A part of BMT in Energy and Environment
Final Report: Development of The Murraylands E2/WaterCast Catchment Model
R.B16618.001.02.doc May 2009
Final Report: Development of The Murraylands E2/WaterCast Catchment Model
Offices
Brisbane Denver Karratha Melbourne Prepared For: South Australian Environment Protection Authority Morwell Newcastle Perth Prepared By: BMT WBM Pty Ltd (Member of the BMT group of companies) Sydney Vancouver
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC
DOCUMENT CONTROL SHEET
BMT WBM Pty Ltd BMT WBM Pty Ltd Level 11, 490 Upper Edward Street Document : R.B16618.001.02.doc Brisbane 4000 Queensland Australia PO Box 203 Spring Hill 4004 Project Manager : Tony Weber
Tel: +61 7 3831 6744 Fax: + 61 7 3832 3627
ABN 54 010 830 421 002 Client : South Australian EPA www.wbmpl.com.au
Client Contact: Luke Mosley
Client Reference
Title : Final Report: Development Of The Murraylands E2/WaterCast Model Author : Dr Joel Stewart Synopsis : This report outlines the development of an E2/WaterCast/WaterCast catchment model to describe the rainfall-runoff and pollutant export dynamics of the South Australian Murray Darling Basin catchment and investigates land management options.
REVISION/CHECKING HISTORY
REVISION DATE OF ISSUE CHECKED BY ISSUED BY NUMBER 0 24/10/07 TRW JPS 1 05/12/08 TRW JPS 2 28/05/09 TRW JPS
DISTRIBUTION
DESTINATION REVISION 0 1 2 3
South Australia EPA pdf Pdf Pdf BMT WBM File pdf Pdf Pdf BMT WBM Library pdf pdf Pdf
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC CONTENTS I
CONTENTS
Contents i List of Figures ii List of Tables iv
EXECUTIVE SUMMARY V
1 INTRODUCTION 1-1
2 CATCHMENT CHARACTERISTICS 2-1
2.1 Land Use and Catchment Management 2-3 2.2 Hydrology 2-4 2.3 Water Quality 2-6
3 METHODOLOGY 3-1
3.1 Catchment Model Background 3-1
4 DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-1
4.1 Overview 4-1 4.2 Step 1 - Spatial Representation of the Catchment 4-1 4.2.1 Creating a Pit Filled DEM 4-1 4.2.2 Developing the E2/WaterCast Subcatchment Map. 4-2 4.2.3 Modifying the Subcatchment Map. 4-2 4.3 Step 2 Creation of the Node-Link Network 4-5 4.4 Step 3 – Functional Unit Definition 4-5 4.4.1 Land Use Map – Creating a Raster for E2/WaterCast. 4-5 4.4.2 Redefining Land Uses for the Murraylands E2/WaterCast Model 4-7 4.5 Step 4 – Node Model selection 4-8 4.6 Step 5 – Link Model Selection 4-8 4.7 Step 6 – Climatic Data Inputs 4-9 4.8 Step 7 – Parameterisation and Calibration 4-9 4.8.1 Flow Gauging Data 4-9
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC LIST OF FIGURES II
4.8.2 Water Quality Data 4-10 4.8.3 Constituents and Generation 4-13 4.8.4 Hydrological Parameterisation 4-14 4.9 Accounting for Instream Losses 4-20 4.9.1 Identification of Overlapping Gauge Records 4-20 4.9.2 Gauged Stream Losses 4-21 4.9.3 Conceptual Representation of In-stream Losses 4-25 4.9.4 Loss Model Implementation 4-26
5 RESULTS AND DISCUSSION 5-1
5.1 Existing Scenario 5-1 5.1.1 Spatial Representation 5-1 5.1.1.1 Annual Rainfall 5-1 5.1.1.2 Flows 5-3 5.1.1.3 Total Suspended Solids 5-6 5.1.1.4 Total Nitrogen 5-7 5.1.1.5 Total Phosphorus 5-8 5.1.2 Temporal Representation 5-9 5.1.2.1 Pollutant Sources (land use loads) 5-12 5.1.2.2 Water Quality Objectives 5-13 5.1.3 Instream Losses 5-15 5.2 Predevelopment Scenario 5-15 5.2.1 Change Scenarios 5-17 5.3 Discussion 5-17 5.4 Summary 5-20
6 REFERENCES 6-1
APPENDIX A: DATA SUPPLIED A-1
APPENDIX B: UNGROUPED LAND USE CLASSES B-1
LIST OF FIGURES
Figure 2-1 Extent of Murraylands E2/WaterCast Model 2-1 Figure 2-2 Lower Murray River and Adjacent Wetlands 2-2 Figure 2-3 The Shores of Lake Alexandrina 2-2
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC LIST OF FIGURES III
Figure 2-4 Finniss River Land Use and Water Monitoring Stations 2-3 Figure 2-5 Grazing Lands of the Eastern Mt Lofty Ranges. 2-4 Figure 2-6 Example Catchment Management Techniques 2-4 Figure 2-7 Angas Weir (Photo SA Department of Water Land and Biodiversity Conservation) 2-5 Figure 2-8 Example Farm Dams 2-5 Figure 2-9 Example Ambient Water Quality Data 2-6 Figure 3-1 Step 1 A Spatial Description of the Catchment is Developed 3-1 Figure 3-2 Step 2 – A Node-Link Network is Constructed 3-2 Figure 3-3 Step 3 Functional Units (land uses) are Defined 3-2 Figure 3-4 Step 4 Node Models are Selected 3-2 Figure 3-5 Step 5 – Selection of Link Models 3-3 Figure 3-6 Combined Node and Link Models Describing the Catchment 3-3 Figure 3-7 Step 6 Climatic Data Inputs Data 3-3 Figure 3-8 Step 7 Parameterisation and Calibration (Gauge AW425530) 3-4 Figure 3-9 Final Murraylands E2/WaterCast Model 3-4 Figure 4-1 3D View of pitfilled DEM 4-2 Figure 4-2 Murraylands Subcatchment Map 4-4 Figure 4-3 Murraylands Model Node-Link Network 4-5 Figure 4-4 Murraylands E2/WaterCast Model Land Use Map 4-6 Figure 4-5 Murraylands E2/WaterCast Calibration Catchments 4-15 Figure 4-6 RRL Calibration and Verification For Gauge AW426403 4-17 Figure 4-7 RRL Calibration of Daily Flows (Small Events) For Gauge AW426403 4-17 Figure 4-8 RRL Calibration of Daily Flows (Large Events) For Gauge AW426403 4-18 Figure 4-9 Hydrological Parameterisation of Murraylands E2/WaterCast Model 4-19 Figure 4-10 Stream Gauge Sites With Overlapping Record 4-21 Figure 4-11 Lower Angas Daily Flow Record 4-21 Figure 4-12 Lower Bremer Daily Flow Record 4-22 Figure 4-13 Cumulative Flow, Angas River 4-23 Figure 4-14 Cumulative Flow, Bremer River 4-23 Figure 4-15 Winter 2006 Angas River Flow 4-24 Figure 4-16 Winter 2006 Bremer River Flow 4-24 Figure 4-17 Angas River instream loss model calibration/verification 4-26 Figure 4-18 Bremer River instream loss model calibration/verification 4-27 Figure 5-1 Mean Annual Rainfall, 1957-2006 5-2 Figure 5-2 Mean Annual Areal Flows 5-3 Figure 5-3 Total Suspended Solids Areal Load Contributions 5-6 Figure 5-4 Total Nitrogen Areal Load Contributions 5-7 Figure 5-5 Total Phosphorus Areal Load Contributions 5-8 Figure 5-6 Predicted Annual Flows at Model Outlet 5-9
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC LIST OF TABLES IV
Figure 5-7 Predicted Annual Total Suspended Solids Loads at Model outlet 5-10 Figure 5-8 Predicted Annual Total Nitrogen Loads at Model Outlet 5-10 Figure 5-9 Predicted Mean Annual Total Phosphorus Loads at Model Outlet 5-11 Figure 5-10 Predicted vs Measured TN for the Finniss River GS 426504 5-11 Figure 5-11 Relative Land Use Loads 5-13 Figure 5-12 Modelled TN Concentration 5-14 Figure 5-13 Modelled TP Concentration 5-14 Figure 5-14 Existing vs Predevelopment TSS Loads 5-16 Figure 5-15 Existing vs Predevelopment TN Loads 5-16 Figure 5-16 Existing vs Predevelopment TP Loads 5-17 Figure 5-17 Nutrient Generation Rates (Extracted From Drewry et al 2006) 5-18 Figure 5-18 Farm Dams in Series, Mt Lofty Ranges, August 2007. 5-19
LIST OF TABLES
Table 4-1 Functional Units of the Murraylands E2/WaterCast Model 4-8 Table 4-2 Flow Gauging Sites Used for Model Calibration 4-10 Table 4-3 Bird In Hand Wastewater Treatment Plant Characteristics 4-11 Table 4-4 Mt Barker Wastewater Treatment Plant Characteristics (irrigation data) 4-12 Table 4-5 Monthly STP Discharge 4-12 Table 4-6 Adopted E2/WaterCast STP Water Quality Characteristics 4-12 Table 4-7 EMC Parameterisation 4-13 Table 4-8 DWC Parameterisation 4-14 Table 4-9 Subcatchment Calibration Summary 4-16 Table 4-10 Subcatchment Calibrated Hydrological Parameters 4-16 Table 4-11 Adopted Hydrological Parameters 4-18 Table 4-12 Gauge Records for Angas and Bremer Rivers 4-20 Table 5-1 Comparison Between CSIRO and E2/WaterCast Calibration Performance 5-4 Table 5-2 Comparison Between CSIRO and E2/WaterCast Predicted Flows 5-4 Table 5-3 Relative Land Use Pollutant Loads 5-12 Table 5-4 Ecosystem Protection Guidelines (ANZECC) 5-14 Table 5-5 Constituent Generation Rates 5-18
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC EXECUTIVE SUMMARY V
EXECUTIVE SUMMARY
The Murraylands E2/WaterCast catchment model extends form the South Australian - Victoria border to the outlet of the Murray River and encompasses approximately 68,000 km2. The landscape covered by the model ranges from arid lands in the north to low rainfall irrigation lands along the Murray River and moderate rainfall grazing and cropping lands in the Eastern Mount Lofty Ranges. The dominant land use in the model is grazing lands, which in turn dominated the total pollutant loads.
The model boundary and subcatchments were generated using both automated pit filled DEM data and hand drawn subcatchments. A total of 122 subcatchments ranging in size from 0.02-13,175km2 were delineated to represent the Murraylands catchment and were connected via a node link network and routed to the catchment outlet.
Due to computational limitations, only 36 years of daily climate data has been incorporated into the model. Each subcatchment has individual rainfall and potential evaporation data. This data is applicable for all land uses within each subcatchment. Smaller subcatchments were built into the model to account for high rainfall gradient zones.
Up to 10 subcatchments, predominantly in the south eastern Mt Lofty Ranges, have been calibrated using daily flow records and gridded SILO rainfall data. The length of calibration/verification record varies for each subcatchment and the calibration was generally performed using the Nash-Sutcliffe criterion on monthly flows followed by manual calibration of the daily hydrograph. Nash-Sutcliffe values ranged from 0.5-0.95 demonstrating good, but variable model calibration. The calibration of several subcatchments provided rainfall-runoff parameter sets that were then adopted for all other un- calibrated areas of the model.
An existing case and predevelopment case model were developed and the results analysed. These models showed that the highest pollutant generation rates are associated with the Eastern Mt Lofty catchments (high rainfall zones) and that these pollutant generation rates are within the accepted range in the literature. Furthermore the model demonstrated seasonality in pollutant concentrations as seen by monitored data (Finniss River) although pollutant concentration peaks for TN may yet be under predicted.
This project thus demonstrates that the Murraylands E2/WaterCast model has the potential to be used as a predictive tool to test further management scenarios. There remain limitations to the model in terms of describing impacts of farm dams and variations in pollutant event mean concentrations, however it appears to provide robust predictions over long time frames and compares well to past catchment pollutant export studies.
Further refinements of subcatchment flow and water quality calibrations are recommended for selected streams in the Eastern Mt Lofty ranges and associated lowlands, particularly with respect to event loads, surface water-groundwater interactions and in stream losses. In addition to this, it is recommended that further scenario development be undertaken in response to specific catchment management approaches and appropriate stakeholder engagement.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC INTRODUCTION 1-1
1 INTRODUCTION
This report details the development of an E2/WaterCast catchment flow and water quality model for the Murraylands Region of South Australia. The model extends over an area of approximately 68,000 km2 and incorporates the lands from the South Australian – Victorian Border in the east to Lake Alexandrina and the eastern Mt Lofty catchments in the south-west.
The model has been constructed using existing land use data and topographic/catchment information for the delineation of catchments. Where possible, flow gauge records were used to calibrate rainfall/runoff relationships for up to 10 subcatchments. Up to 36 years of climate data have been incorporated into the model allowing the simulation of a wide range of events that deliver flows and associated pollutant loads to the lower Murray River and potentially to Lake Alexandrina.
Section 2 of this report outlines the catchment encompassed by the Murraylands E2/WaterCast model, typical land use, hydrology and water quality for the region and Section 5 contains the results of the initial model simulations and scenario comparisons.
Two scenarios were created using the Murraylands E2/WaterCast model including the existing case model, and predevelopment model. The existing case model was created based on current land use characteristics and has been calibrated to relatively recent hydrological records, thus it represents the pollutant loads and flows that may be expected from the region today. The predevelopment model assumes the same catchment and hydrological behaviour as the existing case model but differs in the land use classifications by reverting all land uses to ‘native vegetation’ thereby changing pollutant export relationships.
The model results are generally in agreement with literature values in terms of nutrient generation rates and in stream water quality, indicating that further scenarios can now be developed in consultation with stakeholders to test land management actions.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC CATCHMENT CHARACTERISTICS 2-1
2 CATCHMENT CHARACTERISTICS
The area considered for the Murraylands E2/WaterCast catchment model encompasses all of the catchments contributing to the Murray River from the border of South Australia and Victoria/New South Wales. Figure 2-1 shows the extent of the Murraylands E2/WaterCast model.
Figure 2-1 Extent of Murraylands E2/WaterCast Model
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC CATCHMENT CHARACTERISTICS 2-2
The Murraylands E2/WaterCast model encompasses a wide range of landscapes from the dry arid northern regions to irrigation districts along the river flats and relatively high rainfall Eastern Mt Loft Ranges. The key area of interest for the Murraylands E2/WaterCast model is the Eastern Mt Lofty Ranges and catchments contributing to lower Murray River (Figure 2-2) and Lake Alexandrina (Figure 2-3).
Figure 2-2 Lower Murray River and Adjacent Wetlands
Figure 2-3 The Shores of Lake Alexandrina
The Eastern Mt Lofty catchments receive the highest annual rainfall of approximately 800-900mm/yr, substantially higher than the majority of other areas covered by the other areas of the model, which generally receive less than 400mm/yr. This area contains the major tributaries entering into the River Murray and Lower Lakes (e.g. Marne, Angas, Bremer, and Finniss rivers; Tookayerta and Currency creeks).
A steep rainfall gradient is evident from the Eastern Mt Lofty Ranges to the surrounding lowlands indicating that the majority of runoff and therefore pollutant generation and transport to the Murray River and Lakes is likely to originate from these catchments. Subsequently, higher attention to model resolution was required in this region including:
• Smaller and more detailed subcatchment delineation;
• Consideration for catchment lands uses and management techniques;
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC CATCHMENT CHARACTERISTICS 2-3
• Hydrological calibration where possible; and
• Comparison of modelled water quality and pollutant loads with literature values.
2.1 Land Use and Catchment Management
The total area covered by the model is approximately 68,000 km2 and this is made up predominantly by grazing and cropping land uses, although a small strip of land either side of the Murray River supports a range of horticultural and irrigation based activities. Detailed Land use maps, such as that shown in Figure 2-4 are available for a number of subcatchments in the Eastern Mt Lofty Ranges, as is a GIS layer for the entire catchment (see website www.epa.sa.gov.au),. These land use maps have been provided for this study and provide a basis for the generation of pollutant exports
Figure 2-4 Finniss River Land Use and Water Monitoring Stations
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC CATCHMENT CHARACTERISTICS 2-4
Figure 2-4 provides an example of grazing land use for the Eastern Mt Lofty Ranges near Mt Pleasant. The figure clearly shows the incised drainage lines with the potential to scour and transport suspended sediment during runoff events.
Figure 2-5 Grazing Lands of the Eastern Mt Lofty Ranges.
Observed catchment management techniques include stream bank gully erosion control through riparian plantings as shown in Figure 2-6.
Figure 2-6 Example Catchment Management Techniques
2.2 Hydrology
A number of stream gauges record continuous flows from the relatively well watered catchments of the Eastern Mt Lofty Ranges, however in the low rainfall areas to the east and north of the Murray River, very few gauging sites are available, perhaps reflecting the lack of major river systems and intermittent nature of the majority of drainage lines in this area. Figure 2-7 shows the Angas Weir flow monitoring station as an example of a flow gauging station in the Eastern Mt Lofty Ranges. Gauges such as these provide useful model calibration data.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC CATCHMENT CHARACTERISTICS 2-5
Figure 2-7 Angas Weir (Photo SA Department of Water Land and Biodiversity Conservation)
Stream network maps provided for this study do not show permanent streams to the east and north of the Murray River and the very flat nature of the terrain in these regions also supports the notion that runoff from these areas is only intermittent.
In the higher rainfall zones, numerous farm dams intercept the drainage lines as shown in Figure 2-8 and are likely to have a significant impact on both hydrological response and pollutant export. Data has been provided to describe farm dam density for the Eastern Mt Lofty Ranges however no investigation have yet been undertaken to correlate farm dam density to catchment runoff.
Figure 2-8 Example Farm Dams
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC CATCHMENT CHARACTERISTICS 2-6
The Murraylands region receives the majority of rainfall during the winter months, thus flows to Lake Alexandrina and the Lower Murray River are also seasonal. During winter periods increases in base flows are observed in gauge records, however event based flows constitute the majority of runoff (and hence pollutant load) from the catchment.
2.3 Water Quality
A number of sites in the Eastern Mt Lofty Ranges have extended records describing ambient water quality through a range of physicochemical parameters. Examples of this data produced by the South Australian EPA are shown in Figure 2-9 for the Bremer River for Total Nitrogen (TN) in mg/L and the associated ANZECC water quality guideline value. As the data shows, water quality guideline values are often exceeded, potentially indicating an impacted waterway.
Figure 2-9 Example Ambient Water Quality Data
Ambient water quality data has been provided for this study for up to 12 sites mostly in the Eastern Mt Lofty Ranges and corresponding lowlands.
Two sewage treatment plants (STPs), Bird in Hand STP and Mt Barker STP have been identified for the Murraylands E2/WaterCast model region, with the remainder of treatment plants understood to practice land application. Recent data characterising the flows from these STPs has also been provided for this investigation.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC METHODOLOGY 3-1
3 METHODOLOGY
3.1 Catchment Model Background
The use of catchment decision support tools has been facilitated greatly through the availability of modelling tools provided by the former CRC for Catchment Hydrology and the current eWater CRC through the Catchment Modelling Toolkit (see www.toolkit.net.au). The tools available on the toolkit website allow a catchment modeller to define catchments, calibrate hydrology and develop simulations of catchment responses.
The E2/WaterCast modelling framework provides the ability to simulate current catchment characteristics and responses, in addition to evaluating impacts of land use change and the implementation of best management practices. The E2/WaterCast framework is not one model, but a framework in which groups of different models can be selected and linked such that the most suitable model to describe a particular aspect of the catchment can be used.
To construct a catchment model within E2/WaterCast requires the user to define which model components are required and how they should be linked together. The underlying data within the model is some spatial description of the catchment, whether simply a subcatchment map, or a digital elevation model. These are then joined together via a node-link network, which is then parameterised and calibrated to complete the catchment model. The steps describing this model construction process are outlined below:
Step 1 – The catchment and streams are described spatially using either a digital elevation model or from topographical data (Figure 3-1)
Figure 3-1 Step 1 A Spatial Description of the Catchment is Developed
Step 2 – A node-link network is built either automatically from the digital elevation model, or manually from the data obtained in Step 1 (Figure 3-2).
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC METHODOLOGY 3-2
Figure 3-2 Step 2 – A Node-Link Network is Constructed
Step 3 – Information about each subcatchment is described and within this step, land use data is used to describe the “Functional Units” (FUs) within each subcatchment where different FUs have particular runoff and constituent generation characteristics (Figure 3-3). These are typically a common set for the same functional unit within the entire catchment, though the areal extent differs within each subcatchment. When the functional units are defined, constituents are then selected that will be common across all subcatchments and functional units.
Figure 3-3 Step 3 Functional Units (land uses) are Defined
Step 4 – Particular models are selected which are best suited to the subcatchment/node and these then describe (through different parameters) how each functional unit responds to climatic inputs (Figure 3-4).
Figure 3-4 Step 4 Node Models are Selected
Step 5 – Each link in the stream network is defined using an appropriate model in a similar way to the subcatchments in Step 4 (Figure 3-5).
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC METHODOLOGY 3-3
Figure 3-5 Step 5 – Selection of Link Models
These link models, combined with the models describing the subcatchments/nodes so that groups of models are linked together to describe the catchment as shown below (Figure 3-6).
Figure 3-6 Combined Node and Link Models Describing the Catchment
Step 6 – Climatic Data is selected (Figure 3-7). This can be either from individual stations, or interpolated gridded data (e.g. SILO, PET Atlas). The E2/WaterCast framework then interrogates this data for each model run performed.
Figure 3-7 Step 6 Climatic Data Inputs Data
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC METHODOLOGY 3-4
Step 7 – All models are parameterised and calibrated (Figure 3-8). This is usually accomplished through comparison with some observed data, such as flow gauging stations and storm event water quality
Figure 3-8 Step 7 Parameterisation and Calibration (Gauge AW425530)
Once the model has been appropriately parameterised, and this checked through calibration, it is ready for use. In most cases, the model is set up to represent the existing case. An example of the final model is shown in Figure 3-9.
Figure 3-9 Final Murraylands E2/WaterCast Model
Results of various scenarios can then be extracted for all constituents used in the model and displayed on screen, or exported to other programs such as Excel for compilation or reprocessing.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-1
4 DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL
4.1 Overview
The Murraylands E2/WaterCast model was developed according to the steps set out in the previous section, however several variations in each step were required in order to develop the final model. This is typical of most catchment model developments in that the optimal model is developed through several iterations where the most appropriate combinations of node and link models are selected which best describe the catchment processes and responses.
4.2 Step 1 - Spatial Representation of the Catchment
Initial data sets were obtained for the Murraylands E2/WaterCast Catchment from the SA EPA. These data sets included the files described in Appendix A. Important data sets include the Digital Elevation Model (DEM), subcatchment map (hand drawn), land use raster, flow gauging data and water quality monitoring data.
4.2.1 Creating a Pit Filled DEM
The DEM forms a basis from which to delineate subcatchments and streams according to topography. Within E2/WaterCast, the terrain wizard allows the user to import a pit-filled DEM as the basis for setting catchment and subcatchment boundaries based on stream lines and terrain. The DEMs (not pit filled) supplied for this investigation were at 15, 25 and 90m resolution. Given the very large area to be covered by the Murraylands E2/WaterCast model, the 90m DEM was first chosen to undertake the initial model building steps of developing a pit-filled DEM and undertaking preliminary catchment delineation in E2/WaterCast.
In E2/WaterCast, the DEM must be “pit-filled” so as to make it hydrologically sound (i.e. water from the top of the catchment will drain to the catchment outlet without being trapped in holes or depressions in the DEM). Pitfilling of the DEM was undertaken in TauDEM (a plug-in in MapWindow 3.1) and Streambuilder (Plug-in for MapInfo). Pitfilling was a critical step in the model building process to ensure that E2/WaterCast was able to map streams and develop subcatchments for comparison with hand drawn maps. Following pitfilling, the Digital elevation model was trimmed (to reduce file size) and resized from a 90m grid to a 150m grid so to as to optimise model processing.
Figure 4-1 shows a pitfilled DEM for the Murraylands region, highlighting the vast flat terrain that characterises the majority of area covered by the model.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-2
Figure 4-1 3D View of pitfilled DEM
4.2.2 Developing the E2/WaterCast Subcatchment Map.
The pitfilled DEM was then loaded into E2/WaterCast and subcatchments (and streams) delineated by selecting a node at the outlet to Lake Alexandrina and allowing the model to calculate subcatchments according to a designated stream threshold (100km2). The stream threshold established the amount of detail (number of subcatchments represented by a model) to be included in a model. For example choosing a 100km2 stream threshold for a 1000km2 catchment will result in approximately 10 subcatchments.
4.2.3 Modifying the Subcatchment Map.
The subcatchment map delineated using E2/WaterCast was then compared to the subcatchments as suppled by SA EPA, with areas of inconsistency checked with stream mapping for later correction. In conjunction with this comparison, the subcatchment delineation was compared to the locations of stream gauges. Where stream gauge data existed to allow model calibration, a more detailed subcatchment delineation was undertaken however for regions where no stream gauge data existed, larger subcatchments sufficed. The third consideration for defining subcatchments was the rainfall
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-3
gradient. Subcatchments containing areas of both high and low average annual rainfall zones were avoided (for example by including several smaller sub-catchments within the Eastern Mt Lofty ranges tributary catchments). With these three key amendments in mind, the subcatchments map was extracted from E2/WaterCast and modified using GIS for subsequent reloading into E2/WaterCast, thus being consistent with existing subcatchments maps, location of gauges and rainfall gradients.
The modified subcatchments map is shown in Figure 4-2. As this figure shows, the modified subcatchment map indicates some areas of high detail (i.e. the Mt Lofty Ranges) and areas of relatively low detail (northern catchments). The key discretising elements used in developing a model with a large range of subcatchment areas included:
• Low to very low rainfall zones (200-300mm/yr) in northern and eastern regions resulting in additional model complexity without adding to overall model results;
• Lack of flow gauges, stream network map and water quality information in northern and eastern regions resulting in uncalibratable model regions.
• Lack of land use data in some regions
• Lack of relief or resolution in the DEM within flat areas (e.g. Mallee region) to allow the model to predict water flows to the Murray River during large rainfall events. This is primarily due to a large number of pits and depressions in the raw DEM. These depressions can be pitfilled as the remainder of the DEM, however the accuracy of doing this in map regions where no land use map or stream network was available could have resulted in highly erroneous results. Cross checking the overall catchment outline with Australian river basins data (Geoscience Australia, 1997), revealed overall catchment delineation to be consistent with the South Australian components of the Lower Murray and Wimmera Mallee basins with omitted areas to the south belonging to the Millicent Coast basin.
• Aerial photography and satellite imagery for some of the regions outside of the modelled regions do not show clear steam lines (particularly in very flat regions), creating areas in the study area where model built stream networks may be highly inaccurate.
Thus it was decided to focus the modelling effort and detail to regions with good flow and water quality data availability and higher rainfall zones (where surface runoff is more likely to occur). A total of 122 subcatchments make up the Murraylands E2/WaterCast model ranging in size from 0.02- 13,175km2, with median catchment size of approximately 76km2.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-4
Figure 4-2 Murraylands Subcatchment Map
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-5
4.3 Step 2 Creation of the Node-Link Network
The node-link network defines the way subcatchments are linked and forms the pathway through which flow is routed to the model outlet. As the subcatchment map for the Murraylands E2/WaterCast model was created manually, the node-link network must also be created manually. This step was undertaken by referring to the river and stream network as supplied in the original data sets as well as the original DEM based network definition delineated from the pit filled DEM. The node-link creation step is shown in Figure 4-3.
Figure 4-3 Murraylands Model Node-Link Network
4.4 Step 3 – Functional Unit Definition
4.4.1 Land Use Map – Creating a Raster for E2/WaterCast.
The land use map supplied for this investigation was highly detailed and contained a large number of land use classes representing between 30 and 102 land uses, depending on the classification system adopted. A complete list of all 102 available land use classes relevant to the Murraylands E2/WaterCast region is contained in Appendix B. The land use raster used for the Murraylands E2/WaterCast model was reclassed to combine similar land uses resulting in a final raster containing approximately 25 different land uses as shown in Figure 4-4. This shows that the majority of area covered by the model is classed as grazing intermixed with cropping, along with more intensive land uses along the Murray River and Southern Mt Lofty Ranges.
This land use map was then modified further for use in the E2/WaterCast model by reducing the number of land use classifications (through lumping). This process is outlined below and was essential to reduce the model complexity and achieve acceptable model run times.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-6
Figure 4-4 Murraylands E2/WaterCast Model Land Use Map
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-7
4.4.2 Redefining Land Uses for the Murraylands E2/WaterCast Model
Land use classes (or Functional Units, denoted FUs in E2/WaterCast) in excess of 10-15 are the current limit in the Murraylands E2/WaterCast modelling system due to computational limitations resulting from both the large number of subcatchments and large spatial area covered by the model. Therefore, as many of the land uses classified in the map only cover a small area compared to other land uses, grouping has been undertaken to reduce model complexity and achieve acceptable run times.
Functional Units provide the means through which pollutants are generated and transported to the catchment outlet, thus it is important to be able to define individual pollutant generation characteristics for each FU. Reliable data describing the pollutant generation rates are not available for all 102 land use classes in the Murraylands E2/WaterCast region, thus providing additional reasoning to lump FUs into a smaller number of classes that do have pollutant generation data available.
Thus, in lumping the land use classes (FUs) for the Murraylands E2/WaterCast model, consideration was given to both the computation limits of the modelling framework as well as the availability of pollutant generation data to describe these FUs.
The process to reduce the land use classes and produce a land use map for the Murraylands E2/WaterCast model is described below:
1. The land use map classes were analysed for similarities and grouped accordingly (such as different forms of urban land use were grouped as one ‘urban’ class);
2. The Murray River and lake Alexandrina outline were stamped into the grouped land use map;
3. The land use map was converted to a land use raster with specific integer classes;
4. A null land use class was assigned to all land uses outside the land use raster to eliminate any ‘holes’ in the raster (these holes can effect the smooth running of E2/WaterCast); and
5. The land use raster was resized to an appropriate grid size (in this case 100m cell size)
Table 4-1 describes the 13 land use classes or that have been incorporated into the Murraylands E2/WaterCast model.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-8
Table 4-1 Functional Units of the Murraylands E2/WaterCast Model
Functional Unit (FU) Incorporating land use classes No assigned land use Areas included in the model that fell outside the available land use map Grazing, natural vegetation Grazing, modified pasture Incorporating irrigated modified pastures Minimal uses Incorporating mining Urban Incorporating residential, roads, transport, commercial services, airports, utilities, electrical generation/transmission, Cropping Included all forms of cropping Water Including reservoirs, aqueducts, lakes, evaporation basins, river, farm dams, and aquaculture Nature conservation Including national parks, habitiat/species management, natural feature protection, conservation area Plantation forest Including irrigated plantation forestry and hardwood production Marsh Including wetlands Intensive animal production Included dairy, poultry and pigs and cattle production (not grazing) Horticulture Included all forms of seasonal and perennial, irrigated and non irrigated horticulture Intensive horticulture
For each FU within the Murraylands E2/WaterCast network, models needed to be defined and parameterised to describe the hydrology within the catchment. Consideration of the catchment response given the variability of terrain and long periods without rainfall was made and the SimHyd runoff generation model was selected for all FUs with the EMC/DWC constituent generation model being used for all functional units.
4.5 Step 4 – Node Model selection
Node models were not incorporated in the Murraylands E2/.WaterCast model.
4.6 Step 5 – Link Model Selection
Nodes within the Murraylands E2/WaterCast model were joined through links to the catchment outlet. These links route both flow and constituents to downstream nodes and as such can be configured in several different ways to represent in stream processes (Argent et al. 2006).
Initially the links were configured with a Muskingham-Cunge flow routing model so as to represent the lag in flows from the top of the catchment to the downstream outlet. However insufficient information was available for the majority of subcatchments to fully describe (calibrate) the routing of flow and any nutrient transformation. Therefore for almost all model links, both the decay and routing models were disabled for all scenarios, providing catchment pollutant load exports rather than indications of any in- stream pollutant flow transformation. The exception to this situation was for links in the lower Angas and Bremer rivers. Several flow gauging stations are positioned at intervals on these rivers providing
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-9
sufficient information to undertake flow routing and account for instream losses. The routing procedure adopted for these rivers is outlined in further detain in Section 4.9.
The modelling of daily pollutant concentrations was undertaken by reintroducing lag flow. This was primarily to smooth the peak concentration data and did not affect pollutant load exports.
4.7 Step 6 – Climatic Data Inputs
Given the large areas covered by the Murraylands E2/WaterCast model, gridded SILO data was used for the generation of both daily evaporation and daily rainfall for all model subcatchments. The SILO rainfall grids provide interpolated climate data on a 5km x 5km grid. This data is processed within the E2/WaterCast modelling framework and averaged across each FU in each subcatchment to provide a single rainfall time series for each FU in each subcatchment. This process thus accounts for rainfall variability both within each subcatchment and between subcatchments.
Due to the very large spatial area covered by the Murraylands E2/WaterCast model, computation of a daily rainfall time series for each FU in each subcatchment would result in approximately 1500 rainfall time series averaged from anywhere between 1 and 200 SILO daily rainfall series (covering 36 years of data). This volume of computation far exceeds the current E2/WaterCast model framework capability, thus only one rainfall and PET time series per subcatchment was calculated for the model. To compensate for this potential loss of resolution in the model, subcatchments in known high rainfall zones (or high rainfall gradient zones) were split or made smaller to capture the spatial variability of climate.
4.8 Step 7 – Parameterisation and Calibration
Parameterisation and calibration of the Murraylands E2/WaterCast model is a very important step in the model construction process as the results of this step largely dictate the runoff generated and associated constituent loads exported from each subcatchment.
4.8.1 Flow Gauging Data
Flow gauging data was made available for a number of rivers and streams in the Murraylands E2/WaterCast model region. These gauge records were analysed for suitability in model calibration and converted to time series that could be imported into the Rainfall Runoff Library (RRL) for calibration. The following gauge sites were found to be most appropriate for model calibration and verification.
In addition to the flow gauging sites provided in Table 4-2, a large number of other flow gauging locations were available. However these records were deemed unsuitable as they usually contained water elevation data only (unaccompanied by a rating table) and/or were located on irrigation channels
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-10
Table 4-2 Flow Gauging Sites Used for Model Calibration
Site Station name Commenced Ceased A4261028 SAUNDERS CREEK @ Saunders Gorge 27/05/2002 AW426503 ANGAS RIVER @ Angas Weir 29/05/1964 AW426504 FINNISS RIVER @ 4Km East Of Yundi 8/06/1965 AW426529 MARNE RIVER @ U/S Cambrai 6/12/1972 1/05/1989 AW426530 CURRENCY CREEK @ near Higgins 6/06/1972 23/08/1993 AW426533 BREMER RIVER @ near Hartley 11/05/1973 AW426536 BURRA CREEK @ Worlds End 15/01/1974 AW426557 MOUNT BARKER CREEK @ D/S Mt. Barker 24/04/1979 AW426629 ANGAS RIVER @ Angas Plains 7/06/1991 2/12/2002 AW426662 JANE CREEK @ Brukunga Mine 6/08/1993 AW426679 MT BARKER CREEK @ U/S Bremer River 11/06/1997 AW426688 BREMER RIVER @ U/S Mt Barker Confluence 15/10/1997 -
4.8.2 Water Quality Data
Water quality data was made available for 12 sampling stations on a number of rivers and streams in the Murraylands E2/WaterCast model region. These records generally represent periodic grab sampling over a number of years with the longest record available being that from the Finniss River (1970-2007) of approximately 400 sampling occasions. The water quality monitoring stations were as follows:
• Angas River - downstream Strathalbyn GS426564 and Angas weir GS426503
• Bremer River - Wanstead Rd and Hartley GS426533
• Burra Creek: Worlds End GS426536
• Dawesley Creek: @ Dawesley GS426558
• Finniss River - 4km east of Yundi GS426504 and Winery Rd
• Marne River - Mannum Rd and upstream Cambrai GS426529 and south of Cambrai
• Mt Barker Creek: downstream Mt Barker GS426557
In addition to ambient water quality, water quality and flows were also provided for the Bird in Hand Wastewater Treatment Plant Discharging to Dawesley Creek as shown in Table 4-3 and the Mt Barker STP shown in Table 4-4.
Typical monthly flow rates for the Bird in Hand and Mt Barker STPs are shown in. Table 4-5. The Bird in Hand treatment plant has an approximate annual discharge of 230 ML/yr and the Mt Barker STP has an approximate discharge of 344ML/yr. These monthly flows were used to represent the inflow time series in the Murraylands E2/WaterCast model. Water quality parameters applied in the model were derived from median data from STP discharges (where available) and are shown in Table 4-6.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-11
Table 4-3 Bird In Hand Wastewater Treatment Plant Characteristics
Bird In Hand Wastewater treatment plant_EPALIC#1852, Discharge to Dawesley Creek Water Quality 2005-2006 ANALYTE Units No. Samples MAX MIN MEDIAN LOAD (kg/d) PHYSICAL pH 9 8.8 6.7 8.1 Conductivity uS/cm 9 16001180 1300 TDS (by EC) mg/L 9 882 650 715.5 451.14 BIOCHEMICAL BOD mg/L 9 100 12 22.6 23.37 BOD sol mg/L 9 17.5 2 3.5 4.13 COD mg/L 9 362 124 162 109.54 SS mg/L 9 140 25 47.5 36.12 NUTRIENTS MBAS mg/L 9 0.52 0.13 0.28 0.16 GREASE mg/L 8 10 1 2.5 2
NH3-N mg/L 9 37.3 1.43 26.2 13.92 OxN mg/L 9 11 0.02 0.29 1.82 TKN mg/L 9 43.8 14.3 32.55 19.57 TOTAL P mg/L 9 12 7.5 9.98 6.15 CATIONS CALCIUM mg/L 9 30.9 23.2 25.9 15.84 MAGNESIUM mg/L 9 19.4 12.6 15.6 9.66 SODIUM mg/L 9 256 148 177 111.96 METALS ALUMINIUM mg/L 9 0.8530.092 0.246 0.2099 BORON mg/L 2 0.3090.145 0.227 0.1317 CADMIUM mg/L 2 0.00050.0005 0.0005 0.0003 COPPER mg/L 9 0.02190.0015 0.002 0.003 IRON mg/L 2 1.17 0.2220.696 0.4192 LEAD mg/L 2 0.00050.0005 0.0005 0.0003 MERCURY mg/L 9 0.00050.0003 0.0003 0.0002 NICKEL mg/L 2 0.00390.0005 0.0022 0.0011 SILVER mg/L 9 0.0020.002 0.002 0.0013 ZINC mg/L 9 0.0580.003 0.006 0.0075 OTHER SAR 9 9.32 5.83 6.5 MICROBIOLOGY E. coli /100 mL 15 3600 9 62
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-12
Table 4-4 Mt Barker Wastewater Treatment Plant Characteristics (irrigation data)
ANALYTE Units No. Samples MAX MIN MEDIAN PHYSICAL TDS (by EC) mg/L 11 1100 630 830 BIOCHEMICAL BOD mg/L 11 82 3 SS mg/L 11 882 7 NUTRIENTS
NH3-N mg/L 11 22.50.13 2.25 OxN mg/L 11 16.91.70 9.87 TKN mg/L 11 * * * TOTAL P mg/L 11 * * * MICROBIOLOGY E. coli /100 mL 11 230 0 60 *Not available in the data set
Table 4-5 Monthly STP Discharge
Month Bird in Hand STP Mt Barker STP (ML/month) (ML/month) January 1.17 0 February 0 0 March 0 0 April 21.98 0 May 46.98 9.5 June 22.17 67.1 July 16.88 85.7 August 31.04 62.0 September 28.49 52.6 October 33.96 42.9 November 26.96 24.3 December 0 0 Total 229.63 344.1
Table 4-6 Adopted E2/WaterCast STP Water Quality Characteristics
Constituent Units Bird in Hand STP Mt Barker STP TSS mg/L 47.5 2 TN mg/L 32.5 12.1 NOx mg/L 0.29 9.87 NH4 mg/L 26.2 2.25 TP mg/L 9.98 0.14 FRP mg/L 0.98 0.07 TDS mg/L 715.5 830 Ecoli /100ml 62 60
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-13
4.8.3 Constituents and Generation
The E2/WaterCast model constituent generation and export is based on event runoff monitoring of individual land uses. These models only allow for constant values to be set for base flow and event flow conditions (the Event Mean Concentration and Dry Weather Concentration approach).
The water quality constituents modelled were sediment; Total Suspended Solids (TSS), nutrients,
Total Nitrogen (TN), Ammonia (NH4), Oxidised Nitrogen (NOx), Total Phosphorus (TP), Filtered Reactive Phosphorus (FRP), salt; Total Dissolved Solids (TDS), and bacteria; E.coli, and Cryptosporidium.
The water quality data was assessed for suitability in model calibration, and although extensive at some locations, the lack of specific event data currently prevented the derivation of catchment specific EMC and DWC values. Instead, the EMC and DWC values used in the Murraylands E2/WaterCast model have largely been sourced from work undertaken through the development of Mt Lofty Ranges EMSS and E2/WaterCast modelling, however some data came from additional literature sources and monitoring data. Specifically nutrient speciation was derived as a ratio of total nutrient data based upon ambient monitoring data provided. NOx, NH4 and FRP proportions of total nutrient were 0.04, 0.1 and 0.24 respectively and represent median proportions from all available ambient data. The constituents modelled and their respective EMC and DWC values for each land use are presented in Table 4-7
Table 4-7 EMC Parameterisation TSS TN NOx NH4 TP FRP E.Coli Crypto TDS Land Use mg/L mg/L mg/L mg/L mg/L mg/L cfu/L oocysts/L Mg/L No assigned land use 20 0.8 0.032 0.08 0.20 0.048 6900 0.021 1250 Grazing, natural vegetation 140 1.6 0.064 0.16 0.28 0.067 36000 0.53 1250 Grazing, modified pasture 140 1.95 0.078 0.195 0.36 0.086 36000 0.53 1250 Minimal uses 140 0.8 0.032 0.08 0.20 0.048 6900 0.021 1250 Urban 140 1.6 0.064 0.16 0.28 0.067 63000 9.8 1250 Cropping 140 2.1 0.084 0.21 0.36 0.086 14000 0.042 1250 Nature conservation 20 0.8 0.032 0.08 0.20 0.048 6900 0.021 1250 Plantation forest 20 0.8 0.032 0.08 0.20 0.048 6900 0.021 1250 Marsh 20 0.8 0.032 0.08 0.20 0.048 6900 0.021 1250 Intensive animal production 550 2.2 0.088 0.22 1.0 0.240 63000 9.8 1250 Horticulture 140 2.7 0.108 0.27 0.45 0.108 36000 0.53 1250 Intensive horticulture 550 5.2 0.208 0.52 1.0 0.240 36000 0.53 1250
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-14
Table 4-8 DWC Parameterisation TSS TN NOx NH4 TP FRP E.Coli Crypto TDS Land Use mg/L mg/L mg/L mg/L mg/L mg/L cfu/L oocysts/L Mg/L No assigned land use 7 0.4 0.016 0.04 0.03 0.007 0 0 1620 Grazing, natural vegetation 10 0.7 0.028 0.07 0.07 0.017 0 0 1620 Grazing, modified pasture 10 0.7 0.028 0.07 0.07 0.017 0 0 1620 Minimal uses 10 0.4 0.016 0.04 0.03 0.007 0 0 1620 Urban 10 0.7 0.028 0.07 0.07 0.017 0 0 1620 Cropping 10 0.7 0.028 0.07 0.07 0.017 0 0 1620 Nature conservation 7 0.4 0.016 0.04 0.07 0.017 0 0 1620 Plantation forest 7 0.4 0.016 0.04 0.03 0.007 0 0 1620 Marsh 7 0.4 0.016 0.04 0.07 0.017 0 0 1620 Intensive animal production 10 0.7 0.028 0.07 0.07 0.017 0 0 1620 Horticulture 10 0.7 0.028 0.07 0.07 0.017 0 0 1620 Intensive horticulture 10 0.7 0.028 0.07 0.07 0.017 0 0 1620
4.8.4 Hydrological Parameterisation
The Rainfall Runoff Library (RRL) tool was utilised to undertake the optimisation of parameter values within SimHyd and thus perform subcatchment calibration. RRL is another product available through the Catchment Modelling Toolkit and contains several different hydrological models, plus a dedicated calibration window that allows the user to calibrate the hydrology using a variety of objective functions and manual methods in order to obtain suitable parameters for use in the selected model.
Numerous iterations of the RRL calibration tool were undertaken to identify parameters that would yield suitable calibration results based on the observed data at flow gauges. In the majority of iterations, the Nash-Sutcliffe criterion (Coefficient of Efficiency) was used as the primary basis for determine initial calibration performance with runoff difference being used as secondary criteria. Optimising the model efficiency was first performed undertaken on the monthly flow time series and secondary manual calibrations were undertaken manually on the daily flow time series. This calibration strategy was found to work best for the Murraylands E2/WaterCast catchments and can be thought of as first calibrating for volume of runoff and then for hydrograph shape.
10 subcatchments were calibrated using the RRL tool and an additional 2 catchments (made up from smaller calibration catchments) were also checked against monitoring records as a form of model verification. Figure 4-5 show the relative locations of the Murraylands E2/WaterCast calibration catchments. Note that no suitable stream calibration data was available east of the Murray River and that the majority of data was available for the Mt Lofty region.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-15
Figure 4-5 Murraylands E2/WaterCast Calibration Catchments
Summary parameters for each of the model calibrations are provided in Table 4-6 and Table 4-10. Figure 4-6 - Figure 4-8 show how SimHyd was used to successfully calibrate subcatchment 3 represented by Gauge AW426403. Generally, the south eastern Mt Lofty catchments appear to calibrate better than the northern catchments.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-16
Table 4-9 Subcatchment Calibration Summary Nash-Sutcliffe Efficiency Catchment Catchment Modelling period Site (Monthly flow volume) reference** area Calibration Verification Calibration Verification A4261028 10 66.5 km2 11/02 - 6/06 na - na AW426503 3 61 km2 3/72 - 12/87 1/89 - 12/06 0.9 0.7 AW426504 2 191 km2 4/70 - 2/85 1/88 - 2/06 0.84 0.83 AW426529 11 240 km2 12/72 - 12/06 na 0.52 na AW426530 1 58 km2 4/74 - 1/82 1/83 - 12/92 0.95 0.8 AW426533* 9 481 km2 na 1/73 – 12/06 na na AW426536 12 560 km2 4/74 - 12/05 na 0.63 na AW426557 5 87 km2 1/81 - 2/92 1/82 - 12/06 0.86 0.77 AW426629* 4 190 km2 na 1/97 - 12/01 na na AW426662 6 26.7 km2 1/94 - 12/99 12/99 - 12/06 0.59 0.82 AW426679 8 231 km2 9/97 - 6/02 6/02 - 12/06 0.72 0.73 AW426688 7 197 km2 7/98 - 12/01 12/01 - 12/06 0.82 0.74 *denote model verification catchments ** Refer to Figure 4-5 for catchment reference locations na = not applicable
Table 4-10 Subcatchment Calibrated Hydrological Parameters
SimHyd Parameters* Site Ref Area bc i ic is itc pf risc rc smsc A4261028 10 66.5 km2 0.15 5.0 228 5.2 0.00 1.0 5.0 0.53 395 AW426503 3 61 km2 0.240 5.0 256 6.2 0.01 1.0 5.0 0.44 415 AW426504 2 191 km2 0.120 5.0 256 6.2 0.09 1.0 5.0 0.13 500 AW426529 11 240 km2 0.110 5.0 304 3.9 0.20 1.0 4.5 0.34 80 AW426530 1 58 km2 0.130 5.0 256 4.3 0.10 1.0 3.6 0.36 415 AW426536 12 560 km2 0.007 5.0 210 7.0 0.00 1.0 0.0 0.05 410 AW426557 5 87 km2 0.072 5.0 132 5.6 0.15 1.0 0.5 0.48 240 AW426662 6 26.7 km2 0.120 5.0 400 6.1 0.02 1.0 2.0 0.22 500 AW426679 8 231 km2 0.530 5.0 188 3.0 0.02 1.0 5.0 0.13 250 AW426688 7 197 km2 0.240 5.0 136 5.0 0.01 1.0 1.1 0.08 240 * bc = baseflow coefficient, i = impervious threshold, ic = infiltration coefficient, is = infiltration shape, itc = interflow coefficient, pf = pervious fraction, risc = rainfall interception storage capacity, rc = recharge coefficient, smsc = soil moisture store capacity.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-17
Figure 4-6 RRL Calibration and Verification For Gauge AW426403
Figure 4-7 RRL Calibration of Daily Flows (Small Events) For Gauge AW426403
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-18
Figure 4-8 RRL Calibration of Daily Flows (Large Events) For Gauge AW426403
The Murraylands E2/WaterCast subcatchment model was carefully constructed to contain subcatchment outlets at gauge stations used for model calibration. Thus parameterisation of the calibration catchments was made relatively easily by adopting these calibration parameters. Hydrological parameterisation of the remainder of the model involved the adoption of parameter sets from nearby calibrated catchments.
Figure 4-9 shows the spatial application of calibration parameter sets and Table 4-11 shows the adopted hydrological parameters for the remaining ungauged catchments represented by the model.
Table 4-11 Adopted Hydrological Parameters
SYMHYD Parameters Zone Gauge Ref. bc i ic is itc pf risc rc smsc 1 AW426536 0.007 5.0 210 7.0 0.00 1.0 0.0 0.05 410 2 A4261028 0.15 5.0 228 5.2 0.00 1.0 5.0 0.53 395 3 AW426688 0.240 5.0 136 5.0 0.01 1.0 1.1 0.08 240 4 AW426679 0.142 5.0 136 5.6 0.04 1.0 3.5 0.27 240 5 AW426503 0.240 5.0 256 6.2 0.01 1.0 5.0 0.44 415 6 AW426504 0.120 5.0 256 6.2 0.09 1.0 5.0 0.13 500 7 AW426530 0.130 5.0 256 4.3 0.10 1.0 3.6 0.36 415
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-19
Figure 4-9 Hydrological Parameterisation of Murraylands E2/WaterCast Model
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-20
4.9 Accounting for Instream Losses
The model calibration process highlighted some apparent discrepancies in selected stream flow. The Bremer and Angas rivers in the EMLR have a number of gauges located in succession in the upper and lower parts of these rivers. These records indicated that a loss of volume was occurring between upper and lower gauge sites potentially impacting on the volume of flow to the lake system. This section details the identification and development of a conceptual model to describe the loss of flow in the lower EMLR streams and the implementation in a ‘link’ in the Murraylands E2/WaterCast Model.
4.9.1 Identification of Overlapping Gauge Records
The stream gauge sites that indicate in-stream losses are those on the lower Angas and Bremer. The available stream gauge records for these rivers are:
• Angas River: AW426503, A4261101, A4261074, AW426629, A4261073;
• Bremer River: AW426533, A4261070, A4261102, A4261072.
Not all of the above sites have overlapping data or extensive records. Table 4-12 summarises gauge data for the above sites. The consistent periods of overlapping data occur between the August 2004- April 2007. The stream gauge sites corresponding to this overlapping period are shown below (Figure 4-10). Three stream gauges with overlapping record are available for both the Angas and Bremer Rivers.
Table 4-12 Gauge Records for Angas and Bremer Rivers
Gauge Start End Comments Angas River Represents only a portion of AW426503 Jun-64 Dec-06 upstream catchment A4261101 Nov-06 Jan-07 73 days, minimal flow data A4261074 Aug-04 Apr-07 - AW426629 Jun-91 Dec-02 Station ceased since 2002 A4261073 Aug-04 Apr-07 - Bremer River Represents only a portion of AW426533 May-73 Mar-07 upstream catchment A4261070 Aug-04 Apr-07 - A4261102 Nov-06 Jan-07 73 days, no flow data A4261072 Aug-04 Apr-07 -
The upstream gauge sites on the Angas and Bremer Rivers (AW426503 and AW426533) represent only a small portion of the total upstream catchment therefore the low-lying gauge sites with proportionally small contributing catchments are best to observe any loss of flows without adding the complication of significant catchment area changes. Therefore gauges used for the in-stream loss comparison are A4261074 and A4261073 on the Angas River and A4261070 and A4261072 on the Bremer River.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-21
Figure 4-10 Stream Gauge Sites With Overlapping Record
4.9.2 Gauged Stream Losses
Plots of flow rate from 2004-2007 for the two stream gauging sites on the Angas and Bremer Rivers are shown in Figure 4-11 and Figure 4-12.
Figure 4-11 Lower Angas Daily Flow Record
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-22
In the lower Angas, the upstream gauge records considerably more flow than the downstream gauge and appears to be consistent across the majority of the available record. The quality of the Angas gauge data is prescribed “Provisional/Theoretical” indicating a theoretical rating curve may have been used to generate the recorded flows and some systematic bias may be embedded in the data as a result of these rating curves. Hydrograph peaks in the downstream record are consistently lower or non-existent and Hydrograph recession and baseflows are also reduced.
Figure 4-12 Lower Bremer Daily Flow Record
In the lower Bremer the two gauge records show large variations, with the lower gauge sometimes recording much higher flows than the upper gauge and visa-versa (for example winter 2005 compared to winter 2006). Like the Angas River gauges, these gauges are also prescribed a quality rating “Provisional/Theoretical”, however systematic bias is not evident between the inter-seasonal data. Hydrograph peaks at the downstream site appear significantly higher in winter 2005 and the hydrograph recession shape appears similar between records.
Given the relative proximity of these gauges such large difference in flows would not be expected, however various factors may contribute to the apparent increase and/or decrease in flows from upstream to downstream sites including:
• Additional rainfall and runoff from small lowland catchments;
• Recording error, and rating curve uncertainty;
• Changes in channel shape as a result of flows (resulting in rating curve uncertainty);
• In-stream contributions from other sources (for example groundwater);
• Water extractions;
• Infiltration; and
• Instream storage.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-23
The magnitude of flow differences is shown in the cumulative flow curves (Figure 4-13, Figure 4-14). The observed difference from upstream to downstream on the Angas River is approximately -50%. The total flow at the end of the catchment is similar to that recorded from 1/3 of the total catchment area in the upper catchment.
Figure 4-13 Cumulative Flow, Angas River
Figure 4-14 Cumulative Flow, Bremer River
The observed difference from upstream to downstream on the lower Bremer River is approximately +25%, even though no substantial inflows are expected at the bottom portion of the catchment.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-24
Despite this anomaly, the recorded flows at the bottom of the catchment are less than that recorded from 80% of the total catchment area in the upper catchment.
The data from the EMLR lowland streams therefore paint a confused picture with the Angas River showing consistent in-stream losses, while the Bremer shows relative gains with occasional losses and minimal flows from the lower portions of the catchment. Some of the flow discrepancies may be attributable to gauge errors, in addition to potential in-stream processes.
The Murraylands E2/WaterCast model can be amended to account for instream losses to better predict flows from the Angas and Bremer Rivers to the Lakes system. This may be particularly critical in dry years such as 2006 when flows to the lake from the EMLR may be significantly less than predicted by the E2/WaterCast model. The stream gauge data for winter 2006 indicates approximately 20% decrease in flow for the Angas River and 40% decrease flow for the Bremer River (Figure 4-15 and Figure 4-16)
Figure 4-15 Winter 2006 Angas River Flow
Figure 4-16 Winter 2006 Bremer River Flow
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-25
4.9.3 Conceptual Representation of In-stream Losses
A conceptual in-stream loss model has been formulated to incorporate into the E2/WaterCast model as an ‘E2/WaterCast storage’ and therefore allow for a loss of flow at a link, particularly in the lower reaches of the Angas River and potentially in the Bremer and other low land river links.
The loss model is conceptualised as an in-stream storage of given volume (Smax) (ML), minimum (MinL) and maximum (MaxL) loss rate (ML/d) to represent seepage through the bottom and sides of the storage, and slow release rate (%) of the storage to the downstream node. The loss rate is calculated in relation to the storage level using linear interpolation between MaxL and MinL. Thus, when the in-stream storage is full the loss rate through seepage is maximised and when the storage is empty the loss rate is minimised. The slow release represents the ‘routing’ of low flows through the storage to the next node. All flows in excess of the storage capacity first fill up the storage and then overtop it by passing straight through to the next node.
The above conceptual model is applied in E2/WaterCast via three relational tables describing:
• Storage level vs. volume;
• Storage level vs. losses; and
• Storage level vs. release rate (outflow).
The storage level vs. volume table was allocated constant area thus area was directly proportional to volume. Flows that overtop the storage pass straight through. The remaining two tables were constructed using the following formulae:
(MaxL − MinL) Losst = Volumet × Equation 1 Smax
Outflow = % × (Volumet − Losst ) Equation 2
Smax = Maximum storage volume
MaxL = Maximum loss rate
MinL = Minimum loss rate
% = Release rate
Losst = Instream loss through seepage at time t
Volumet = Storage volume at time t
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-26
4.9.4 Loss Model Implementation
Individual loss models for the Angas and Bremer Rivers were first implemented in a spreadsheet using recorded data for the period 2005 – 2007. The spreadsheet model was implemented to aid calibration. Model efficiency (Nash Sutcliff Efficiency) was used in addition to volumetric error to judge the performance of the model. Generally, only three seasons of concurrent data were available to test the model. The first two seasons were generally used as calibration data sets and the third season used as a verification data set.
The Angas River model under predicts in 2005/6 but over predicts in 2007. The overall difference in volume for the entire period is just 0.5%. The overall loss of flow predicted by the model (between the upstream gauge and the downstream gauge is approximately 50% or 6200 ML over 3 years.
Figure 4-17 Angas River instream loss model calibration/verification
The Bremer River model over predicts in 2005/6 and in 2007. The overall difference in volume for the entire period is just 3%. The overall loss of flow predicted by the model (between the upstream gauge and the downstream gauge is approximately 30% or 5700 ML over 3 years.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DEVELOPMENT OF THE MURRAYLANDS E2/WATERCAST MODEL 4-27
Figure 4-18 Bremer River instream loss model calibration/verification
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-1
5 RESULTS AND DISCUSSION
Currently, 2 scenarios have been generated for the Murraylands E2/WaterCast model including an existing case and predevelopment case through changes to the land use raster only.
The development of the existing case can now form the basis of any subsequent scenarios and the results from this scenario are outlined in detail below.
5.1 Existing Scenario
The construction of the Murraylands E2/WaterCast model as outlined in Section 4 provides the base model (‘existing’ scenario) from which further scenarios can be generated. This section presents the preliminary results from this existing case.
The ‘existing’ scenario models the entire Murraylands region using:
• Existing land use data (2003);
• Rainfall/runoff SYMHYD parameters from several flow calibrated subcatchments;
• EMC/DWC values associated with existing land use data;
• 36 years of subcatchment averaged SILO rainfall data (January 1970-June 2007); and
• Disaggregated gridded monthly PET data.
The results of this existing scenario are presented in the sections below.
5.1.1 Spatial Representation
The results of the Murraylands E2/WaterCast model can be viewed spatially by extracting results and post processing in a GIS package. This was achieved by extracting individual subcatchment outputs (such as flows and pollutant loads) importing these outputs into a MapInfo table. This data was then converted to gridded data with applied colour shading allowing for quick visual inspection and interpretation of results. Selected model outputs are provided in Figure 5-1 to Figure 5-5 below and include annual rainfall, predicted flows and pollutant loads for TSS, TN and TP.
5.1.1.1 Annual Rainfall
Much of the Lower Murraylands region covered by the E2/WaterCast model receives relatively little annual rainfall. Figure 5-1 shows the results of spatial analysis of the SILO rainfall data averaged across each catchment for the Murraylands E2/WaterCast model. The average annual rainfall for the 36 years (1970-2006) for the entire 68,000 km2 catchment is approximately 305mm, however the rainfall varies across the catchment from 228-876mm/yr.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-2
Figure 5-1 Mean Annual Rainfall, 1957-2006
As Figure 5-1 shows, the areas of highest rainfall are concentrated in the Mt Lofty region, however a relatively steep rainfall gradient exists east of the Mt Lofty ranges and to the north of the basin. The E2/WaterCast model calculates subcatchment rainfall by averaging the 5km x 5km SILO rainfall grid, therefore requiring subcatchments with high rainfall gradients to be split to account avoid losing the rainfall variability. As discussed previously, to account for the rainfall gradient observed east of the Mt Lofty region, smaller subcatchments have been incorporated into this section of the model to capture this variability. The annual rainfall distribution gives a good indication of where the majority of catchment flows will originate.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-3
5.1.1.2 Flows
Figure 5-2 shows the mean annual flow per unit area for the Murraylands E2/WaterCast model and identifies the high rainfall zones as indicated by Figure 5-1 and includes the adjustments in runoff through the application of the individually calibrated SYMHYD models. As expected, on a per unit area basis, the Eastern Mt Lofty Range catchments deliver the highest flows.
Figure 5-2 Mean Annual Areal Flows
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-4
The modelled flows from the Eastern Mount Lofty Catchments of the Marne, Bremer, Angas, Finniss rivers in addition to flows from Tookayerta Creek and Currency Creek have also been independently modelled by CSIRO (2007) This modelling study also used the SimHyd model calibrated against four gauges. Direct comparison between the present study and the CSIRO modelling study shows good agreement between both the calibration performance and the total flows modelled from these rivers as shown in Table 5-1 and Table 5-2.
Table 5-1 shows a comparison between calibration performance as measured by the Nash-Sutcliffe coefficient of efficiency for common stream gauges for the CSIRO (2007) and the present studies. The results show both models performing well with good correlation between measures and predicted flows at these gauges.
Table 5-1 Comparison Between CSIRO and E2/WaterCast Calibration Performance Nash-Sutcliffe Efficiency Monthly flow volume (verification performance) Gauge WaterCast CSIRO (2007) 0.86, 0.59, 0.72, 0.82 42605330* (Bremer) (0.77, 0.82, 0.73, 0.74) 0.82 42605030 (Angas) 0.9 (0.7) 0.78 42605040 (Finniss) 0.84 (0.83) 0.9 42605300 (Currency) 0.95 (0.8) 0.9 *Four upstream gauges were used for calibration for the WaterCast model
Table 5-2 shows a comparison between modelled average annual end of catchment flows for the Marne, Bremer, Angas, Finniss Rivers in addition to flows from Tookayerta Creek and Currency Creek.
Table 5-2 Comparison Between CSIRO and E2/WaterCast Predicted Flows WaterCAST CSIRO Predicted Difference (%) Catchment Predicted GL/yr GL/yr* Marne River 8.9 8.9 0.4% Bremer River 18.9 19.8 -4.5% Angas River 6.3 15.1 -58.0% Finniss River 33.3 42.7 -22.0% Tookayerta River 8.0 16.7 -52.1% Currency Creek 7.2 7.4 -2.9% total 82.7 110.6 -25.3%
The modelling periods over which these average annual flows differ, however some significant differences can be seen in the comparison including the Finniss, Angas, Bremer and Tookayerta flows. The large differences in modelled flows on the Bremer and Angas Rivers are primarily a result of instream losses as modelled by WaterCast rather than significant differences due to surface runoff calibration. Instream loss modelling presented in Section 4.9 shows potential losses on the Angas River of 43% and 22% for the Bremer River.
Differences in the Finniss and Tookayerta flows are likely to be a result of differences in calibration and/or a combination of model evaluation period and subcatchment area contributions. The Tookayerta catchment has been parameterised with Finniss River parameters. This may not have
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-5
been the case for the CSIRO (2007) study which may have use Currency Creek parameters for this catchment.
Overall, the difference in average annual modelled flows between CSIRO (2007) and the present study for the above streams is approximately -25%. Without instream losses, the difference between studies is approximately 16%.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-6
5.1.1.3 Total Suspended Solids
TSS loads (kg/ha/yr) are shown in Figure 5-3 and are generally in agreement with catchment runoff, although in the Eastern Mt lofty Ranges, variations exist indicating TSS exports are also influenced by the land use characterisation in individual subcatchments.
Figure 5-3 Total Suspended Solids Areal Load Contributions
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-7
5.1.1.4 Total Nitrogen
TN loads (kg/ha/yr) are shown in Figure 5-4 and like TSS, are also generally in agreement with catchment runoff. Variations from catchment runoff rates do exist in the Eastern Mt Lofty Ranges demonstrating the influence of land use differences between subcatchments and requiring further investigation.
Figure 5-4 Total Nitrogen Areal Load Contributions
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-8
5.1.1.5 Total Phosphorus
TP loads (kg/ha/yr) are shown in Figure 5-5 and show very similar patterns as previously demonstrated for TSS and TN areal export rates.
Figure 5-5 Total Phosphorus Areal Load Contributions
Clearly, the mean annual areal plots provided above show that the key area of interest is the Eastern Mt Lofty Ranges catchments. Analysis of these catchments is provided in subsequent sections.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-9
5.1.2 Temporal Representation
Figure 5-6 shows the predicted annual flows from the Murraylands E2/WaterCast catchments from 1970-2006. The predicted mean annual flow is 270,000 ML/yr. The two years (1974 and 1993) show the highest predicted flows due to a single event in each of those years across a number of the larger northern and eastern subcatchments, producing the only significant flows from these regions. These wet years predict the export of large pollutant loads however these events are relatively rare. Limitations in the calculation capacity of the E2/WaterCast modelling framework currently prevent the generation and therefore simulation of more than 36 years of climate input data1.
Figure 5-6 Predicted Annual Flows at Model Outlet
The modelled flows tend to be highly variable from year to year, which may reflect the relatively low annual rainfall across most of the Murraylands E2/WaterCast catchments. With so little annual rainfall across much of the catchment area, a single medium to large rainfall event in one or two of these larger catchments may be sufficient to produce an above average yearly flow.
The variability in annual runoff is similarly translated to predictions of catchment pollutant exports of TSS, TN and TP and shown in Figure 5-7 to Figure 5-9. The model predicts mean annual pollutant exports of 17300t/yr of TSS, 330 t/yr of TN and 52t/yr of TP. The higher variability observed in the annual pollutant load plots correspond to higher runoff years indicating the higher proportion of event based runoff (and associated pollutant load) rather than base flows during these years.
1 The calculation of climate input data for the Murraylands E2 model currently requires significant computing time. To create a single rainfall time series for the large subcatchments in the northern and eastern regions of the model, all of the 5km x 5km gridded rainfall files that fall within each subcatchment must be averaged. Currently, we have found that the model will allow the computation of approximately 36 years of data. Attempts to create longer time series typically end in a computer error indicating that for the large scale Murraylands E2 model, 36 years of time series may currently be the computational upper limit until either the model extents are reduced or computing power increases. The recently released Watercast model may enable longer time series to be incorporated.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-10
Figure 5-7 Predicted Annual Total Suspended Solids Loads at Model outlet
Figure 5-8 Predicted Annual Total Nitrogen Loads at Model Outlet
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-11
Figure 5-9 Predicted Mean Annual Total Phosphorus Loads at Model Outlet
Long term water quality data collected for a number of locations modelled by the Murraylands E2/WaterCast model can also be compared in terms of measured versus predicted concentrations. Figure 5-10 shows the 3-day average predicted TN concentration (averaged to aid readability of the plot) and measured TN for the upper Finniss River at station GS 426504. Modelled data was extracted from node 19.
Figure 5-10 Predicted vs Measured TN for the Finniss River GS 426504
The plot shows relatively good agreement between predicted and measured concentrations, including the general cycles that show an increase in TN associated with winter inflows. No routing
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-12
was included in the Murraylands E2/WaterCast model for this data extraction, therefore following rainfall events, all modelled concentrations return to the baseline 0.6mg/L when event based runoff is not predicted. Noticeably, some concentration peaks are not well represented suggesting some TN EMCs in this catchment may warrant increased values. The ANZECC guideline value applicable for TN at this location is 1mg/L and the plot shows that this guideline is typically exceeded during winter runoff periods.
5.1.2.1 Pollutant Sources (land use loads)
An assessment of individual land use loads was undertaken by the development of separate models for a number of the major land uses with the rainfall-runoff model applied only to the land use of interest. This allowed loads of each landuse to be compared as shown in Table 5-3 and presented graphically in Figure 5-11.
Table 5-3 Relative Land Use Pollutant Loads Mean Annual Loads Land Use t/yr TSS TN TP Grazing (includes natural and 14956 263.1 40.81 modified pastures) Cropping 706 19.6 2.53 Urban 1099 15.8 2.46 Horticulture 281 6.7 0.99 STP Bird in Hand 11 7.5 2.30 STP Mt Barker 2 4.2 0.05 All other remaining land uses 460 17.2 3.31 total 17514 334 52
Due to the extensive areas of the modelled region being covered by grazing lands, it comes as no surprise that the majority of total pollutant loads are exported from this land use. The subcatchments with higher rainfall and runoff (eastern Mt Lofty catchments) produce the highest areal load rates as shown in Section 5.1.1.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-13
Figure 5-11 Relative Land Use Loads
5.1.2.2 Water Quality Objectives
Default ANZECC guidelines have been used to assess modelled stream condition. The relevant guidelines refer to Tables 3.9.8 and 3.9.9 from ANZECC (2000) for South Central Australia low rainfall area. Guideline values have been reproduced below in Table 5-4
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-14
Table 5-4 Ecosystem Protection Guidelines (ANZECC)
Constituent Guideline value (mg/L) Total Nitrogen 1.0 Oxidised Nitrogen 0.1 Ammonium 0.1 Total Phosphorus 0.1 Filterable Reactive Phosphorus 0.04 Turbidity 50 ntu
Box and whisker plots have been prepared for modelled TN and TP concentrations for major streams in the Eastern Mt Lofty Ranges as shown in Figure 5-12 and Figure 5-13.
Figure 5-12 Modelled TN Concentration
Figure 5-13 Modelled TP Concentration
These figures show that the most streams meet guideline values, however event based concentrations will likely exceed guideline values for all streams.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-15
The high values observed for the Bremer River reflect the STP concentrations from the Bird in Hand and Mt Barker STPs. No instream processing was incorporated to simulate the reduction in pollutant concentrations downstream of these STP discharges.
The larger boxes observed for the Finniss River is an artefact of modelling resulting from the routing being turned on to develop these plots. The routing method chosen was the simplest (lag flow) which simply delays flows and pollutant concentrations by 1 day for each link between nodes. As both the Bremer and Finniss Rivers have a number of subcatchments above the monitoring point, higher pollutant concentrations associated with events are lagged across more days resulting in higher median and upper quartile values for these streams.
This model artefact demonstrates sensitivity of pollutant concentration results to routing methodologies. Further refinements to the routing components of the model will again likely alter these plots highlighting the need for further consideration to routing should stream health assessments be a focus for model output.
5.1.3 Instream Losses
The instream loss model implemented on the lower Angas and Bremer Rivers results in significant reductions in mean annual flows. For the Bremer River, the model indicates a reduction of 22% across all years and for the Angas River, the modelled flow reduction is much greater at 43%. These two rivers without instream losses provide approximately 45000Ml/yr to the lake Alexandrina, however with instream losses, this average annual volume reduces to approximately 33000Ml/yr. This average annual volume of unaccounted for (lost) water represents approximately 2cm of depth across Lake Alexandrina (610km2)
5.2 Predevelopment Scenario
The predevelopment scenario has been developed from the existing case model by changing all pollutant export characteristics for all land uses to that of ‘nature conservation’ (i.e, the minimum pollutant export rates). This scenario is very simplistic and also uncertain due to the lack of data available to characterise both the likely differences in hydrological responses such as that due to changes in land use and pollutant exports before catchment development. These land use changes include the potential impacts of reafforestation and removal of farm dam impacts (Zhang et al 2007) However this scenario does provide for a useful comparison against the existing case by providing an estimate of pollutant export results prior to European settlement through changes to EMC/DWC values only.
Figure 5-14 - Figure 5-16 show comparative pollutant loads between the existing case and predevelopment case. The predevelopment case predicts less 79% reduction in average annual TSS load, 50% reduction in average annual TN load and 33% reduction in average annual TP load.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-16
Figure 5-14 Existing vs Predevelopment TSS Loads
Figure 5-15 Existing vs Predevelopment TN Loads
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-17
Figure 5-16 Existing vs Predevelopment TP Loads
5.2.1 Change Scenarios
Further scenarios can now be modelled using the Murraylands E2/WaterCast model. These scenarios could include use of riparian buffers, alterations to farm dam density, reductions in point source inputs and changes to land management techniques (such as fertilizer application) or land use or climate change. These potential scenarios and others have not yet been constructed and assessed for the Murraylands Region and further consultation is recommended before such scenarios are developed and assessed.
5.3 Discussion
The Murraylands E2/WaterCast Model was constructed to encompass a large area. Due to the limited availability of stream flow data, only gauged streams in the upper Eastern Mt Lofty catchments have been used for flow calibration, therefore model results for large portions of the model are considered highly uncertain.
However, for the Eastern My Lofty catchments, flow calibration was relatively successful, indicating that the results derived from this portion of the model can be interpreted with higher confidence. This region of the model also experiences the highest annual rainfall and is a significant contributor to the overall pollutant load to the Murray River and lower lakes in an average year. Pollutant areal loading rates are highest for this part of the model.
Maximum areal pollutant generation rates for the Murraylands E2/WaterCast model were 150kg/ha/yr TSS, 2.2 kg/ha/yr TN and 0.4 kg/ha/yr TP. These pollutant generation rates can be compared to literature values to determine if further model adjustment of constituent parameters is warranted.
Marston et al (1995) summarised nutrient generation rates for a number of Australian catchments. Table 5-5 shows these and other nutrient generation rates for rural and natural landscapes compared to those generated for the Murraylands E2/WaterCast model.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-18
Table 5-5 Constituent Generation Rates
Constituent TN (kg/ha/yr) TP (Kg/ha/yr) Marston et al (1995) (Australia) 0.6-26 0.1-6.4 Wood (1986) (Mt Lofty Ranges) 1.5-5.5 0.1-0.39 Murraylands E2/WaterCast maximum value 2.2 0.4
Similarly, Drewry et al (2006) summarises more recent Australian nutrient generation data which is reproduced below in Figure 5-17.
Figure 5-17 Nutrient Generation Rates (Extracted From Drewry et al 2006)
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-19
Both the data in Table 5-5 and Figure 5-17 show that the nutrient generation rates modelled by the Murraylands E2/WaterCast model are comparable to literature values, particularly those associated with sheep and cattle grazing.
However there may be sound reason to increase the model TN EMC values based upon locally specific data (for example Wood 1986) where the TN generation rates for similar catchments were approximately double those derived from the Murraylands E2/WaterCast model. This evidence is supported by the comparison of in stream TN concentration data with modelled data, which showed good agreement with seasonal variation in TN concentrations, but did not manage to replicate the higher TN concentrations often observed as a result of winter runoff.
It is however likely that the difference in TN export rates between literature values and the Murraylands E2/WaterCast model will be a result of a combination of predicted hydrological (runoff) differences and pollutant concentration data. Nutrient generation is sensitive to the under prediction of catchment flows, therefore, further subcatchment specific investigation of flows may be warranted where more detailed water quality calibrations can be made, perhaps even using smaller subcatchments and actual rain gauge data rather than gridded data. These further investigations are outside of the current study scope.
Incorporation of hydrological and water quality processes associated with farm dams has been a significant challenge within the current Murraylands E2/WaterCast model framework. Although hydrological calibration was relatively successful for a number of subcatchments, these calibrations only superficially incorporate the filling and emptying of on-farm storages. These storages are numerous in the upper My Lofty catchments, often with a number of storages in series as shown in Figure 5-18.
Figure 5-18 Farm Dams in Series, Mt Lofty Ranges, August 2007.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-20
More detailed analysis, building upon the experiences gained through the development of the Murraylands E2/WaterCast model should be focused upon the lower Murraylands and eastern Mt Lofty Ranges to better understand and quantify the effects of these farm dams. Also the losses occurring in the lower reaches of some of the rivers (e.g. Angas and Bremer) warrant further investigation.
Pollutant loads contributions from different land uses reflect the land use map with the order of pollutant contribution being grazing>urban>cropping>horticulture>other. This classification may change, should just the lower Murraylands become a focus of future model scenarios. Comparison of the generated pollutant loading rates with literature values indicates that the model estimates catchment loads with the expected range.
The influence of STP inflows on pollutant exports from catchments has been estimated by the Murraylands model. Data was available for the Bird in Hand AND Mt Barker STPs and showed that the nutrient load from this STP is roughly equivalent to all urban land uses (for TN and TP). The influence of these STP flows on stream hydrology and subsequent model calibration was not investigated in detail by accounting for these recoded flows in the hydrological calibrations. More detailed model scenarios focussed upon the lower Murraylands region should take greater consideration for these and other potential inflows and the impact that they may have on rainfall- runoff model calibration.
The model results show that guideline values for nutrients are generally achieved, with the majority of flow concentrations for TN and TP falling below guideline values. Event based loads however generally exceed guideline values.
Murraylands E2/WaterCast scenarios are currently existing and predevelopment. The predevelopment scenario indicated catchment loads have been increased 5 fold for TSS, 2 fold for TN and 50% for TP. Further model scenarios can now be developed based on further stakeholder engagement targeted to answer specific catchment management questions.
5.4 Summary
The Murraylands E2/WaterCast catchment model was developed that extends form the South Australian - Victoria border to the outlet of the Murray River and encompasses approximately 68,000 km2. The landscape covered by the model ranges from arid lands in the north to low rainfall irrigation lands along the Murray River and moderate rainfall grazing and cropping lands in the Eastern Mount Lofty Ranges. The dominant land use in the model is grazing lands, which in turn dominated the total pollutant loads.
The model boundary and subcatchments were generated using both automated pit filled DEM data and hand drawn subcatchments. A total of 122 subcatchments ranging in size from 0.02-13,175km2 were delineated to represent the Murraylands catchment and were connected via a node link network and routed to the catchment outlet.
Due to computational limitations, only 36 years of daily climate data has been incorporated into the model. Each subcatchment has individual rainfall and PET data. This data is applicable for all land uses within each subcatchment. Smaller subcatchments were built into the model to account for high rainfall gradient zones.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC RESULTS AND DISCUSSION 5-21
Up to 10 subcatchments, predominantly in the south eastern Mt Lofty Ranges, have been calibrated using daily flow records and gridded SILO rainfall data. The length of calibration/verification record varies for each subcatchment and the calibration was generally performed using the Nash-Sutcliffe criterion on monthly flows followed by manual calibration of the daily hydrograph. Nash-Sutcliffe values ranged from 0.5-0.95 demonstrating good, but variable model calibration. The calibration of several subcatchments provided rainfall-runoff parameter sets that were then adopted for all other un- calibrated areas of the model.
An existing case and predevelopment case model were developed and the results analysed. These models showed that the highest pollutant generation rates are associated with the Eastern Mt Lofty catchments (high rainfall zones) and that these pollutant generation rates are within the accepted range in the literature. Furthermore the model demonstrated seasonality in pollutant concentrations as seen by monitored data (Finniss River) although pollutant concentration peaks for TN may yet be under predicted.
This project thus demonstrates that the Murraylands E2/WaterCast model has the potential to be used as a predictive tool to test further management scenarios. There remain limitations to the model in terms of describing impacts of farm dams and variations in pollutant EMCs, however it appears to provide robust predictions over long time frames and compares well to past catchment pollutant export studies.
Further refinements of subcatchment flow and water quality calibrations are recommended for selected streams in the Eastern Mt Lofty ranges and associated lowlands, particularly with respect to event loads, surface water-groundwater interactions and in stream losses. In addition to this, it is recommended that further scenario development be undertaken in response to specific catchment management approaches and appropriate stakeholder engagement.
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC REFERENCES 6-1
6 REFERENCES
ANZECC. 2000. Australian and New Zealand guidelines for fresh and marine water quality. Australian and New Zealand Environment and Conservation Council, Canberra.
Argent, R.M., Murray, N., Podger, G.M., Perraud, J.-M., Newham, L., 2006. E2 catchment modelling software component modules, CRC for Catchment Hydrology, Australia.
Drewey, J., Newham, L., Greene, R., Jakeman, A., Croke., B., 2006. A review of nitrogen and phosphorus export to waterways: context for catchment modelling. Marine and Freshwater Research. No. 58 pp 757-774.
Geoscience Australia. 2007. Australia’s River Basins 1997. National Mapping Division, Geoscience Australia.
Marston, F., Young, W., Davis, R. 1995. Nutrient data book. Nutrient generation rates data book, 2nd Edition. Unpublished CMSS Reference Manual. CSIRO Division Of Water Resources, LWRRDC and SWB, Canberra.
Wood, G., 1986. My Lofty Ranges Watershed: Impact of land use on water quality and implications on reservoir water quality management. EWS Dept, 86/19.
Zhang L., Vertessy R., Walker G., Gilfedder M. and Hairsine P. (2007) Afforestation in a catchment context: understanding the impacts on water yield and salinity. Industry report 1/07, eWater CRC, Melbourne, Australia
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DATA SUPPLIED A-1
APPENDIX A: DATA SUPPLIED Folder Data File Name(s) Type of Year Producer Description & How collected Accuracy Name Data Catchment SAMDB_Catchment_Bound GIS Unknown DEH South Australian Murray Darling Basin (SAMDB) Catchment boundary – study extent Spatially correct _Boundary ary Sub_Catchments GIS 2005 DWLBC Based on 1:50,000 topographic map –major tributary sub-catchments (mainly in lower reaches) Unknown Climate STILL TO BE SUPPLIED Excel SILO rainfall data Contours Contour10kMDB GIS 5m contour lines - incomplete in upper NE corner of catchment Contour50kMDB GIS 10m contour lines - incomplete in upper NE corner of catchment Contour250kMDB GIS 50m contour lines – complete catchment coverage DEMs DEM_15m GIS/raster 2007 PIRSA 15m Digital Elevation Model from local modelled data – incomplete in upper NE corner of catchment Accurate to 15m DEM_25m GIS/raster 2007 PIRSA 30m Digital Elevation Model from local modelled data – incomplete in upper NE corner of catchment Accurate to 25m DEM_90m GIS/raster 2007 USGS 90m GeoTiff data from Shuttle Radar Topography Mission (SRTM) converted to DEM - complete Accurate to 90m EPA_activiti EPA_licences GIS EPA database Accurate as at 23/5/07 es EPA_activities GIS EPA database Accurate as at 23/5/07 Farm_Dam Pool _dam_less_than_15m GIS Apr-03 DWLBC Aerial videography, desktop coding point feature Accurate within 10 -100m s (average <50m) Pool_dam_greater_than_15 GIS Apr-03 DWLBC Aerial videography, desktop coding line feature Accurate within 10 -100m m (average <50m) Farm_dam_intensity GIS 2004 DWLBC Estimation of intensity of water use based on water capture. Indicative of intensity of water Desk top aerial photography assessment of farm dam location (1999 data). capture Irrigation_bore_intensity GIS 2004 DWLBC Combined data from irrigation / bore usage and farm dam development (1999 data). Indicative of intensity of water use only Aerial (SEE SEPARATE CD) GIS/raster 2003- DEH Various aerial images – mostly concentrated along River Murray & Lakes Spatially correct Imagery 2005 Flow_gaugi Watercourse Gauging GIS Various DWLBC Field assessed Accurate to the GPS location ng Stations Stream_flow_data.xls Excel 2007 DWLBC Data extracted from DWLBC database Obtained May 2007 Land_Use MDB2003_Landuse GIS 2003 DWLBC Murray Darling Basin Land Use 2003 – ALUMv4 classified – see metadata pdf in folder 2003 STATE03_ALUMv6_GEO_ GIS 1999- DWLBC Clip from Statewide Land Use Dataset 2003 – ALUMv6 classified – see metadata pdf in folder 2003 GDA94 2003 PIRSA_Land_Use GIS 2006 PIRSA Murray Darling Basin Land Use – from PIRSA – simplified landuse layer Obtained 2006 Parcels Parcels GIS Feb-07 DEH/DAIS Land Parcels/Cadastre in SAMDB catchment Spatially correct Point_Sour Point_Source_Pollution GIS 2007 EPA Point source pollution from township wastewater treatment plants (EPA licensed) EPA data 31/5/07 ce_Pollutio n Point_Source_Pollution_W Excel 2007 EPA Point source pollution from township wastewater treatment plants (EPA licensed) EPA data 31/5/07 Q_Data.xls River_Murr River_Murray_Lakes GIS Unknown DEH River Murray & Lower Lakes boundaries (polygon layer) – Boundary for receiving water model Spatially correct ay MurrayFlood1956 GIS 1956 DEH River Murray & Lower Lakes 1956 Flood Boundary – delineates floodplain under major flood Spatially correct Roads Roads GIS Feb-07 DEH Roads in SAMDB catchment Soils (SEE SEPARATE CD) GIS DWLNC See DWLBC Land & Soil Spatial Data for Southern Australia CD – includes water holding capacity Data Current at Dec. 2005
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC DATA SUPPLIED A-2 Streams Streams_1st_2nd_order GIS Apr-03 DWLBC Aerial videography, desktop coding line feature extracted for 1st & 2nd order streams – should be used for Accurate within 10 -100m burn-in of DEM? (average <50m) Streams_3rd_order GIS Apr-03 DWLBC Aerial videography, desktop coding line feature extracted for ³ 3rd order streams - should be used for burn-in Accurate within 10 -100m of DEM? (average <50m) Streams GIS Feb-07 DWLBC Some gaps in line data but better coverage for northern region of catchment Accurate within 10 -100m (average <50m) Baseflow GIS Apr-03 DWLBC Aerial videography, desktop coding line feature Accurate within 10 -100m (average <50m) Streamworks GIS Apr-03 DWLBC Aerial videography, desktop coding point feature Accurate within 10 -100m (average <50m) Towns Towns GIS Unknown DWLBC Towns in SAMDB catchment Spatially correct Water MDB_Stream_WQ_Monitori GIS May-07 EPA EPA water quality monitoring sites spatial layer (for excel datafile in same folder) From EDMS database Quality ng_Sites 30/5/07 Murray_Darling_Basin_Stre Excel May-07 EPA EPA Water quality monitoring data for sites in GIS layer – note: some sites have limited data From EDMS database ams.xls 30/5/07
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC UNGROUPED LAND USE CLASSES B-1
APPENDIX B: UNGROUPED LAND USE CLASSES
Number of Classes - 102 1 Grazing Natural Vegetation 35 Rehabilitation 69 Irrigated legume/grass mixtures 2 Shrubland 36 Irrigated modified pastures 70 Irrigated hardwood production 3 Grazing modified pastures 37 Perennial horticulture 71 Hay and silage 4 Navigation and communication 38 Oleaginous fruits 72 Intensive animal production 5 Other minimal use 39 Residual native cover 73 Cattle 6 Woodland 40 Marsh/wetland – production 74 Pasture legumes 7 Urban residential 41 Effluent pond 75 Irrigated flowers and bulbs 8 Commercial services 42 River – intensive use 76 Irrigated plantation nurseries 9 Native/exotic pasture mosaic 43 Irrigated tree fruits 77 Vine fruits 10 Railways 44 Irrigated vine fruits 78 Irrigates shrub nuts fruits and berries 11 Public Services 45 Irrigated perennial horticulture 79 Tree fruits 12 Mallee 46 Irrigated seasonal horticulture 80 Hardwood production 13 Quarries 47 Irrigated oleaginous fruits 81 Marsh/wetland – intensive use 14 Cereals 48 Irrigated cropping 82 Glasshouses 15 Reservoir/dam 49 Irrigated vegetables and herbs 83 Sewage 16 Roads 50 Woody fodder plants 84 Tree nuts 17 Grassland 51 Ports and water transport 85 Mining 18 Legumes 52 Irrigated hay and silage 86 Dairy 19 Strict nature reserves 53 Waste treatment and disp9osal 87 Defence 20 Rural residential 54 Irrigated tree nuts 88 Irrigated oil seeds 21 Manufacturing and industrial 55 Irrigated pasture legumes 89 Intensive horticulture 22 Natural feature protection 56 Electricity 90 Pasture legume/grass mixtures generation/transmission 23 Solid garbage 57 Irrigated cereals 91 Lake – production 24 Irrigated sown grasses 58 Poultry 92 Irrigated legumes 25 Oil seeds 59 Irrigated plantation forestry 93 Traditional indigenous uses 26 Softwood production 60 Lake 94 Aquaculture 27 Other conserved area 61 Marsh/wetland – conservation 95 Sown grasses 28 Recreation and culture 62 Plantation forestry 96 Cropping 29 Airports/aerodromes 63 Habitat/species – management 97 Research facilities area 30 Services 64 Evaporation basin 98 Lake – saline 31 Marsh/wetland 65 Lake – conservation 99 Reservoir 32 Water storage – intensive 66 Shadehouses 100 National park use/farm dams 33 Pigs 67 Drainage channels/aqueduct 101 Landfill 34 Utilities 68 Irrigated woody fodder plants 102 Channel/aqueduct
G:\ADMIN\B16618.G.TRW_MURRAYLANDS_E2\R.B16618.001.02.DOC
BMT WBM Brisbane Level 11, 490 Upper Edward Street Brisbane 4000 PO Box 203 Spring Hill QLD 4004 Tel +61 7 3831 6744 Fax +61 7 3832 3627 Email [email protected] Web www.wbmpl.com.au
BMT WBM Denver 14 Inverness Drive East, #B132 Englewood Denver Colorado 80112 USA Tel +1 303 792 9814 Fax +1 303 792 9742 Email [email protected] Web www.wbmpl.com.au
BMT WBM Melbourne Level 5, 99 King Street Melbourne 3000 PO Box 604 Collins Street West VIC 8007 Tel +61 3 9614 6400 Fax +61 3 9614 6966 Email [email protected] Web www.wbmpl.com.au
BMT WBM Morwell Cnr Hazelwood Drive & Miners Way Morwell 3840 PO Box 888 Morwell VIC 3840 Tel +61 3 5135 3400 Fax +61 3 5135 3444 Email [email protected] Web www.wbmpl.com.au
BMT WBM Newcastle 126 Belford Street Broadmeadow 2292 PO Box 266 Broadmeadow NSW 2292 Tel +61 2 4940 8882 Fax +61 2 4940 8887 Email [email protected] Web www.wbmpl.com.au
BMT WBM Perth 1 Brodie Hall Drive Technology Park Bentley 6102 Tel +61 8 9328 2029 Fax +61 8 9486 7588 Email [email protected] Web www.wbmpl.com.au
BMT WBM Sydney Suite 206, 118 Great North Road Five Dock 2046 PO Box 129 Five Dock NSW 2046 Tel +61 2 9713 4836 Fax +61 2 9713 4890 Email [email protected] Web www.wbmpl.com.au
BMT WBM Vancouver 1190 Melville Street #700 Vancouver British Columbia V6E 3W1 Canada Tel +1 604 683 5777 Fax +1 604 608 3232 Email [email protected] Web www.wbmpl.com.au