, J.C., Priestley, G., Stevenson, J.G., 2003b. Predicting sheetwash and rill erosion over the Australian continent. Australian Journal of Soil Research 41, 1037-1062.

Ludwig, W., Probst, J., 1996. A global modelling of the climatic, morphological, and lithological control of river sediment discharges to the oceans. Erosion and Sediment Yield, publ. no. 236. IAHS, Exeter.

Madsen, A.T., Murray, A.S., Anderson, T.J., Pejrup, M., Breuning-Madsen, H., 2005. Optically stimulated dating of young estuarine sediments: a comparison with 210Pb and 137Cs dating. Marine Geology 214, 251-268.

Makaske, B., Smith, D.G., Berendsen, H.J.A., 2002. Avulsions, channel evolution and floodplain sedimentation rates of the anastomosing upper Columbia River, British Columbia, Canada. Sedimentology 49, 1049-1071.

Mantua, N.J., Hare, S.R., 2002. The Pacific Decadal Oscillation. Journal of Oceanography 58(1), 35-44.

Matisoff, G., Bonniwell, E.C., Whiting, P.J., 2002. Soil erosion and sediment sources in an Ohio watershed using Beryllium-7, Cesium-137, and Lead-210. Journal of Environmental Quality 31(1), 54-61.

May, D.W., 2003. Properties of a 5500-year-old flood-plain in the Loup River basin, Nebraska. Geomorphology 56, 243-254.

169 References

McCulloch, M., Fallon, S., Wyndham, T., Hendy, E., Lough, J., Barnes, D., 2003. Coral record of increased sediment flux to the inner Great Barrier Reef since European settlement. Nature 421, 727-730.

McIvor, J.G., Williams, J., Gardener, C.J., 1995. Pasture management influences runoff and soil movement in the semi-arid tropics. Australian Journal of Experimental Agriculture 35, 55-65.

McKergow, L.A., Prosser, I.P., Hughes, A.O., Brodie, J., 2005. Sources of sediment to the Great Barrier Reef World Heritage Area. Marine Pollution Bulletin 51, 200- 211.

McLaughlin, C.J., Smith, C.A., Buddemeier, R.W., Bartley, J.D., Maxwell, B.A., 2003. Rivers, runoff and reefs. Global and Planetary Change 39(1-2), 191-199.

Merritt, W.S., Letcher, R.A., Jakeman, A.J., 2003. A review of erosion and sediment transport models. Environmental Modelling & Software 18, 761-799.

Middelkoop, H., 2002. Reconstructing floodplain sedimentation rates from heavy metal profiles by inverse modelling. Hydrological Processes 16, 47-64.

Montgomery, D.R., Foufoula-Georgiou, E., 1993. Channel network source representation using digital elevation models. Water Resources Research 29(12), 3925-3934.

Moss, A.J., Rayment, G.E., Reilly, N., Best, E.K., 1992. A preliminary assessment of sediment and nutrient exports from Queensland coastal catchments. Department of Environment and Heritage Technical Report No. 4, Department of Environment and Heritage, Brisbane.

Motha, J.A., Wallbrink, P.J., Hairsine, P.B., Grayson, R.B., 2003. Determining the sources of suspended sediment in a forested catchment in southeastern Australia. Water Resources Research 39(3), doi:10.1029/2001WR000794.

Murray, A.S., Marten, R., Johnston, A., Martin, P., 1987. Analysis for naturally occuring radionuclides at environmental concentrations by gamma spectrometry. Journal of Radioanalytical and Nuclear Chemistry 115(2), 263-288.

170 Chapter 6 - Summary and conclusions

Murray, A.S., Olley, J., 2002. Precision and accuracy in the optically stimulated luminescence dating of sedimentary quartz: a status review. Geochronometria 21, 1-16.

Nanson, G.C., Barbetti, M., Taylor, G., 1995. River stabilisation due to changing climate and vegetation during the late Quaternary in western Tasmania, Australia. Geomorphology 13, 145-158.

Nanson, G.C., Croke, J.C., 2002. Emerging issues in flood plain research. In: Dyer, F., Thoms, M., Olley, J.M. (Eds.), The Structure, Function and Management Implications of Fluvial Sedimentary Systems. IAHS, pp. 271-280.

Neil, D.T., Orpin, A.R., Ridd, P.V., Yu, B., 2002. Sediment yield and impacts from river catchments to the Great Barrier Reef lagoon. Marine and Freshwater Research 53, 733-752.

NLWRA, 2001. Australian Agricultural Assessment 2001, Volume 1, National Land and Water Resources Audit, Canberra.

Norrish, K., Hutton, J.T., 1969. An accurate x-ray spectrographic method for the analysis of a wide range of geological samples. Geochimica et Comochimica Acta 33, 31-453.

Nott, J., Price, D., Nanson, G.C., 2002. Stream response to Quaternary climate change: evidence from the Shoalhaven River catchment, southeastern highlands, temperate Australia. Quaternary Science Reviews 21, 965-974.

O'Reagain, P.J., Brodie, J., Fraser, G., Bushell, J.J., Holloway, C.H., Faithful, J.W., Haynes, D., 2005. Nutrient loss and water quality under extensive grazing in the upper Burdekin River catchment, North Queensland. Marine Pollution Bulletin 51, 37-50.

Oldfield, F., Clark, R.L., 1990. Lake sediment-based studies of soil erosion. In: Boardman, J., Foster, I.D.L., Dearing, J.A. (Eds.), Soil Erosion on Agricultural Land. John Wiley & Sons Ltd.

171 References

Olley, J., Caitcheon, G., 2000. Major element chemistry of sediments from the Darling- Barwon River and its tributaries: implications for sediment and phosphorous sources. Hydrological Processes 14, 1159-1175.

Olley, J., Pietsch, T., Roberts, R.G., 2004. Optical dating of Holocene sediments from a variety of geomorphic settings using single grains of quartz. Geomorphology 60, 337-358.

Olley, J.M., Murray, A.S., Mackenzie, D.M., Edwards, K., 1993. Identifying sediment sources in a gullied catchment using natural and anthropogenic radioactivity. Water Resources Research 29, 1037-1043.

Olley, J.M., Wasson, R.J., 2003. Changes in the flux of sediment in the Upper Murrumbidgee catchment, Southeastern Australia, since European settlement. Hydrological Processes 17, 3307-3320.

Ormerod, L.M., 1998. Estimating sedimentation rates and sources in a partially urbanised catchment using caesium-137. Hydrological Processes 12, 1009-1020.

Owens, P.N., Walling, D.E., 1996. Spatial variability of Caesium-137 inventories at reference sites: an example from two contrasting sites in England and Zimbabwe. Applied Radiation and Isotopes 47(7), 699-707.

Owens, P.N., Walling, D.E., He, Q., 1996. The behaviour of bomb-derived Caesium- 137 fallout in catchment soils. Journal of Environmental Radioactivity 32(3), 169-191.

Owens, P.N., Walling, D.E., Leeks, G.J.L., 1999. Use of floodplain sediment cores to investigate recent historical changes in overbank sedimentation rates and sediment sources in the catchment of the River Ouse, Yorkshire, UK. Catena 36, 21-47.

Owens, P.N., Walling, D.E., 2002. Changes in sediment sources and floodplain deposition rates in the catchment of the River Tweed, Scotland, over the last 100 years: the impact of climate and land use change. Earth Surface Processes and Landforms 27, 403-423.

172 Chapter 6 - Summary and conclusions

Page, K., Frazier, P., Pietsch, T., Dehaan, R., 2007. Channel change following European settlement: Gilmore Creek, southeastern Australia. Earth Surface Processes and Landforms 32, 1398-1411.

Page, K.J., Carden, Y.R., 1998. Channel adjustment following the crossing of a threshold: Tarcutta Creek, southeastern Australia. Australian Geographical Studies 36(3), 289-311.

Pandey, A., Chowdary, V.M., Mal, B.C., Billib, M., 2008. Runoff and sediment yield modeling from a small agricultural watershed in India using the WEPP model. Journal of Hydrology 348(3-4), 305-319.

Pasternack, G.B., Brush, G.S., Hilgartner, W.B., 2001. Impact of historic land-use change on sediment delivery to a Chesapeake Bay subestuarine delta. Earth Surface Processes and Landforms 26, 409-427.

Pavlovic, G., Barisic, D., Lovrencic, I., Orescanin, V., Prohic, E., 2005. Use of fallout 137Cs for documenting the chronology of overbank sediments from the river Sava, Croatia, and interpreting their geochemical patterns. Environmental Geology 47, 475-481.

Pfitzner, J., Brunskill, G., Zagorskis, I., 2004. 137Cs and excess 210Pb deposition patterns in estuarine and marine sediment in the central region of the Great Barrier Reef Lagoon, north-eastern Australia. Journal of Environmental Radioactivity 76, 81- 102.

Phillips, J., 1997. Human agency, Holocene sea level, and floodplain accretion in coastal plain rivers. Journal of Coastal Research 13(3), 854-866.

Phillips, J.D., 1991. Fluvial sediment budgets in the North Carolina Piedmont. Geomorphology 4, 231-241.

Pierson, T.C., 2007. Dating young geomorphic surfaces using age of colonizing Douglas fir in southwestern Washington and northwestern Oregon, USA. Earth Surface Processes and Landforms 32(6), 811-831.

173 References

Poesen, J., Nachtergaele, J., Verstraeten, G., Valentin, C., 2003. Gully erosion and environmental change: importance and research needs. Catena 50(2-4), 91-133.

Polyakov, V., Fares, A., Kubo, D., Jacobi, J., Smith, C., 2007. Evaluation of a non-point source pollution model, AnnAGNPS, in a tropical watershed. Environmental Modelling & Software 22(11), 1617-1627.

Power, S., Casey, T., Folland, C., Colman, A., Mehta, V., 1999. Inter-decadal modulation of the impact of ENSO on Australia. Climate Dynamics 15(5), 319- 324.

Prescott, J., Hutton, J., 1994. Cosmic-ray contributions to dose-rates for luminescence and ESR dating - large depths and long-term time variations. Radiation Measurements 23, 497-500.

Preston, S.D., Bierman, V.J., Silliman, S.E., 1989. An evaluation of methods for the estimation of tributary mass loads. Water Resources Management 25(6), 1379- 1389.

Prosser, I.P., 1990. Fire, humans and denudation at Wangrah Creek, southern Tablelands, N.S.W. Australian Geographical Studies 28(1), 77-95.

Prosser, I.P., Chappell, J., Gillespie, R., 1994. Holocene valley aggradation and gully erosion in headwater catchments, south-eastern highlands of Australia. Earth Surface Processes and Landforms 19, 465-480.

Prosser, I.P., Winchester, S.J., 1996. History and processes of gully initiation and development in Australia. Zeitschrift für Geomorphologie Supplement Band 105, 91-109.

Prosser, I.P., Rustomji, P., Young, W.J., Moran, C., Hughes, A.O., 2001a. Constructing river basin sediment budgets for the National Land and Water Resources Audit. CSIRO Land and Water Technical Report 15/01, CSIRO Land and Water, Canberra.

174 Chapter 6 - Summary and conclusions

Prosser, I.P., Rutherfurd, I.D., Olley, J.M., Young, W.J., Wallbrink, P.J., Moran, C.J., 2001b. Large-scale patterns of erosion and sediment transport in river networks, with examples from Australia. Marine and Freshwater Research 52, 81-99.

Purvis-Smith, D., 2007. Flood inundation and floodplain sedimentation at a valley constriction on Medway Creek, in the Fitzroy basin, Queensland. Unpublished Bachelor of Science (Hons) Thesis, University of New South Wales, Canberra.

Quine, T.A., 1999. Use of caesium-137 data for validation of spatially distributed erosion models: the implications of tillage erosion. Catena 37, 415-430.

Refsgaard, J.C., 1997. Parameterisation, calibration and validation of distributed hydrological models. Journal of Hydrology 198(1-4), 69-97.

Renard, K.G., G.A., F., D.K., W., McCool, D.K., Yoder, D.C., 1997. Predicting soil erosion by water: A guide to conservation planning with the revised universal soil loss equation, United States Department of Agriculture, Washington DC.

Restrepo, J.D., Zapata, P., Diaz, J.A., Garzon-Ferreira, J., Garcia, C.B., 2006. Fluvial fluxes into the Caribbean Sea and their impact on coastal ecosystems: The Magdalena River, Colombia. Global and Planetary Change 50(1-2), 33-49.

Ritchie, J.C., McHenry, J.R., 1990. Application of radiation fallout caesium-137 for measuring soil erosion and sediment accumulation rates and patterns: a review. Journal of Environmental Quality 19, 215-233.

Ritchie, J.C., Finney, V.L., Oster, K.J., Ritchie, C.A., 2004. Sediment deposition in the flood plain of Stemple Creek Watershed, northern California. Geomorphology 61, 347-360.

Rittenour, T.M., Goble, R.J., Blum, M.D., 2003. An optical age chronology of Late Pleistocene fluvial deposits in the northern lower Mississippi valley. Quaternary Science Reviews 22, 1105-1110.

Ritter, D.F., Kinsey, W.F., Kauffman, M.E., 1973. Overbank sedimentation in the Delaware River valley during the last 6000 years. Science 179(4071), 374-375.

175 References

Roache, P.J., 1982. Computational Fluid Dynamics. Hermosa Publishers, Albuquerque, NM, USA.

Rowland, J.C., Lepper, K., Dietrich, W.E., Wilson, C.J., Sheldon, R., 2005. Tie channel sedimentation rates, oxbow formation age and channel migration rate from optically stimulated luminescence (OSL) analysis of floodplain deposits. Earth Surface Processes and Landforms 30, 1161-1179.

Rumsby, B., 2000. Vertical accretion rates in fluvial systems: a comparison of volumetric and depth-based estimates. Earth Surface Processes and Landforms 25, 617-631.

Russell-Smith, R., Lucas, D., Gapindi, M., Gunbunuka, B., Kapirigi, N., Namingu, G., 1997. Aboriginal resource utilization and fire management practice in western Arnhem Land, monsoonal northern Australia: notes for prehistory, lessons for the future. Human Ecology 25(2), 159-195.

Russell, M.A., Walling, D.E., Hodgkinson, R.A., 2001. Suspended sediment sources in two small lowland agricultural catchments in the UK. Journal of Hydrology 252, 1-24.

Rustomji, P., 2006. Analysis of gully dimensions and sediment texture from southeast Australia for catchment sediment budgeting. Catena 67, 119-127.

Rustomji, P., Pietsch, T., Wilkinson, S., 2006. Pre and Post-European Settlement Patterns of Floodplain Deposition in the Lake Burragorang Catchment. CSIRO Land and Water Science Report 38/06, CSIRO Land and Water, Canberra, Australia.

Rustomji, P., Pietsch, T., 2007. Alluvial sedimentation rates from southeastern Australia indicate post-European settlement landscape recovery. Geomorphology 90, 73- 90.

Rustomji, P., Caitcheon, G., Hairsine, P., 2008. Combining a spatial model with geochemical tracers and river station data to construct a catchment sediment

176 Chapter 6 - Summary and conclusions

budget. Water Resources Research 44, W01422, do1:10.1029/2007WR006112, 2008.

Rutherfurd, I., 2000. Some human impacts on Australian stream channel morphology. In: Brizga, S., Finlayson, B. (Eds.), River Management: The Australasian Experience. John Wiley & Sons, Chichester, pp. 2-52.

Sarangi, A., Cox, C.A., Madramootoo, C.A., 2007. Evaluation of the AnnAGNPS model for prediction of runoff and sediment yields in St Lucia watersheds. Biosystems Engineering 97(2), 241-256.

Scanlan, J.C., Pressland, A.J., Myles, D.J., 1996. Run-off and soil movement on mid- slopes in north-east Queensland grazed woodlands. Rangeland Journal 18(1), 33-46.

Schaefli, B., Gupta, H.V., 2007. Do Nash values have value? Hydrological Processes 21, 2075-2080.

Schuller, P., Voigt, G., Handl, J., Ellies, A., Oliva, L., 2002. Global weapons' fallout 137Cs in soils and transfer to vegetation in south-central Chile. Journal of Environmental Radioactivity 62, 181-193.

Seidel, J., Mäckel, R., 2007. Holocene sediment budgets in two river catchments in the Southern Upper Rhine Valley, Germany. Geomorphology 92, 198-207.

Sheldon, F., Thoms, M.C., 2006. In-channel geomorphic complexity: The key to the dynamics of organic matter in large dryland rivers? Geomorphology 77, 270- 285.

Simons, M., Podger, R., Cooke, R., 1996. IQQM - A hydrologic modelling tool for water resource and salinity management. Environmental Software 11(1-3), 185- 192.

Spezzano, P., 2005. Distribution of pre- and post-Chernobyl radiocaesium with particle size fractions of soils. Journal of Environmental Radioactivity 83, 117-127.

177 References

Stefano, C.D., Ferro, V., Porto, P., Rizzo, S., 2005. Testing a spatially distributed delivery model (SEDD) in a forested basin by cesium-137 technique. Journal of Soil and Water Conservation 60(3), 148-157.

Sutherland, R.A., 1996. Caesium-137 soil sampling and inventory variability in reference locations: a literature survey. Hydrological Processes 10, 43-53.

Syvitski, J.P.M., Vörösmarty, C.J., Kettner, A.J., Green, P., 2005. Impact of humans on the flux of terrestrial sediment to the global coastal ocean. Science 308, 376-380.

Szerbin, P., Koblinger-Bokori, U.E., Koblinger, L., Végvári, I., Ugron, Á., 1999. Caesium-137 migration in Hungarian soils. The Science of the Total Environment 227(3-4), 215-227.

Takken, I., Beuselinck, L., Nachtergaele, J., Govers, G., Poesen, J., Degraer, G., 1999. Spatial evaluation of a physically-based distributed erosion model (LISEM). Catena 37, 431-447.

Taylor, G., Woodyer, K.D., 1978. Bank deposition in suspended-load streams. In: Miall, A.D. (Ed.), Fluvial Sedimentology. Canadian Society of Petroleum Geologists, Calgary, pp. 257-275.

Thoms, M.C., Parsons, M.E., Foster, J.M., 2007. The use of multivariate statistics to elucidate patterns of floodplain sedimentation at different spatial scales. Earth Surface Processes and Landforms 32, 672-686.

Trevithick, R., Herring, M., Dougall, C., Denham, R., Pringle, M., 2008. Gully density mapping and modelling for the Fitzroy Basin, Queensland, Australia, Queensland Department of Natural Resources and Water, Brisbane.

Trimble, S.W., 1983. A sediment budget for Coon Creek Basin in the Driftless area, Wisconsin, 1853-1977. American Journal of Science 283, 454-474.

Tyler, A.N., Carter, S., Davidson, D.A., Long, D.J., Tipping, R., 2001. The extent and significance of bioturbation on 137Cs distributions in upland soils. Catena 43, 81- 99.

178 Chapter 6 - Summary and conclusions

UNSCEAR, 2000. Sources and effects of ionizing radiation. Volume I: Sources, Annex C: Exposures from man-made sources of radiation, United Nations Committee on the Effects of Atomic Radiation, http://www.unscear.org/unscear/en/publications/2000_1.html.

Van Rompaey, A., Verstraeten, G., Van Oost, K., Govers, G., Poesen, J., 2001. Modelling mean annual sediment yield using a distributed approach. Earth Surface Processes and Landforms 26, 1221-1236. van Woesik, R., Tomascik, T., Blake, S., 1999. Coral assemblages and physico- chemical characteristics of the Whitsunday Islands: evidence of recent community changes. Marine and Freshwater Research 50, 427-440.

Vertessy, R.A., Watson, F.G.R., Rahman, J.M., Cuddy, S.D., Seaton, S.P., Chiew, F.H., Marston, F.M., Lymburner, L., Jeanelle, S., Verbunt, M., 2001. New software to aid water quality management in the catchments and waterways of the south- east Queensland region. In: Rutherfurd, I., Sheldon, F., Brierley, G., Kenyon, C. (Eds.), Third Australian Stream Management Conference, Brisbane Cooperative Research Centre for Catchment Hydrology, pp. 611-616.

Viney, N.R., Sivapalan, M., 1999. A conceptual model of sediment transport: application to the Avon River Basin in Western Australia. Hydrological Processes 13, 727-743.

Wallbrink, P.J., Murray, A.S., 1993. Use of fallout radionuclides as indicators of erosion processes. Hydrological Processes 7, 297-304.

Wallbrink, P.J., Murray, A.S., Olley, J.M., Olive, L.J., 1998. Determining sources and transit times of suspended sediment in the Murrumbidgee River, New South Wales, Australia, using fallout 137Cs and 210Pb. Water Resources Research 34, 879-887.

Wallbrink, P.J., 2004. Quantifying the erosion processes and land uses which dominate fine sediment supply to Moreton Bay, Southeast Queensland, Australia. Journal of Environmental Radioactivity 76, 67-80.

179 References

Walling, D.E., 1983. The sediment delivery problem. Journal of Hydrology 65, 209- 237.

Walling, D.E., 1988. Erosion and sediment yield research - some recent perspectives. Journal of Hydrology 100, 113-141.

Walling, D.E., He, Q., 1992. Interpretation of caesium-137 profiles in lacustrine and other sediments: the role of catchment-derived inputs. Hydrobiologia 235/236(1), 219-230.

Walling, D.E., He, Q., Nicholas, A.P., 1996. Floodplains as suspended sediment sinks. In: Anderson, M.G., Walling, D.E., Bates, P.D. (Eds.), Floodplain Processes. John Wiley & Sons Ltd., Chichester, pp. 399-440.

Walling, D.E., He, Q., 1997. Use of fallout 137Cs in investigations of overbank sediment deposition on river floodplains. Catena 29, 263-282.

Walling, D.E., 1999. Linking land use, erosion and sediment yields in river basins. Hydrobiologia 410, 223-240.

Walling, D.E., He, Q., 1999. Improved models for estimating soil erosion rates from Cesium-137 measurements. Journal of Environmental Quality 28(2), 611-622.

Walling, D.E., Fang, D., 2003. Recent trends in the suspended sediment loads of the world's rivers. Global and Planetary Change 39, 111-126.

Walling, D.E., He, Q., Whelan, P.A., 2003a. Using 137Cs measurements to validate the application of the application of the AGNPS and ANSWERS erosion and sediment yield models in two small Devon catchments. Soil and Tillage Research 69, 27-43.

Walling, D.E., Owens, P.N., Foster, I.D.L., Lees, J.A., 2003b. Changes in fine sediment dynamics of the Ouse and Tweed basins in the UK over the last 100-150 years. Hydrological Processes 17, 3245-3269.

Walling, D.E., 2005. Tracing suspended sediment sources in catchments and river systems. Science of the Total Environment 344, 159-184.

180 Chapter 6 - Summary and conclusions

Wallinga, J., 2002. Optically stimulated luminescence dating of fluvial deposits: a review. Boreas 31, 303-322.

Wasson, R.J., Mazari, R.K., Starr, B., Clifton, G., 1998. The recent history of erosion and sedimentation on the Southern Tablelands of south-eastern Australia: sediment flux dominated by channel incision. Geomorphology 24, 291-308.

Wasson, R.J., Caitcheon, G.G., Murray, A.S., McCulloch, M., Quade, J., 2002. Sourcing sediment using multiple tracers in the catchment of Lake Argyle, northwestern Australia. Environmental Management 29(5), 634-646.

Wicks, J.M., Bathurst, J.C., 1996. SHESED: a physically based, distributed erosion and sediment yield component for the SHE hydrological modelling system. Journal of Hydrology 175, 213-238.

Wilkinson, S., Henderson, A., Chen, Y., Sherman, B., 2004. SedNet User guide. CSIRO Land and Water Client Report, CSIRO Land and Water, Canberra.

Wilkinson, S.N., Prosser, I.P., Hughes, A.O., 2006. Predicting the distribution of bed material accumulation using river network sediment budgets. Water Resources Research 42, W10419, doi10.1029/2006WR004958.

Williams, G.P., 1989. Sediment concentration versus water discharge during single hydrologic events. Journal of Hydrology 111, 89-106.

Wintle, A.G., 1997. Luminescence dating: laboratory procedures and protocols. Radiation Measurements 27, 769- 817.

Xu, J., 2003. Sedimentation rates in the lower Yellow River over the past 2300 years as influenced by human activities and climate change. Hydrological Processes 17(16), 3359-3371.

Yang, S., Zhao, Q., Belkin, I.M., 2002. Temporal variation in the sediment load of the Yangtze river and the influences of human activities. Journal of Hydrology 263, 56-71.

181 References

Yeager, K.M., Santschi, P.H., Phillips, J.D., Herbert, B.E., 2005. Suspended sediment sources and tributary effects in the lower reaches of a coastal plain stream as indicated by radionuclides, Loco Bayou, Texas. Environmental Geology 47, 382-395.

Zhang, J.F., Zhou, L.P., Yue, S.Y., 2003. Dating fluvial sediments by optically stimulated luminescence: selection of equivalent doses for age calculation. Quaternary Science Reviews 22, 1123-1129.

Zierholz, C., Prosser, I.P., Fogarty, P.J., Rustomji, P., 2001. In-stream wetlands and their significance for channel filling and the catchment sediment budget, Jugiong Creek, New South Wales. Geomorphology 38(3-4), 221-235.

182 APPENDIX 1

Monte Carlo model mean difference data

Appendix 1 – Monte Carlo model mean difference data

Table A1.1. Mean difference between Monte Carlo model estimates and measured geochemical and radionuclide tracing data

Mean relative Mean relative Sample site/core difference (%) – difference (%) – depth geochemical tracing radionuclide tracing TC1 30.48 22.40 TC2 6.56 14.50 SC1 3.71 3.50 SC2 6.56 15.60 SC3 4.85 30.30

0-2cm 10.80 - 2-4cm 11.47 - 4-6cm 17.88 - 6-8cm 22.42 - 8-10cm 19.17 - 10-12cm 15.73 - 12-14cm 10.87 - 14-16cm 14.48 - 16-18cm 8.98 - 18-20cm 17.36 - 20-22cm 15.56 - 22-24cm 21.51 - 24-26cm 25.32 - 26-28cm 28.83 - 28-30cm 27.52 - 30-32cm 19.01 - 32-34cm 25.35 - 34-36cm 16.37 - 36-38cm 22.44 - 38-40cm 16.41 - 40-42cm 18.54 - 42-44cm 25.42 - 44-46cm 21.97 - 46-48cm 20.93 - 48-50cm 20.20 - 50-52cm 15.93 - 52-54cm 18.01 - 54-56cm 15.12 - 56-58cm 17.07 - 58-60cm 11.89 -

184 APPENDIX 2

Sediment source geochemistry data

Appendix 2 – Sediment source geochemistry data

Table A2.1. Basic descriptive statistics for the sediment source geochemistry data

Major element concentrations (%) Element Granite 1σ Metasediment 1σ Basalt 1σ Clastic 1σ (n =10) (n = 10) (n =11) (n =10) MgO 2.35 0.23 2.92 0.72 4.75 0.56 1.37 0.55

K2O 1.87 0.19 3.86 1.35 0.70 0.16 2.09 0.96

P2O5 0.15 0.04 0.16 0.05 0.22 0.07 0.25 0.14

TiO2 1.31 0.16 1.13 0.24 1.98 0.54 1.43 0.44 Minor element concentrations (ppm) Element Granite 1σ Metasediment 1σ Basalt 1σ Clastic 1σ (n =10) (n = 10) (n =11) (n =10)

CeO2 194.98 28.85 171.77 60.49 76.66 19.39 171.44 38.14 NiO 60.97 23.23 152.04 45.05 247.38 33.28 129.14 31.58 SrO 258.78 146.01 87.24 24.44 271.82 51.84 137.77 35.13

V2O5 462.69 69.92 447.81 94.76 326.36 40.86 332.10 37.10

ZrO2 102.12 25.21 249.53 42.85 211.78 37.72 305.31 65.50

186 APPENDIX 3

Summary of over-bankfull discharge data for gauging station 130210A

Appendix 3 – Over-bankfull discharge summary data

Table A3.1. Summary of over-bankfull discharge data for gauging station 130210A.

Event start date Event end date Duration above 660 Peak discharge m3 sec-1 (m3 sec-1) 01:05_21/12/1973 17:16_21/12/1973 16.2 hours 850 00:12_22/01/1976 07:46_22/01/1976 7.6 hours 701 17:21_01/02/1978 05:22_04/02/1978 2.5 days 1174 15:51_21/05/1983 14:01_22/05/1983 22.2 hours 701 21:24_08/03/1994 15:05_09/03/1994 17.7 hours 807 18:12_04/01/1999 08:34_05/01/1999 14.4 hours 721

188 APPENDIX 4

137Cs data for floodplain and reference site cores

Appendix 4 – 137Cs data for floodplain and reference core sites

Table A4.1. Caesium-137 data for floodplain and reference core sites

Core Depth Mass(g) Count time Cs-137 2 σ (cm) (secs) (Bq kg-1) CC-F 0-2 53.64 162652 0.70 0.80 CC-F 2-4 61.19 262800 0.46 0.45 CC-F 4-6 49.2 175004 0.81 0.81 CC-F 6-8 87.25 173088 0.56 0.65 CC-F 8-10 83.12 165351 0.90 0.64 CC-F 10-12 74.38 262800 2.63 0.55 CC-F 12-14 73.15 149123 5.20 0.78 CC-F 14-16 69.46 174972 3.36 0.68 CC-F 16-18 96.13 164195 1.19 0.66 CC-F 18-20 88.12 262800 0.39 0.51 LTC-F 0-2 57.4 170742 1.27 0.82 LTC-F 2-4 81.81 176199 2.02 0.69 LTC-F 4-6 81.06 178055 3.04 0.71 LTC-F 6-8 92.93 179171 4.54 0.77 LTC-F 8-10 128.82 173842 3.19 0.73 LTC-F 10-12 95.58 173842 1.52 0.73 LTC-F 12-14 90.14 175673 0.93 0.70 LTC-F 14-16 90.37 170920 0.81 0.73 LTC-F 16-18 65.33 171475 0.71 0.74 LTC-F 18-20 79.53 178881 0.72 0.73 UTC-F 0-2 75.19 168050 1.74 0.79 UTC-F 2-4 90.06 167201 2.03 0.78 UTC-F 4-6 94.75 167563 3.67 0.81 UTC-F 6-8 93.11 170214 2.35 0.79 UTC-F 8-10 79.24 262800 0.66 0.65 UTC-F 10-12 58.17 166821 0.68 0.88 UTC-F 12-14 51.47 171627 0.25 0.93 UTC-F 14-16 74.62 82800 0.57 1.14 UTC-F 16-18 88.31 82800 0.05 1.07 UTC-F 18-20 81.88 82800 0.55 1.12 REF1 0-2 40.04 82800 6.31 1.47 REF1 2-4 52.33 82800 7.94 1.29 REF1 4-6 49.57 82800 6.28 1.28 REF1 6-8 65.77 82800 2.41 0.99 REF1 8-10 62.7 82800 1.07 1.00 REF2 0-2 64.69 164539 8.21 0.85 REF2 2-4 62.76 164700 7.33 0.85 REF2 4-6 135.87 82800 0.63 0.97 REF2 6-8 99.28 82800 0.43 0.97 REF2 8-10 74.24 82800 0.97 0.97 REF3 0-2 52.3 172800 7.14 1.00 REF3 2-4 96.86 172800 4.61 0.92 REF3 4-6 79.3 172800 1.62 0.84 REF3 6-8 94.84 172800 0.79 0.82 REF3 8-10 86.34 172800 0.80 0.83

190 APPENDIX 5

Channel cross-sections

Appendix 5 – Channel cross-sections

352

350

348 UTC-F

346

344 UTC-B 342 Height above msl (m) 340

338 170 190 210 230 250 270 290 310 330 Distance (m)

Figure A5.1. Channel cross-section at the upper Theresa Creek site (UTC) showing the sediment core positions

267 266 265 LTC-F 264 263 262 261 LTC-B 260 259 Height above msl (m) 258 257 256 160 180 200 220 240 260 280 Distance (m)

Figure A5.2. Channel cross-section at the lower Theresa Creek site (LTC) showing the sediment core positions

192 Appendix 5 – Channel cross-sections

295

294

293 CC-F

292

291

290 Height above msl(m) above Height 289

288 0 50 100 150 200 Distance (m)

Figure A5.3. Channel cross-section at the Capella Creek site (CC) showing the sediment core positions

193

APPENDIX 6

Radial plots for single-grain OSL data Appendix 6 – OSL radial plots

Figure A6.1. Radial plot of single-grain De estimates (Gy) for sample LTC-B-38

Figure A6.2. Radial plot of single-grain De estimates (Gy) for sample LTC-B-95

196

Figure A6.3. Radial plot of single-grain De estimates (Gy) for sample LTC-F-22

Figure A6.4. Radial plot of single-grain De estimates (Gy) for sample LTC-F-35

197 Appendix 6 – OSL radial plots

Figure A6.5. Radial plot of single-grain De estimates (Gy) for sample LTC-F-50

Figure A6.6. Radial plot of single-grain De estimates (Gy) for sample LTC-F-95

198

Figure A6.7. Radial plot of single-grain De estimates (Gy) for sample UTC-B-40

Figure A6.8. Radial plot of single-grain De estimates (Gy) for sample UTC-B-90

199 Appendix 6 – OSL radial plots

Figure A6.9. Radial plot of single-grain De estimates (Gy) for sample CC-F-20

Figure A6.10. Radial plot of single-grain De estimates (Gy) for sample CC-F-40

200

Figure A6.11. Radial plot of single-grain De estimates (Gy) for sample CC-F-60

Figure A6.12. Radial plot of single-grain De estimates (Gy) for sample UTC-F-15

201 Appendix 6 – OSL radial plots

Figure A6.13. Radial plot of single-grain De estimates (Gy) for sample UTC-F-60

Figure A6.14. Radial plot of single-grain De estimates (Gy) for sample UTC-F-100

202 APPENDIX 7

Site data for OSL samples

Appendix 6 – OSL radial plots

Table A7.1 Site data for OSL samples

Sample ID Depth Latitude Longitude Altitude Density Water Cosmic dose rate Total Terrestrial (m) (degrees) (degrees) (m) (kg m-3) Content (%) (Gy ka-1) Dose rate (Gy ka-1) LTC-B-38 0.38 -23.23 147.99 260 1.8 7.5 ± 5 0.2059 2.7491 LTC-B-95 0.95 -23.23 147.99 260 1.8 7.5 ± 5 0.1917 2.8165

LTC-F-22 0.22 -23.23 147.99 264 1.8 7.5 ± 5 0.2101 2.8439 LTC-F-35 0.5 -23.23 147.99 264 1.8 7.5 ± 5 0.2028 3.1172 LTC-F-50 0.5 -23.23 147.99 264 1.8 7.5 ± 5 0.2028 3.5024 LTC-F-95 0.95 -23.23 147.99 264 1.8 7.5 ± 5 0.1917 3.3528

UTC-B-40 0.4 -23.06 147.48 342 1.8 7.5 ± 5 0.2081 3.4467 UTC-B-90 0.9 -23.06 147.48 342 1.8 7.5 ± 5 0.1954 2.6549

UTC-F-15 0.6 -23.06 147.48 347 1.8 7.5 ± 5 0.2031 4.0004 UTC-F-60 0.6 -23.06 147.48 347 1.8 7.5 ± 5 0.2031 3.5782 UTC-F-100 1 -23.06 147.48 347 1.8 7.5 ± 5 0.1932 3.5202

CC-F-20 0.2 -23.01 148.04 292 1.8 7.5 ± 5 0.2114 0.6025 CC-F-40 0.4 -23.01 148.04 292 1.8 7.5 ± 5 0.2061 0.5435 CC-F-60 0.6 -23.01 148.04 292 1.8 7.5 ± 5 0.2010 0.5382

204