Water Availability in the Barwon-Darling

Transcription

1 Water Availability in the Barwon-Darling A report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project June 2008

2 Murray-Darling Basin Sustainable Yields Project acknowledgments The Murray-Darling Basin Sustainable Yields project is being undertaken by CSIRO under the Australian Government's Raising National Water Standards Program, administered by the National Water Commission. Important aspects of the work were undertaken by Sinclair Knight Merz; Resource & Environmental Management Pty Ltd; Department of Water and Energy (New South Wales); Department of Natural Resources and Water (Queensland); Murray-Darling Basin Commission; Department of Water, Land and Biodiversity Conservation (South Australia); Bureau of Rural Sciences; Salient Solutions Australia Pty Ltd; ewater Cooperative Research Centre; University of Melbourne; Webb, McKeown and Associates Pty Ltd; and several individual sub-contractors. Murray-Darling Basin Sustainable Yields Project disclaimers Derived from or contains data and/or software provided by the Organisations. The Organisations give no warranty in relation to the data and/or software they provided (including accuracy, reliability, completeness, currency or suitability) and accept no liability (including without limitation, liability in negligence) for any loss, damage or costs (including consequential damage) relating to any use or reliance on that data or software including any material derived from that data and software. Data must not be used for direct marketing or be used in breach of the privacy laws. Organisations include: Department of Water, Land and Biodiversity Conservation (South Australia), Department of Sustainability and Environment (Victoria), Department of Water and Energy (New South Wales), Department of Natural Resources and Water (Queensland), Murray-Darling Basin Commission. CSIRO advises that the information contained in this publication comprises general statements based on scientific research. The reader is advised and needs to be aware that such information may be incomplete or unable to be used in any specific situation. No reliance or actions must therefore be made on that information without seeking prior expert professional, scientific and technical advice. To the extent permitted by law, CSIRO (including its employees and consultants) excludes all liability to any person for any consequences, including but not limited to all losses, damages, costs, expenses and any other compensation, arising directly or indirectly from using this publication (in part or in whole) and any information or material contained in it. Data is assumed to be correct as received from the Organisations. Citation CSIRO (2008). Water availability in the Barwon-Darling. A report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project. CSIRO, Australia. 106pp. Publication Details Published by CSIRO 2008 all rights reserved. This work is copyright. Apart from any use as permitted under the Copyright Act 1968, no part may be reproduced by any process without prior written permission from CSIRO. ISSN X Photo on cover: The Darling River near Bourke, NSW. Courtesy of the Murray-Darling Basin Commission.

3 Director s Foreword Following the November 2006 Summit on the Southern Murray-Darling Basin, the then Prime Minister and Murray-Darling Basin state Premiers commissioned CSIRO to report on sustainable yields of surface and groundwater systems within the Murray-Darling Basin. This report from the CSIRO Murray-Darling Basin Sustainable Yields Project details the assessments for one of 18 regions that encompass the Basin. The CSIRO Murray-Darling Basin Sustainable Yields Project is providing critical information on current and likely future water availability. This information will help governments, industry and communities consider the environmental, social and economic aspects of the sustainable use and management of the precious water assets of the Murray-Darling Basin. The project is the first rigorous attempt worldwide to estimate the impacts of catchment development, changing groundwater extraction, climate variability and anticipated climate change, on water resources at a basin-scale, explicitly considering the connectivity of surface and groundwater systems. To do this, we are undertaking the most comprehensive hydrologic modelling ever attempted for the entire Basin, using rainfall-runoff models, groundwater recharge models, river system models and groundwater models, and considering all upstream-downstream and surfacesubsurface connections. We are complementing this work with detailed surface water accounting across the Basin never before has surface water accounting been done in such detail in Australia, over such a large area, and integrating so many different data sources. To deliver on the project CSIRO is drawing on the scientific leadership and technical expertise of national and state government agencies in Queensland, New South Wales, Victoria, the Australian Capital Territory and South Australia, as well as the Murray-Darling Basin Commission and Australia s leading industry consultants. The project is dependent on the cooperative participation of over 15 government and private sector organisations contributing over 100 individuals. The project has established a comprehensive but efficient process of internal and external quality assurance on all the work performed and all the results delivered, including advice from senior academic, industry and government experts. The project is led by the Water for a Healthy Country Flagship, a CSIRO-led research initiative which was set up to deliver the science required for sustainable management of water resources in Australia. The Flagship goal is to achieve a tenfold increase in the social, economic and environmental benefits from water by By building the capacity and capability required to deliver on this ambitious goal, the Flagship is ideally positioned to accept the challenge presented by this complex integrative project. CSIRO has given the Murray-Darling Basin Sustainable Yields Project its highest priority. It is in that context that I am very pleased and proud to commend this report to the Australian Government. Dr Tom Hatton Director, Water for a Healthy Country National Research Flagships CSIRO

4

5 Executive Summary Background The CSIRO Murray-Darling Basin Sustainable Yields Project is providing governments with a robust estimate of water availability for the entire Murray-Darling Basin (MDB) on an individual catchment and aquifer basis, taking into account climate change and other risks. This report describes the assessment undertaken for the Barwon-Darling region. While key aspects of the assessment and modelling methods used in the project are contained in this report, fuller methodological descriptions will be provided in a series of project technical reports. The Barwon-Darling region is in northwestern New South Wales and covers 13 percent of the total area of the MDB. The region is based around the Barwon and Darling rivers. The population is 50,000 or 2.5 percent of the MDB total, concentrated in the centres of Collarenebri, Walgett, Brewarrina, Bourke, Cobar and Wilcannia. The major land use is dryland pasture used for beef and sheep grazing. Approximately 63,000 ha of land were irrigated in 2000 including 57,900 ha for cotton production on the western plains. The Talyawalka area is a nationally important wetland located on the southern boundary of the region on the Darling Riverine Plains between Wilcannnia and Menindee. The area comprises the wetlands of the Talyawalka Anabranch of the Darling River and its distributary, Teryawynia Creek. It is representative of a semi-arid inland floodplain wetland system fringed by Black Box woodland. The Paroo Overflow Lakes, although located within the Barwon-Darling region, are described and assessed in the project report for the Paroo region as they depend on flow from the Paroo River. Executive Summary The region uses 3 percent of the total surface water diverted for irrigation in the MDB and groundwater use is less than 1 percent of the MDB total. The Barwon and Darling rivers are unregulated but flows are influenced by development and large major public water storages on upstream tributary rivers. There are no major public water storages within the region. Key Messages The key messages relating to climate, surface water resources, groundwater and the environment are presented below for scenarios of current and possible future conditions. The scenarios assessed are defined in Chapter 1. Scenario A is the baseline for comparison with all other scenarios. Historical climate and current development (Scenario A) The annual rainfall and modelled runoff averaged over the Barwon-Darling region are 328 mm and 6 mm respectively. Rainfall is low throughout the year but highest in summer and runoff is highest in summer and early autumn. The region generates 2.8 percent of the total MDB runoff. Current average surface water availability for the entire Darling Basin (assessed at Bourke) is 3515 GL/year and 99 percent of this water is generated in regions upstream of the Barwon-Darling region (including 23 percent from the Namoi, 22 percent from the Macquarie-Castlereagh and 20 percent from the Border Rivers). Current average surface water use in the Barwon-Darling region is 230 GL/year. Licences within the region are fully utilised. Note however, access rules for the Barwon-Darling irrigators were altered in July 2006 to cap use at 173 GL/year. Current total average surface water use across the entire Darling Basin reduces streamflow at Bourke by 1365 GL/year. The relative level of use for the Darling Basin is thus 39 percent. This is a high level of development. Water resource development across the Darling Basin has not significantly altered the seasonality of streamflow in the Barwon-Darling region but has reduced the magnitude of two-year average return interval floods by 41 percent, and the magnitude of five- and ten-year average return interval floods by around 30 percent. These reductions in flood magnitude are considerably greater than the reductions likely under the best estimate 2030 climate. Groundwater use within the region in 2004/05 is estimated at 10 GL (4 percent of total within-region water use), excluding use of the confined aquifers of the Great Artesian Basin (GAB). Groundwater extraction for each of the nine groundwater management units (GMUs) assessed in the region is less than 10 percent of rainfall recharge. This is a low level of groundwater development which poses no major concerns. CSIRO 2008 June 2008 Water availability in the Barwon-Darling i

6 The reduction in streamflow due to current groundwater extraction is around 1.3 GL/year. A more complete analysis (outside the scope of this project) could account for flood recharge as an additional water source to the shallow alluvial aquifers. Water resource development in this and upstream regions has nearly doubled the average and maximum periods between substantial flows to the Talyawalka Anabranch system. The average period between substantial inflows is now 4 years and the maximum period is 19 years. Individual events are now larger on average but the total volume is lower because there are far fewer flow events. Water resource development has more than doubled the average period between events that drown out Bourke Weir (thereby allowing fish passage). Individual drown-out events at the weir are slightly larger and last slightly longer. Recent climate and current development (Scenario B) Executive Summary The average annual rainfall and runoff over the ten-year period 1997 to 2006 are 3 and 8 percent higher respectively than the long-term (1895 to 2006) average values. However, because of the inter-annual variability and the relatively short ten-year period used as the basis for comparison, the 1997 to 2006 rainfall and runoff are not significantly different to the long term mean values. Similarly, in the upstream contributing regions from whence most of the flow is sourced, the rainfall and runoff over the last ten years are not significantly different to the long-term average values. A scenario based on the last ten years was therefore not modelled for the region. Future climate and current development (Scenario C) Under the best estimate (median) 2030 climate average annual runoff would be reduced by 2 percent. The extreme estimates (which come from the high global warming scenario) range from a 22 percent reduction to a 50 percent increase in average annual runoff. The range from the low global warming scenario is from an 8 percent reduction to a 12 percent increase. However, due to changed runoff in contributing upstream regions, average surface water availability (assessed at Bourke) would, under the best estimate 2030 climate, be reduced by 8 percent and end-of-system flows would be reduced by 10 percent. Total average surface water use within the region would increase by 2 percent due to increased evaporation from on-farm storages. The impacts of climate change vary between water products: water use under Class A, B and C licences would increase by 11, 2 and less than 1 percent, respectively. Floodplain harvesting would reduce by 2 percent. The relative level of use for the entire Darling Basin would then be a very high 41 percent. Under the wet extreme 2030 climate, average surface water availability would increase by 31 percent, surface water use within the region would increase by 3 percent and end-of-system flows would increase by 47 percent. Under the dry 2030 climate extreme, average surface water availability would decrease by 27 percent, surface water use within the region would increase by 5 percent and end-of-system flows would decrease by 35 percent. Groundwater recharge would fall slightly under the best estimate 2030 climate but not enough to significantly change the ratio of extraction to recharge. Under the best estimate 2030 climate Talyawalka Anabranch system inflow events would be slightly more frequent but smaller in volume. The average period between inflows and the individual event volume would be reduced by 8 and 18 percent, respectively. The average period between drown-outs at Bourke Weir would increase by 18 percent. The average event volume at the weir would be unaffected. Under the dry extreme 2030 climate there would be large changes to flows to the Talyawalka Anabranch system. The average period between inflows would increase by over 30 percent and the average volume of these events would be reduced by more than 30 percent. The average period between drown-outs at Bourke Weir would be increased by 66 percent and the average event volume would be reduced by 10 percent. The ecological consequences of these changes would be likely to be severe. Under the wet extreme 2030 climate Talyawalka Anabranch system inflows would be similar to without-development conditions. Drown-outs at Bourke Weir would be more frequent. Event volumes at the weir would be slightly smaller. ii Water availability in the Barwon-Darling June 2008 CSIRO 2008

7 Future climate and future development (Scenario D) There are no commercial forestry plantations in the region and none are projected for the future. The total farm dam storage volume is projected to increase by 13.2 GL (14 percent) by ~2030. The projected increase in farm dams will reduce mean annual runoff by less than 0.5 percent. Projected 2030 development (groundwater extraction and additional farm dams in upstream regions) would reduce inflows to the Barwon-Darling region (under the best estimate 2030 climate) by 3 percent or 78 GL/year on average. Additional farm dams and groundwater extraction would be responsible for about equal shares of this impact. Diversions would be reduced by 1 percent compared to current conditions and end-of-system flows would be reduced by a 3 percent in addition to best estimate climate change impacts. The relative level of use for the entire Darling Basin would then be 42 percent. Regional groundwater extraction is projected to increase 24-fold to 240 GL/year by 2030, moving groundwater use from 4 percent to just over 50 percent of the total average annual water use in the region. The ratio of extraction of recharge would similarly rise significantly, although the rate of growth is highly uncertain. The greatest estimated increase would occur in the Gunnedah Basin portion of the region, in which extraction would then be close to 50 percent of recharge (a medium level of development) under the best estimate 2030 climate. Growth in extraction in the GAB Alluvial, Western Murray Porous Rock and Kanmantoo Fold Belt GMUs also lead to a medium level of development (extraction exceeding 30 percent of recharge) under the best estimate 2030 climate. The Lower Darling Alluvium GMU would only reach this level of development under reduced recharge of the dry 2030 climate extreme. Projected groundwater extraction would reduce streamflow by 37 GL/year once equilibrium conditions are reached. Future development would have minor additional effects on the hydrology of the Talyawalka Anabranch system and the frequency and magnitude of Bourke Weir drown-outs. Executive Summary Uncertainty The largest sources of uncertainty for future climate results are the climate change projections (global warming level) and the modelled implications of global warming on regional rainfall. The results from 15 global climate models were used but there are large differences amongst these models in terms of regional rainfall predictions. There are also considerable uncertainties associated with the future projections of farm dams and commercial forestry plantations in the upstream regions which impact on future flows in the region. Future developments could differ considerably from these projections if governments were to impose different policy controls. The river model generally reproduces observed streamflow patterns well and produces estimates that agree reasonably well with water accounts. The projected changes in flows due to future climate are greater than model uncertainty under the wet extreme future climate scenarios, but are similar to model uncertainty under the dry extreme and best estimate future climate scenarios, partly due to the model bias in some reaches. The model provides strong evidence of changes in flow pattern due to prior development, but the projected changes due to future development are very small. While the model is well suited for the purpose of this project, changes in low flows are not simulated well by the model. A simple water balance approach was used to assess the region s groundwater. This is appropriate given the low priority of the GMUs in the context of the project. However, the approach would not be adequate for addressing local groundwater management issues. The estimated impacts of groundwater extraction on streamflow are assigned a low level of confidence. The estimates are sensitive to the connectivity factor used and may be overestimated given the methodology. The environmental assessments of this project only consider a subset of the important assets for this region and are based on limited hydrology parameters with no direct quantitative relationships for environmental responses. Considerably more detailed investigation is required to provide the necessary information for informed management of the environmental assets of the region. CSIRO 2008 June 2008 Water availability in the Barwon-Darling iii

8 Table of Contents 1 Introduction Background Project methodological framework Climate and development scenarios Rainfall-runoff modelling River system modelling Monthly water accounts Groundwater modelling Environmental assessment References Overview of the region The region Environmental description Surface water resources Groundwater References Rainfall-runoff modelling Summary Modelling approach Modelling results Discussion of key findings References River system modelling Summary Modelling approach Modelling results Discussion of key findings References Uncertainty in surface water modelling results Summary Approach Results Discussion of key findings References Groundwater assessment Summary Approach Groundwater management units Groundwater levels Surface groundwater connectivity Water balance Discussion of key findings References Environment Summary Approach Results Discussion of key findings References...91 Appendix A Rainfall-runoff results for all subcatchments Appendix B River modelling reach mass balances Appendix C River system model uncertainty assessment by reach Water availability in the Barwon-Darling June 2008 CSIRO 2008

9 Tables Table 1-1. River system models in the Murray-Darling Basin...7 Table 2-1. Summary of land use in the year 2000 within the Barwon-Darling region...17 Table 2-2. Ramsar wetlands and wetlands of national importance located within the Barwon-Darling region. Bold text indicates that the wetland is a part of the assessed Talyawalka Anabranch and Teryawynia Creek system Table 2-3. Categorisation of groundwater management units, including extraction, entitlement and recharge details...24 Table 2-4. Summary of groundwater management plans...25 Table 3-1. Summary results under the 45 Scenario C simulations (numbers show percent change in mean annual rainfall and runoff under Scenario C relative to Scenario A)...32 Table 3-2. Water balance over the entire region by scenario...34 Table 4-1. Storages in the river system model...42 Table 4-2. Modelled water use configuration...42 Table 4-3. Model water management...42 Table 4-4. Model setup information...43 Table 4-5. River system model average annual water balance under scenarios O, P, A0, A, C and D...44 Table 4-6. Annual water availability under scenarios A and C (assessed for without-development conditions, which for Scenario A is synonymous with Scenario P)...45 Table 4-7. Contribution of each region to water availability under Scenario A...46 Table 4-8. Change in total diversions in each subcatchment under scenario C and D relative to Scenario A...47 Table 4-9. Relative level of use under scenarios A, C and D...50 Table Indicators of use during dry periods under scenarios A, C and D...51 Table Average reliability of water products under scenarios C and D relative to Scenario A...52 Table Daily flow event frequency at Bourke gauge (425003) under scenarios P, A, C and D...54 Table Percentage of time flow occurs at the end-of-system under scenarios P, A, C and D...55 Table Relative level of available water not diverted for use for the entire Darling Basin aggregated to Bourke under scenarios A, C and D...55 Table 5-1. Framework for considering implications of assessed uncertainties...60 Table 5-2. Comparison of water accounting reaches with subcatchment codes used in the river model...60 Table 5-3. Some characteristics of the gauging network of the Barwon-Darling region (142,173 km 2 ) compared with the entire Murray-Darling Basin (1,062,443 km 2 )...63 Table 5-4. Streamflow gauging stations for which data were used in Barwon-Darling IQQM calibration...66 Table 5-5. Regional water balance modelled and estimated on the basis of water accounting...69 Table 6-1. Categorisation of groundwater management units in the region, including annual extraction, entitlement and recharge details...75 Table 6-2. Estimated current and future groundwater extraction for the assessed groundwater management units in the Barwon- Darling region...80 Table 6-3. Summary results from the 45 Scenario C simulations. Numbers show percent change in mean annual rainfall and recharge under Scenario C relative to Scenario A. Those in bold type have been selected for further modelling...82 Table 6-4. Change in mean annual recharge for groundwater management units in the Barwon-Darling region under Scenario C relative to Scenario A...82 Table 6-5. Scaled recharge for assessed groundwater management units under scenarios A and C...83 Table 6-6. Comparison of current groundwater extraction with scaled rainfall recharge for assessed groundwater management units under scenarios A and C...83 Table 6-7. Comparison of future groundwater extraction with scaled rainfall recharge for assessed groundwater management units under Scenario D...84 Table 6-8. Surface groundwater connectivity showing an estimate of the volumetric impact extraction has on streamflow...84 Table 7-1. Definition of environmental indicators...90 Table 7-2. Environmental indicator values under scenarios P and A, and percent change (from Scenario A) in indicator values under scenarios C and D...90 Figures Figure 1-1. Region by region map of the Murray-Darling Basin...2 Figure 1-2. Methodological framework for the Murray-Darling Basin Sustainable Yields Project...3 Figure 1-3. Timeline of groundwater use and resultant impact on river...8 Figure annual and monthly rainfall averaged over the region. The curve on the annual graph shows the low frequency variability...15 CSIRO 2008 June 2008 Water availability in the Barwon-Darling

10 Figure 2-2. Map of dominant land uses of the Barwon-Darling region with inset showing the region s location within the Murray-Darling Basin. The map only shows wetlands that are assessed in the project (Chapter 7). A full list of key assets associated with the region is in Table Figure 2-3. Historical surface water diversions...20 Figure 2-4. Generalised north-west south-east cross-section of the hydrogeology of the Barwon-Darling region; from Evans et al. (1994)...21 Figure 2-5. Map of groundwater management units within the Barwon-Darling region...23 Figure 3-1. Map of the modelling subcatchments. The calibration catchments are not shown as they are outside the region...28 Figure 3-2. Modelled and observed monthly runoff and daily flow duration curve for the calibration catchments...29 Figure 3-3. Spatial distribution of mean annual rainfall and modelled runoff averaged over Figure annual rainfall and modelled runoff averaged over the region (the curve shows the low frequency variability)...30 Figure 3-5. Mean monthly rainfall and modelled runoff (averaged over for the region)...31 Figure 3-6. Percent change in mean annual runoff under the 45 Scenario C simulations (15 global climate models and three global warming scenarios) relative to Scenario A runoff...32 Figure 3-7. Mean annual rainfall and modelled runoff under scenarios A, Cdry, Cmid and Cwet...33 Figure 3-8. Mean monthly rainfall and modelled runoff under scenarios A and C averaged over across the region (C range is based on the consideration of each month separately the lower and upper limits in C range are therefore not the same as scenarios Cdry and Cwet)...35 Figure 3-9. Daily flow duration curves under scenarios A and C averaged over the region (C range is based on the consideration of each rainfall and runoff percentile separately the lower and upper limits in C range are therefore not the same as scenarios Cdry and Cwet)...35 Figure 4-1. River system map showing subcatchments, inflow and demand nodes, links and gauge locations...41 Figure 4-2. Transect of total river flow under scenarios A and C...45 Figure 4-3. Water availability under Scenario A (assessed for without-development conditions)...46 Figure 4-4. Time series of change in total surface water availability under Scenario C relative to Scenario A (assessed for without-development conditions)...47 Figure 4-5. Total average annual diversions for subcatchments under (a) scenarios A and C and (b) scenarios A and D...48 Figure 4-6. Total diversions under (a) Scenario A and difference between total water use under (b) Scenario Cwet, (c) Scenario Dwet, (d) Scenario Cmid, (e) Scenario Dmid, (f) Scenario Cdry and (g) Scenario Ddry...49 Figure 4-7. Reliability of Class A access for irrigators under scenarios A, C and D...52 Figure 4-8. Daily flow duration curves at Bourke gauge (425003) under scenarios P, A, C and D...53 Figure 4-9. Average monthly flow at Bourke gauge (425003) under scenarios P, A, C and D...53 Figure Daily flow duration curves for the combined end-of-system flow at Menindee under scenarios P, A, C and D...54 Figure Seasonal flow curves for the combined end-of-system flows at Menindee under scenarios P, A, C and D...54 Figure Comparison of diverted and non-diverted shares of water for the entire Darling Basin aggregated to Bourke under scenarios P, A, C and D...56 Figure 5-1. Map showing the subcatchments used in modelling, accounting reaches and contributing catchments. Black dots and red lines are nodes and links in the river model respectively...61 Figure 5-2. Map showing the rainfall, streamflow and evaporation observation network, along with the subcatchments used in modelling...64 Figure 5-3. The fraction of inflows/gains, outflows/losses and the total of water balance components that is (a) gauged or (b) could be attributed in the water accounts...68 Figure 5-4. Changes in the model efficiency (the performance of the river model in explaining observed streamflow patterns) along the length of the river (numbers refer to reach)...70 Figure 5-5. Pattern along the river of the ratio of the projected change over the river model uncertainty under scenarios P, C and D modelled for (a) monthly and (b) annual flows...71 Figure 6-1. Map of groundwater management units and key observation bores in the region...75 Figure 6-2. Hydrographs for Bores GW and GW monitoring the upper and lower aquifers respectively of the Upper Darling Alluvium near Bourke. These show a response to flood events in the upper aquifer with a lesser response in the lower aquifer...76 Figure 6-3. Hydrographs for Bores GW and GW monitoring the Upper and Lower aquifers respectively of the Upper Darling Alluvium near Wilcannia. These show a slightly downward trend with a strong similarity between the upper and lower aquifers...77 Figure 6-4. Hydrograph for Bore GW monitoring the Narrabri Formation near the Castlereagh-Barwon Junction. This hydrograph shows a response to flood events...77 Figure 6-5. Map of surface groundwater connectivity showing losing and gaining river reaches...78 Figure 6-6. Comparison of groundwater levels and surface water height for the Lower Darling River at Louth showing low to medium losing conditions...79 Figure 6-7. Percent change in mean annual recharge from the 45 Scenario C simulations (15 GCMs and three global warming scenarios) relative to Scenario A recharge...81 Figure 7-1. Location map of environmental assets assessed in this chapter...88 Figure 7-2. Satellite image of Talyawalka Lakes, showing in yellow the areas included for this site listed in the Directory of Important Wetlands in Australia (Environment Australia, 2001)...89 Water availability in the Barwon-Darling June 2008 CSIRO 2008

11 1 Introduction 1.1 Background Australia is the driest inhabited continent on Earth, and in many parts of the country including the Murray-Darling Basin water for rural and urban use is comparatively scarce. Into the future, climate change and other risks (including catchment development) are likely to exacerbate this situation and hence improved water resource data, understanding and planning and management are of high priority for Australian communities, industries and governments. On 7 November, 2006, the then Prime Minister of Australia met with the First Ministers of Victoria, New South Wales, South Australia and Queensland at a water summit focussed primarily on the future of the Murray-Darling Basin (MDB). As an outcome of the Summit on the Southern Murray-Darling Basin, a joint communiqué called for CSIRO to report progressively by the end of 2007 on sustainable yields of surface and groundwater systems within the MDB, including an examination of assumptions about sustainable yield in light of changes in climate and other issues. The subsequent Terms of Reference for what became the Murray-Darling Basin Sustainable Yields Project specifically asked CSIRO to: 1 Introduction estimate current and likely future water availability in each catchment and aquifer in the MDB considering: o climate change and other risks o surface groundwater interactions compare the estimated current and future water availability to that required to meet the current levels of extractive use. The Murray-Darling Basin Sustainable Yields Project is reporting progressively on each of 18 contiguous regions that comprise the entire MDB. These regions are primarily the drainage basins of the Murray and the Darling rivers Australia s longest inland rivers, and their tributaries. The Darling flows southwards from southern Queensland into New South Wales west of the Great Dividing Range into the Murray River in southern New South Wales. At the South Australian border the Murray turns southwesterly eventually winding to the mouth below the Lower Lakes and the Coorong. The regions for which the project assessments are being undertaken and reported are the Paroo, Warrego, Condamine-Balonne, Moonie, Border Rivers, Gwydir, Namoi, Macquarie-Castlereagh, Barwon-Darling, Lachlan, Murrumbidgee, Murray, Ovens, Goulburn-Broken, Campaspe, Loddon-Avoca, Wimmera and Eastern Mount Lofty Ranges (see Figure 1-1). CSIRO 2008 June 2008 Water availability in the Barwon-Darling 1

12 1 Introduction Figure 1-1. Region by region map of the Murray-Darling Basin The Murray-Darling Basin Sustainable Yields Project will be the most comprehensive MDB-wide assessment of water availability undertaken to-date. For the first time: daily rainfall-runoff modelling has been undertaken at high spatial resolution for a range of climate change and development scenarios in a consistent manner for the entire MDB the hydrologic subcatchments required for detailed modelling have been precisely defined across the entire MDB the hydrologic implications for water users and the environment by 2030 of the latest Intergovernmental Panel on Climate Change climate projections, the likely increases in farm dams and commercial forestry plantations and the expected increases in groundwater extraction have been assessed in detail (using all existing river system and groundwater models as well new models developed within the project) river system modelling has included full consideration of the downstream implications of upstream changes between multiple models and between different States, and quantification of the volumes of surface groundwater exchange detailed analyses of monthly water balances for the last ten to twenty years have been undertaken using available streamflow and diversion data together with additional modelling including estimates of wetland evapotranspiration and irrigation water use based on remote sensing imagery (to provide an independent crosscheck on the performance of river system models). 2 Water availability in the Barwon-Darling June 2008 CSIRO 2008

13 The successful completion of these outcomes, among many others, relies heavily on a focussed collaborative and teamoriented approach between CSIRO, State government natural resource management agencies, the Murray-Darling Basin Commission, the Bureau of Rural Sciences, and leading consulting firms each bringing their specialist knowledge and expertise on the MDB to the project. 1.2 Project methodological framework The methodological framework for the project is shown in Figure 1-2. This also indicates in which chapters of this report the different aspects of the project assessments and results are presented. 1 Introduction Figure 1-2. Methodological framework for the Murray-Darling Basin Sustainable Yields Project The first steps in the sequence of the project are definition of the reporting regions and their composite subcatchments, and definition of the climate and development scenarios to be assessed (including generation of the time series of climate data that describe these scenarios). The second steps are rainfall-runoff modelling and rainfall-recharge modelling for which the inputs are the climate data for the different scenarios. Catchment development scenarios for farm dams and commercial forestry plantations are modifiers of the modelled runoff time series. Next, the runoff implications are propagated through river system models and the recharge implications propagated through groundwater models for the major groundwater resources or considered in simpler assessments for minor groundwater resources. The connectivity of surface and groundwater is assessed and the actual volumes of surface groundwater exchange under current and likely future groundwater extraction are quantified. Uncertainty levels of the river system models are then assessed based on monthly water accounting. The results of scenario outputs from the river system model are used to make limited hydrological assessments of ecological relevance to key environmental assets. Finally, the implications of the scenarios for water availability and water use under current water sharing arrangements are assessed, synthesised and reported. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 3

14 1.3 Climate and development scenarios The project is assessing the following four scenarios of historical and future climate and current and future development, all of which are defined by daily time series of climate variables based on different scalings of the historical 1895 to 2006 climate sequence: historical climate and current development recent climate and current development future climate and current development future climate and future development. These scenarios are described in some detail below with full details provided in Chiew et al. (2008a). 1 Introduction Historical climate and current development Historical climate and current development referred to as Scenario A is the baseline against which other climate and development scenarios are compared. The historical daily rainfall time series data that are used are taken from the SILO Data Drill of the Queensland Department of Natural Resources and Water database which provides data for a 0.05 o x 0.05 o (5 km x 5 km) grid across the continent (Jeffrey et al., 2001; and Areal potential evapotranspiration (PET) data are calculated from the SILO climate surface using Morton s wet environment evapotranspiration algorithms ( and Chiew and Leahy, 2003). Current development for the rainfall-runoff modelling is the average of 1975 to 2005 land use and small farm dam conditions. Current development for the river system modelling is the dams, weirs and licence entitlements in the latest State agency models, updated to 2005 levels of large farm dams. Current development for groundwater models is 2004 to 2005 levels of licence entitlements. Surface groundwater exchanges in the river and groundwater models represent an equilibrium condition for the above levels of surface and groundwater development Recent climate and current development Recent climate and current development referred to as Scenario B is used for assessing future water availability should the climate in the future prove to be similar to that of the last ten years. Climate data for 1997 to 2006 is used to generate stochastic replicates of 112-year daily climate sequences. The replicate which best produces a mean annual runoff value closest to the mean annual runoff for the period 1997 to 2006 is selected to define this scenario. Scenario B is only analysed and reported upon where the mean annual runoff for the last ten years is statistically significantly different to the long-term average Future climate and current development Future climate and current development referred to as Scenario C is used to assess the range of likely climate conditions around the year Three global warming scenarios are analysed in 15 global climate models (GCM) to provide a spectrum of 45 climate variants for the The scenario variants are derived from the latest modelling for the fourth assessment report of the Intergovernmental Panel on Climate Change (IPCC, 2007). Two types of uncertainties in climate change projections are therefore taken into account: uncertainty in global warming mainly due to projections of greenhouse gas emissions and global climate sensitivity to the projections; and uncertainty in GCM modelling of climate over the MDB. Results from each GCM are analysed separately to estimate the change per degree global warming in rainfall and other climate variables required to calculate PET. The change per degree of global warming is then scaled by a high, medium and low global warming by 2030 relative to 1990 to obtain the changes in the climate variables for the high, medium and low global warming scenarios. The future climate and current development Scenario C considerations are therefore for 112-year rainfall and PET series for a greenhouse enhanced climate around 2030 relative to 1990 and not for a forecast climate at Water availability in the Barwon-Darling June 2008 CSIRO 2008

15 The method used to obtain the future climate and current development Scenario C climate series also takes into account different changes in each of the four seasons as well as changes in the daily rainfall distribution. The consideration of changes in the daily rainfall distribution is important because many GCMs indicate that extreme rainfall in an enhanced greenhouse climate is likely to be more intense, even in some regions where projections indicate a decrease in mean seasonal or annual rainfall. As the high rainfall events generate large runoff, the use of traditional methods that assumes the entire rainfall distribution to change in the same way will lead to an underestimation of mean annual runoff in regions where there is an increase, and an overestimation of the decrease in mean annual runoff where there is a decrease (Chiew, 2006). All 45 future climate and current development Scenario C variants are used in rainfall-runoff modelling; however, three variants a dry, a mid (best estimate median) and a wet variant are presented in more detail and are used in river and groundwater modelling Future climate and future development Future climate and future development referred to as Scenario D considers the dry, mid and wet climate variants from the future climate and current development Scenario C together with likely expansions in farm dams and commercial forestry plantations and the changes in groundwater extractions anticipated under existing groundwater plans. 1 Introduction Farm dams here refer only to dams with their own water supply catchment, not those that store water diverted from a nearby river, as the latter require licences and are usually already included within existing river system models. A 2030 farm dam development scenario for the MDB has been developed by considering current distribution and policy controls and trends in farm dam expansion. The increase in farm dams in each subcatchment is estimated using simple regression models that consider current farm dam distribution, trends in farm dam (Agrecon, 2005) or population growth (Australian Bureau of Statistics, 2004; and Victorian Department of Sustainability and Enviroment (DSE), 2004) and current policy controls (Queensland Government, 2000; New South Wales Government, 2000; Victoria Government, 1989; South Australia Government, 2004). Data on the current extent of farm dams is taken from the 2007 Geosciences Australia Man-made Hydrology GIS coverage and from the 2006 VicMap 1:25,000 topographic GIS coverage. The former covers the eastern region of Queensland MDB and the northeastern and southern regions of the New South Wales MDB. The latter data covers the entire Victorian MDB. A 2030 scenario for commercial forestry plantations for the MDB has been developed using regional projections from the Bureau of Rural Sciences which takes into account trends, policies and industry feedbacks. The increase in commercial forestry plantations is then distributed to areas adjacent to existing plantations (which are not natural forest land use) with the highest biomass productivity estimated from the PROMOD model (Battaglia and Sands, 1997). Growth in groundwater extractions has been considered in the context of existing groundwater planning and sharing arrangements and in consultation with State agencies. For groundwater the following issues have been considered: growth in groundwater extraction rates up to full allocation improvements in water use efficiency due to on-farm changes and lining of channels water buy-backs. 1.4 Rainfall-runoff modelling The adopted approach provides a consistent way of modelling historical runoff across the MDB and assessing the potential impacts of climate change and development on future runoff. The lumped conceptual daily rainfall-runoff model, SIMHYD, with a Muskingum routing method (Chiew et al., 2002; Tan et al., 2005), is used to estimate daily runoff at 0.05 o grids (~ 5 km x 5 km) across the entire MDB for the four scenarios. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 5

16 The model is calibrated against 1975 to 2006 streamflow data from about 200 unregulated catchments of 50 km 2 to 2000 km 2 across the MDB (calibration catchments). Although unregulated, streamflow in these catchments for the calibration period may in fact reflect some low levels of water diversion and the effects of historical land use change. The calibration period is a compromise between a shorter period that would better represent current development and a longer period that would better account for climatic variability. In the model calibration, the six parameters in SIMHYD are optimised to maximise an objective function that incorporates the Nash-Sutcliffe efficiency (Nash and Sutcliffe, 1970) of monthly runoff and daily flow duration curve, together with a constraint to ensure that the total modelled runoff over the calibration period is within 5 percent of the total recorded runoff. The resulting optimised model parameters are therefore identical for all cells within a calibration catchment. 1 Introduction The runoff for non-calibration catchments is modelled using optimised parameter values from the geographically closest calibration catchment, provided there is a calibration catchment point within 250 km. Once again the parameter values for each grid cell within a non-calibration catchment are identical. For catchments more than 250 km from a calibration catchment default parameter values are used. The default parameter values are taken from the entire MDB modelling run (identical parameters across the entire MDB are chosen to ensure a realistic runoff gradient across the drier parts of the MDB) which best matched observed flows at calibration points. The places these default values are used are therefore all areas of very low runoff. As the parameter values come from calibration against streamflow from 50 km 2 to 2000 km 2 catchments, the runoff defined here is different, and can be much higher, than streamflow recorded over very large catchments where there can be significant transmission losses (particularly in the western and northwestern parts of the MDB). Almost all of the catchments available for model calibration are in the higher runoff areas in the eastern and southern parts of the MDB. Runoff estimates are therefore generally more accurate in the eastern and southern parts of the MDB and are comparatively poor elsewhere. The same model parameter values are used for all the simulations. The future climate Scenario C simulations therefore do not take into account the effect on forest water use of global warming and enhanced atmospheric CO 2 concentrations. There are compensating positive and negative global warming impacts on forest water use, and it is difficult to estimate the net effect because of the complex climate-biosphere-atmosphere interactions and feedbacks. This is discussed in Marcar et al. (2006) and in Chiew et al. (2008b). Bushfire frequency is also likely to increase under the future climate Scenario C. In local areas where bushfires occur, runoff would reduce significantly as forests regrow. However, the impact on runoff averaged over an entire reporting region is unlikely to be significant (see Chiew et al., 2008b). For Scenario D (future climate and future development scenario) the impact of additional farm dams on runoff is modelled using the CHEAT model (Nathan et al., 2005) which takes into account rainfall, evaporation, demands, inflows and spills. The impact of additional plantations on runoff is modelled using the FCFC model (Forest Cover Flow Change) (Brown et al, 2006; The rainfall-runoff model SIMHYD is used because it is simple and has relatively few parameters and, for the purpose of this project, provides a consistent basis (that is automated and reproducible) for modelling historical runoff across the entire MDB and for assessing the potential impacts of climate change and development on future runoff. It is possible that, in data-rich areas, specific calibration of SIMHYD or more complex rainfall-runoff models based on expert judgement and local knowledge as carried out by some state agencies would lead to better model calibration for the specific modelling objectives of the area. Chiew et al. (2008b) provide a more detailed description of the rainfall-runoff modelling, including details of model calibration, cross-verification and regionalisation with both the SIMHYD and Sacramento rainfall-runoff models and simulation of climate change and development impacts on runoff. 6 Water availability in the Barwon-Darling June 2008 CSIRO 2008

17 1.5 River system modelling The project is using river system models that encapsulate descriptions of current infrastructure, water demands, and water management and sharing rules to assess the implications of the changes in inflows described above on the reliability of water supply to users. Given the time constraints of the project and the need to link the assessments to State water planning processes, it is necessary to use the river system models currently used by State agencies, the Murray-Darling Basin Commission and Snowy Hydro Ltd. The main models in use are IQQM, REALM, MSM-Bigmod, WaterCress and a model of the Snowy Mountains Hydro-electric Scheme. The modelled runoff series from SIMHYD are not used directly as subcatchment inflows in these river system models because this would violate the calibrations of the river system models already undertaken by State agencies to different runoff series. Instead, the relative differences between the daily flow duration curves of the historical climate Scenario A and the remaining scenarios (scenarios B, C and D respectively) are used to modify the existing inflows series in the river system models (separately for each season). The scenarios B, C and D inflow series for the river system modelling therefore have the same daily sequences but different amounts as the Scenario A river system modelling series. Table 1-1. River system models in the Murray-Darling Basin Model Description Rivers modelled 1 Introduction IQQM Integrated Quantity-Quality Model: hydrologic modelling tool developed by the NSW Government for use in planning and evaluating water resource management policies. Paroo, Warrego, Condamine-Balonne (Upper, Mid, Lower), Nebine, Moonie, Border Rivers, Gwydir, Peel, Namoi, Castlereagh, Macquarie, Marthaguy, Bogan, Lachlan, Murrumbidgee, Barwon-Darling REALM Resource Allocation Model: water supply system simulation tool package for modelling water supply systems configured as a network of nodes and carriers representing reservoirs, demand centres, waterways, pipes, etc. Ovens (Upper, Lower), Goulburn, Wimmera, Avoca, ACT water supply. MSM-BigMod Murray Simulation Model and the daily forecasting model Murray BigMod: purpose-built by the Murray-Darling Basin Commission to manage the Murray River system. MSM is a monthly model that includes the complex Murray accounting rules. The outputs from MSM form the inputs to BigMod, which is the daily routing engine that simulates the movement of water. WaterCress Water Community Resource Evaluation and Simulation System: PC-based water management platform incorporating generic and specific hydrological models and functionalities for use in assessing water resources and designing and evaluating water management systems. Eastern Mt Lofty Ranges (six separate catchments) SMHS Snowy Mountains Hydro-electric Scheme model: purpose built by Snowy Hydro Ltd to guide the planning and operation of the SMHS. Snowy Mountains Hydro-electric Scheme A few areas of the MDB have not previously been modelled and hence some new IQQM or REALM models have been implemented. In some cases ancillary models are used to estimate aspects of water demands of use in the river system model. An example is the PRIDE model used to estimate irrigation for Victorian REALM models. River systems that do not receive inflows or transfers from upstream or adjacent river systems are modelled independently. This is the case for most of the river systems in the MDB and for these rivers the modelling steps are: model configuration model warm-up to set initial values for all storages in the model, including public and private dams and tanks, river reaches and soil moisture in irrigation areas using scenario climate and inflow time series, run the river model for all climate and development scenarios CSIRO 2008 June 2008 Water availability in the Barwon-Darling 7

18 where relevant, extract initial estimates of surface groundwater exchanges and provide this to the groundwater model where relevant, use revised estimates of surface groundwater exchanges from groundwater models and re-run the river model for all scenarios. For river systems that receive inflows or transfers from upstream or adjacent river systems, model inputs for each scenario were taken from the upstream models. In a few cases several iterations were required between upstream and downstream models because of the complexities of the water management arrangements. An example is the connections between the Murray, Murrumbidgee and Goulburn regions and the Snowy Mountains Hydro-electric Scheme. For all scenarios, the river models are run for the 111-year period 1 July 1895 to 30 June This period therefore ignores the first and last six months of the 112-year period considered in the climate analyses and the rainfall-runoff modelling. 1 Introduction Surface groundwater interactions The project explicitly considers and quantifies the water exchanges between rivers and groundwater systems. The approaches used are described below. The river models used by State agencies have typically been calibrated by State agencies to achieve mass balance within calibration reaches over relatively short time periods. When the models are run for extended periods the relationships derived during calibration are assumed to hold for the full modelling period. In many cases, however, the calibration period is a period of changing groundwater extraction and a period of changing impact of this extraction on the river system. That is, the calibration period is often one of changing hydrologic relationships, a period when the river and groundwater systems have not fully adjusted to the current level of groundwater development. To provide a consistent equilibrium basis for scenario comparisons it is necessary to determine the equilibrium conditions of surface and groundwater systems considering their interactions and the considerable lag times involved in reaching equilibrium. Figure 1-3 shows an indicative timeline of groundwater use, impact on river, and how this has typically been treated in river model calibration, and what the actual equilibrium impact on the river would be. By running the groundwater models until a dynamic equilibrium is reached, a reasonable estimate of the ultimate impact on the river of current groundwater use is obtained. A similar approach is used to determine the ultimate impact of future groundwater use. Figure 1-3. Timeline of groundwater use and resultant impact on river 8 Water availability in the Barwon-Darling June 2008 CSIRO 2008

19 For some groundwater management units particularly fractured rock aquifers there is significant groundwater extraction but no model available for assessment. In these cases there is the potential for considerable impacts on streamflow. At equilibrium, the volume of water extracted must equal the inflows to the aquifer from diffuse recharge, lateral flows and flows from overlying rivers. The fraction that comes from the overlying rivers is determined using a connectivity factor that is estimated from the difference in levels between the groundwater adjacent to the river and the river itself, the conductance between the groundwater pump and the river, and the hydrogeological setting. Given the errors inherent in this method, significant impacts are deemed to be those about 2 GL/year for a subcatchment, which given typical connectivity factors translates to groundwater extraction rates of around 4 GL/year for a subcatchment. 1.6 Monthly water accounts Monthly water accounts provide an independent set of the different water balance components by river reach and by month. The water accounting differs from the river modelling in a number of key aspects: the period of accounting extends to 2006 where possible, which is typically more recent than the calibration and evaluation periods of the river models assessed. This means that a comparison can produce new insights about the performance and assumptions in the river model, as for example associated with recent water resources development or the recent drought in parts of the MDB the accounting is specifically intended to estimate, as best as possible, historical water balance patterns, and used observed rather than modelled data wherever possible (including recorded diversions, dam releases and other operations). This reduces the uncertainty associated with error propagation and assumptions in the river model that were not necessarily intended to reproduce historical patterns (e.g. differences in actual historical and potential future degree of entitlement use) the accounting uses independent, additional observations and estimates on water balance components not used before such as actual water use estimates derived from remote sensing observations. This can help to constrain the water balance with greater certainty. 1 Introduction The water accounting methodology invokes models and indirect estimates of water balance components where direct measurements are not available. These water accounts are not an absolute point of truth. They provide an estimate of the degree to which the river water balance is understood and gauged, and a comparison between river model and water account water balances provides one of several lines of evidence to inform our (inevitably partially subjective) assessment of model uncertainty and its implications for the confidence in findings. The methods for water accounting are based on existing methods and those used by Kirby et al. (2006) and Van Dijk et al. (2008) and are described in detail in Kirby et al. (2008) Wetland and irrigation water use An important component of the accounting is an estimate of actual water use based on remote sensing observations. Spatial time series of monthly net water use from irrigation areas, rivers and wetlands are estimated using interpolated station observations of rainfall and climate combined with remote sensing observations of surface wetness, greenness and temperature. Net water use of surface water resources is calculated as the difference between monthly rainfall and monthly actual evapotranspiration (AET). AET estimates are based on a combination of two methods. The first method uses surface temperature remotely sensed by the AVHRR series of satellite instruments for the period 1990 to 2006 and combines this with spatially interpolated climate variables to estimate AET from the surface energy balance (McVicar and Jupp, 2002). The second method loosely follows the FAO56 crop factor approach and scales interpolated potential evaporation (PET) estimates using observations of surface greenness and wetness by the MODIS satellite instrument (Van Dijk et al., 2008). The two methods are constrained using direct on-ground AET measurements at seven study sites and catchment streamflow observations from more than 200 catchments across Australia. Both methods provide AET estimates at 1 km resolution. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 9

20 The spatial estimates of net water use are aggregated for each reach and separately for all areas classified as either irrigation area or floodplains and wetlands. The following digital data sources were used: land use grids for 2000/01 and 2001/02 from the Bureau of Rural Sciences (adl.brs.gov.au/mapserv/landuse/) NSW wetlands maps from the NSW Department of Environment and Conservation (NSW DEC) hydrography maps, including various types of water bodies and periodically inundated areas, from Geoscience Australia (GA maps; Topo250K Series 3) long-term rainfall and AET grids derived as outlined above LANDSAT satellite imagery for the years 1998 to The reach-by-reach estimates of net water use from irrigation areas and from floodplains and wetlands are subject to the following limitations: 1 Introduction partial validation of the estimates suggested an average accuracy in AET estimation within 15 percent, but probably decreasing with the area over which estimates are averaged. Uncertainty in spatial estimates originates from the interpolated climate and rainfall data as well as from the satellite observations and the method applied errors in classification of irrigation and floodplain/wetland areas may have added an unknown uncertainty to the overall estimates, particularly where subcatchment definition is uncertain or wetland and irrigation areas are difficult to discern estimated net water use cannot be assumed to have been derived from surface water in all cases as vegetation may also have access to groundwater use, either directly or through groundwater pumping estimated net water use can be considered as an estimate of water demand that apparently is met over the long-term. Storage processes, both in irrigation storages and wetlands, need to be simulated to translate these estimates in monthly (net) losses from the river main stem. Therefore, the AET and net water use estimates may be used internally in conceptual water balance models of wetland and irrigation water use that include a simulated storage Calculation and attribution of apparent ungauged gains and losses In a river reach, ungauged gains or losses are the difference between the sum of gauged main stem and tributary inflows, and the sum of main stem and distributary outflows and diversions. This would be equal to measured main stem outflows and water accounting could occur with absolute certainty. The net sum of all gauged gains and losses provides an estimate of ungauged apparent gains and losses. There may be differences between apparent and real gains and losses for the following reasons: apparent ungauged gains and losses will also include any error in discharge data that may originate from errors in stage gauging or from the rating curves associated to convert stage height to discharge ungauged gains and losses can be compensating and so appear smaller than in reality. This is more likely to occur at longer time scales. For this reason water accounting was done on a monthly time scale changes in water storage in the river reach, connected reservoirs, or wetlands can lead to apparent gains and losses that become more important as the time scale of analysis decreases. A monthly time scale has been chosen to reduce storage change effects, but they can still occur. The monthly pattern of apparent ungauged gains and losses are evaluated for each reach in an attempt to attribute them to real components of water gain or loss. The following techniques are used in sequence: analysis of normal (parametric) and ranked (non-parametric) correlation between apparent ungauged gains and losses on one hand, and gauged and estimated water balance components on the other hand. Estimated components included SIMHYD estimates of monthly local inflows and remote sensing-based estimates of wetland and irrigation net water use visual data exploration: assessment of temporal correlations in apparent ungauged gains and losses to assess trends or storage effects, and a comparison of apparent ungauged gains and losses with a time series of estimated water balance components. 10 Water availability in the Barwon-Darling June 2008 CSIRO 2008

21 Based on the above information, apparent gains and losses are attributed to the most likely process, and an appropriate method was chosen to estimate the ungauged gain or loss using gauged or estimated data. The water accounting model includes the following components: a conceptual floodplain and wetland running a water balance model that estimates net gains and losses as a function of remote sensing-based estimates of net water use and main stem discharge observations a conceptual irrigation area running a water balance model that estimates (net) total diversions as a function of any recorded diversions, remote sensing-based estimates of irrigated area and net crop water use, and estimates of direct evaporation from storages and channels a routing model that allows for the effect of temporary water storage in the river system and its associated water bodies and direct open water evaporation a local runoff model that transforms SIMHYD estimates of local runoff to match ungauged gains. These model components are described in greater detail in Kirby et al. (2008) and are only used where the data or ancillary information suggests their relevance. Each component has a small number of unconstrained or partially constrained parameters that need to be estimated. A combination of direct estimation as well as step-wise or simultaneous automated optimisation is used, with the goal to attribute the largest possible fraction of apparent ungauged gains and losses. Any large residual losses and gains suggest error in the model or its input data. 1 Introduction 1.7 Groundwater modelling Groundwater assessment, including groundwater recharge modelling, is undertaken to assess the implications of the climate and development scenarios on groundwater management units (GMUs) across the MDB. A range of methods are used appropriate to the size and importance of different GMUs. There are over 100 GMUs in the MDB, and the choice of methods was based on an objective classification of the GMUs as high, medium or low priority. Rainfall-recharge modelling is undertaken for all GMUs. For dryland areas, daily recharge was assessed using a model that considered plant physiology, water use and soil physics to determine vertical water flow in the unsaturated zone of the soil profile at a single location. This model is run at multiple locations across the MDB in considering the range of soil types and land uses to determine scaling factors for different soil and land use conditions. These scaling factors are used to scale recharge for given changes in rainfall for all GMUs according to local soil types and land uses. For many of the higher priority GMUs, recharge is largely from irrigation seepage. In New South Wales this recharge has been embedded in the groundwater models as a percentage of the applied water. For irrigation recharge, information was collated for different crop types, irrigation systems and soil types, and has been used for the scenario modelling. For high priority GMUs numerical groundwater models are being used. In most cases these already exist but often require improvement. In some cases new models are being developed. Although the groundwater models have seen less effort invested in their calibration than the existing river models, the project has invested considerable effort in model calibration and various cross-checks to increase the level of confidence in the groundwater modelling. For each groundwater model, each scenario is run using river heights as provided from the appropriate river system model. For recent and future climate scenarios, adjusted recharge values are also used, and for future development the 2030 groundwater extractions levels are used. The models are run for two consecutive 111-year periods (to match the 111-year period used for the river modelling). The average surface-groundwater flux values for the second 111-year period are passed back to the river models as the equilibrium flux. The model outputs are used to assess indicators of groundwater use and reliability. For lower priority GMUs no models are available and the assessments are limited to simple estimates of recharge, estimates of current and future extraction, allocation based on State data, and estimates of the current and future impacts of extraction on streamflow where important. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 11

22 1.8 Environmental assessment Environmental assessments on a region by region basis consider the environmental assets already identified by State governments or the Australian Government that are listed in the Directory of Important Wetlands in Australia (Environment Australia, 2001) or the updated on-line database of the directory. From this directory, environmental assets are selected for which there exists sufficient publicly available information on hydrological indicators (such as commenceto-fill levels) which relate to ecological responses such as bird breeding events. 1 Introduction Information sources include published research papers and reports, accessible unpublished technical reports, or advice from experts currently conducting research on specific environmental assets. In all cases the source of the information on the hydrological indicators used in each assessment is cited. The selection of the assets for assessment and hydrologic indicators was undertaken in consultation with State governments and the Australian Government through direct discussions and through reviews by the formal internal governance and guidance structures of the project. The Directory of Important Wetlands in Australia (Environment Australia, 2001) lists over 200 wetlands in the MDB. Information on hydrological indicators of ecological response adequate for assessing scenario changes only exists for around one-tenth of these. More comprehensive environmental assessments are beyond the terms of reference for the project. The Australian Department of Environment and Water Resources has separately commissioned a compilation of all available information on the water requirements of wetlands in the MDB that are listed in the Directory of Important Wetlands in Australia. For regions where the above selection criteria identify no environmental assets, the river channel itself is considered as an asset and ecologically-relevant hydrologic assessments are reported for the channel. The locations for which these assessments are provided are guided by prior studies. In the Victorian regions for example, detailed environmental flow studies have been undertaken which have identified environmental assets at multiple river locations with associated hydrological indicators. In these cases a reduced set of locations and indicators has been selected in direct consultation with the Victorian Department of Sustainability and Environment. In regions where less information is available, hydrological indicators may be limited to those that report on the water sharing targets that are identified in water planning policy or legislation. Because the environmental assessments are a relatively small component of the project, a minimal set of hydrological indicators are used in assessments. In most cases this minimum set includes change in the average period between events and change in the maximum period between events as defined by the indicator. A quality assurance process is applied to the results for the indicators obtained from the river system models which includes checking the consistency of the results with other river system model results, comparing the results to other published data and with the asset descriptions, and ensuring that the river system model is providing realistic estimates of the flows required to evaluate the particular indicators. 1.9 References Agrecon (2005) Agricultural Reconnaissance Technologies Pty Ltd Hillside Farm Dams Investigation. MDBC Project 04/4677DO. Australian Bureau of Statistics (2004) Population projections for Statistical Local Areas 2002 to (ASGC 2001). ABS Catalogue No Available at: Battaglia M and Sands P (1997) Modelling site productivity of Eucalyptus globulus in response to climatic and site factors. Australian Journal of Plant Physiology 24, Brown AE, Podger GM, Davidson AJ, Dowling TI and Zhang L (2006) A methodology to predict the impact of changes in forest cover on flow duration curves. CSIRO Land and Water Science Report 8/06. CSIRO, Canberra. Chiew FHS, Teng J, Kirono D, Frost A, Bathols J, Vaze J, Viney N, Hennessy K and Cai W (2008a) Climate data for hydrologic scenario modelling across the Murray-Darling Basin. A report to the Australian government from the CSIRO Murray-Darling Basin Sustainable Yields Project. CSIRO, Australia. In prep. Chiew FHS, Vaze J, Viney N, Jordan P, Perraud J-M, Zhang L, Teng J, Pena J, Morden R, Freebairn A, Austin J, Hill P, Wiesenfeld C and Murphy R (2008b) Rainfall-runoff modelling across the Murray-Darling Basin. A report to the Australian government from the CSIRO Murray-Darling Basin Sustainable Yields Project. CSIRO, Australia. In prep. Chiew FHS (2006) An overview of methods for estimating climate change impact on runoff. In: Proceedings of the 30th Hydrology and Water Resources Symposium, December 2006, Launceston. Chiew FHS and Leahy C (2003) Comparison of evapotranspiration variables in Evapotranspiration Maps of Australia with commonly used evapotranspiration variables. Australian Journal of Water Resources 7, Water availability in the Barwon-Darling June 2008 CSIRO 2008

23 Chiew FHS, Peel MC and Western AW (2002) Application and testing of the simple rainfall-runoff model SIMHYD. In: Singh VP and Frevert DK (Eds), Mathematical Models of Small Watershed Hydrology and Application. Littleton, Colorado, pp DSE (2004) Victoria in Future 2004 Population projections. Department of Sustainability and Environment, Victoria. Available at: Environment Australia (2001) A directory of important wetlands in Australia. Third edition. Environment Australia, Canberra. Available at: Geosciences Australia (2007) Man made hydrology GIS coverage (supplied under licence to CSIRO). Australian Government, Canberra. IPCC (2007) Climate Change 2007: The Physical Science Basis. Contributions of Working Group 1 to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press. Jeffrey SJ, Carter JO, Moodie KB and Beswick AR (2001) Using spatial interpolation to construct a comprehensive archive of Australian climate data. Environmental Modelling and Software 16, Kirby J, Mainuddin M, Podger G and Zhang L (2006) Basin water use accounting method with application to the Mekong Basin. In: Sethaputra S and Promma K (eds) Proceedings on the International Symposium on Managing Water Supply for Growing Demand, Bangkok, Thailand, October 2006, pp Jakarta: UNESCO. Kirby J et al. (2008) Reach-level water accounting for across the Murray-Darling Basin. A report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project. CSIRO, Australia. In prep. Marcar NE, Benyon RG, Polglase PJ, Paul KI, Theiveyanathan S and Zhang L (2006) Predicting the Hydrological Impacts of Bushfire and Climate Change in Forested Catchments of the River Murray Uplands: A Review. CSIRO Water for a Healthy Country. McVicar TR and Jupp DLB (2002) Using covariates to spatially interpolate moisture availability in the Murray-Darling Basin. Remote Sensing of Environment 79, Nash JE and Sutcliffe JV (1970) River flow forecasting through conceptual models 1: A discussion of principles. Journal of Hydrology 10, Nathan RJ, Jordan PW and Morden R (2005) Assessing the impact of farm dams on streamflows 1: Development of simulation tools. Australian Journal of Water Resources 9, New South Wales Government (2000) Water Management Act 2000 No 92. New South Wales Parliament, December Available at Queensland Government (2000) Water Act Queensland Government, Brisbane. South Australia Government (2004) Natural Resources Management Act The South Australian Government Gazette, Adelaide, September Available at: Tan KS, Chiew FHS, Grayson RB, Scanlon PJ and Siriwardena L (2005) Calibration of a daily rainfall-runoff model to estimate high daily flows. MODSIM 2005 International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand, December 2005, pp ISBN: Van Dijk AIJM et al. (2008) Uncertainty assessments for scenario modelling. A report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project, CSIRO Australia. In prep. VicMap (2007) Topographic data series. State of Victoria. Available at Victoria Government (1989) Water Act 1989, Act Number 80/1989. Parliament of Victoria. 1 Introduction CSIRO 2008 June 2008 Water availability in the Barwon-Darling 13

24 2 Overview of the region The Barwon-Darling region is in northwestern New South Wales and covers 13 percent of the total area of the Murray- Darling Basin (MDB). The region is based around the Barwon and Darling rivers. The population is 50,000 or 2.5 percent of the MDB total, concentrated in the centres of Collarenebri, Walgett, Brewarrina, Bourke, Cobar and Wilcannia. The major land use is dryland pasture used for beef and sheep grazing. Approximately 63,000 ha of land were irrigated in 2000 including 57,900 ha for cotton production on the western plains. The Talyawalka area is a nationally important wetland located on the Darling Riverine Plains between Wilcannia and Menindee. The area comprises the wetlands of the Talyawalka Anabranch of the Darling River and its distributary, Teryawynia Creek. It is representative of a semi-arid inland floodplain wetland system fringed by Black Box woodland. 2 Overview of the region The region uses 3 percent of the surface water diverted for irrigation in the MDB and groundwater use is less than 1 percent of the MDB total. The Barwon and Darling rivers are usually considered to be unregulated rivers and there are no major public water storages within the region. However, there are numerous weirs along these rivers and large storages regulate many of the upstream tributaries. This chapter summarises the region s biophysical features including rainfall, topography, land use and the environmental assets of significance. It outlines the institutional arrangements for the region s natural resources and presents key features of the surface and groundwater resources of the region including historical water use. 2.1 The region The Barwon-Darling region is located predominantly within northwestern New South Wales and covers 142,173 km 2 or 13 percent of the MDB. It is bounded to the west by the internally draining Bulloo catchment and receives flow from the Paroo, Warrego and Condamine-Balonne regions in the north and the Moonie, Border Rivers, Gwydir, Namoi and Macquarie-Castlereagh regions to the east. It is bounded by the Lachlan region to the south and delivers river flow to the Murray region in the south-west. The region terminates upstream of Menindee Lakes at Billilla Homestead located on the Darling River about 80 km downstream of Wilcannia. The region comprises undulating uplands in the eastern, northeastern and northern margins that fall to the south-west and extensive alluvial floodplains along the Barwon-Darling system and its main tributaries. The major water sources in the region include the Barwon and Darling rivers, the Castlereagh River below Coonamble, the lower reaches of the Bogan River and the Marra and Thalaba Creeks, alluvial aquifers of the Great Artesian Basin (GAB), wetlands and water storages. Both public and private infrastructure are associated with these water sources including weirs on the lower Darling River and on-farm water storage. Irrigation is almost entirely from surface water diversion. Annual rainfall is low and averages 328 mm for the entire region but falls to 250 mm in the west. Rainfall is low throughout the year but is higher in summer than winter. The inter-annual variability of rainfall is high and the period 1950 to 2000 was wetter in the region on average than the period 1900 to 1950 (Figure 2-1). The average annual rainfall of 339 mm over the ten-year period 1997 to 2006 is similar to the long-term average. 14 Water availability in the Barwon-Darling June 2008 CSIRO 2008

25 Annual rainfall (mm) Mean monthly rainfall (mm) J F M A M J J A S O N D Figure annual and monthly rainfall averaged over the region. The curve on the annual graph shows the low frequency variability The Barwon-Darling region contributes about 2.8 percent of the total runoff in the MDB. Average annual modelled runoff over the region for the 111-year period is 6 mm and is higher in summer and early autumn. The average annual modelled runoff over the ten-year period 1997 to 2006 was 8 percent higher but not significantly different to the long-term average values. The runoff estimates for the Barwon-Darling region are based on gauged catchments from outside the region so there is less confidence in the modelled outputs, particularly in the drier western half of the region. The region s population is approximately 50,000 which is 2.5 percent of the MDB total. The largest towns include Collarenebri and Walgett in the east, Brewarrina, Bourke and Cobar in the central area, and Wilcannia in the south-west. The major land use is dryland pasture used for broadacre livestock grazing. Almost one-third of the land area remains as native vegetation. Crops are irrigated by supplementary water from large on-farm storages along the river system. The water is harvested from upstream tributary flows in preceding months. Approximately 63,000 ha of irrigated crops were grown in 2000 and over 90 percent was cotton. Hot summer temperatures and generally low but erratic rainfall in the region support less intensive dryland agricultural production than in the rest of New South Wales. 2 Overview of the region CSIRO 2008 June 2008 Water availability in the Barwon-Darling 15

26 2 Overview of the region * Paroo Overflow Lakes are fed from the Paroo River and are thus assessed in the report for the Paroo region. ** Talyawalka Wetland is fed from Teryawynia Creek and the Talyawalka Anabranch of the Darling River and are thus assessed in this report. Figure 2-2. Map of dominant land uses of the Barwon-Darling region with inset showing the region s location within the Murray-Darling Basin. The map only shows wetlands that are assessed in the project (Chapter 7). A full list of key assets associated with the region is in Table 2-2 The land use map (Figure 2-2) and land use area (Table 2-1) are based on the 2000 land use of the MDB grid, derived from 2001 Bureau of Rural Sciences AgCensus data (BRS, 2005). Irrigation estimates are based on crop areas recorded as irrigated in the census. 16 Water availability in the Barwon-Darling June 2008 CSIRO 2008

27 Table 2-1. Summary of land use in the year 2000 within the Barwon-Darling region Land use Area percent ha Dryland crops 5.0% 706,200 Dryland pasture 63.7% 9,030,600 Irrigated crops 0.4% 63,000 Cereals 4.1% 2,600 Cotton 92.0% 57,900 Horticulture 0.3% 200 Orchards 0.3% 200 Pasture and hay 2.7% 1,700 Vine fruits 0.6% 400 Native vegetation 30.2% 4,290,900 Plantation forests <0.1% 2,100 Urban <0.1% 3,500 Water 0.6% 90,000 Total 100.0% 14,186,300 Source: BRS, The Western Catchment Action Plan (CAP) is the strategic framework that will guide natural resource management in the New South Wales portion of the Barwon-Darling region over the next ten years. It was prepared under the Catchment Management Authorities Act 2003 (CMA Act) and commenced in June The Western CAP is a ten-year plan for improving and managing natural resources in the region. The CAP specifies how the Western Catchment Management Authority (CMA) will direct the $19 million approved under the current three-year investment strategy (and future funding) to manage the catchment s natural resources (Western CMA, 2006). It specifies catchment and management targets used to measure the health of the catchment. The targets consider other plans that may affect the catchment, including water sharing plans. The CAP outlines targets for four themes: land and vegetation, rivers and groundwater, biodiversity, and community. The rivers and groundwater theme includes three catchment targets for 2015: 2 Overview of the region the Surface Water System Health Index Rating (HIR) and the Groundwater System HIR improved at 60 percent of relevant monitoring sites and maintained at all other monitoring sites salinity in the Barwon-Darling at Wilcannia less than 800 EC for 80 percent of the time as measured on a daily basis and less than 350 EC for 50 percent of the time salt load in the Barwon-Darling at Wilcannia less than 530,000 tonnes per year for 80 percent of the time and less than 160,000 tonnes per year for 50 percent of the time. Four management programs were established to address these catchment targets: aquatic habitat program water quality and salinity program surface water management program groundwater management program. 2.2 Environmental description The lower Barwon-Darling region includes the Barrier Ranges to the south-west and the Central West Highlands east of Cobar. The Barwon-Darling valley comprises the broad alluvial plain upstream of Bourke and the more restricted alluvial floodplain between Bourke and Menindee. Approximately 30 percent of the region is covered with native vegetation. This includes open woodlands, timbered areas, chenopod shrublands and native pastures. The condition of the vegetation is highly variable, depending on its location and the value of the land for cropping or grazing. The eastern margin of the region is under increasing development pressure from extensive grazing and intensive cropping. Ground cover is variable and perennial species are declining across the region. There are four main soil types in the region: soft red, hard red, the riverine alluvial and the western sand plain. Many of the soils are highly susceptible to wind and water erosion when vegetation is disturbed. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 17

28 The wetlands within the region that have national or international importance are detailed in Table 2-2. The Talyawalka wetland area comprises the wetlands of the Talyawalka Anabranch of the Darling River and its distributary, Teryawynia Creek. They are located between Wilcannia and Menindee on the Darling Riverine Plains outside the region, but are assessed here as they are watered from the Darling River within the region. The wetland system comprises a series of braided channels across the floodplain, interspersed by intermittent wet and dry lake beds. It includes the Teryawynia, Dry, White Water, Eucalyptus/Waterloo, Victoria, Brummeys, Dennys, Brennans, Sayers, Gum, Boolaboolka, North and Ratcatchers lakes, plus associated wetlands. Most of the lakes are inundated by overflow from Teryawynia Creek, but several (Sayers, Gum, Boolaboolka, North and Ratcatchers lakes) are inundated by overflow from other lakes. It is representative of a semi-arid inland floodplain wetland system fringed by Black Box woodland. These lakes provide habitat for large numbers of waterbirds when inundated (DWE, 2007). 2 Overview of the region 18 Water availability in the Barwon-Darling June 2008 CSIRO 2008

29 Table 2-2. Ramsar wetlands and wetlands of national importance located within the Barwon-Darling region. Bold text indicates that the wetland is a part of the assessed Talyawalka Anabranch and Teryawynia Creek system. Site code Directory of Important Wetlands in Australia name Area (1) Ramsar sites ha NSW009 Macquarie Marshes (2) 200,000 yes (3) NSW014 Lake Burkanoko 271 none NSW015 Lake Nichebulka 348 none NSW016 Murphys Lake 1,000 none NSW017 Paroo River Distributary Channels 720,000 none NSW018 Willeroo Lake 113 none NSW019 Yantabulla Swamp (Cuttaburra Basin) 37,200 none NSW096 Blue Lake 237 none NSW097 Gilpoko Lake 436 none NSW099 Green Lake 392 none NSW100 Mullawoolka Basin 2,026 none NSW101 Peery Lake (Peri Lake) 5,026 none NSW102 Poloko Lake (Olepoloko Lake) 3,722 none NSW103 Tongo Lake 524 none NSW104 Yantabangee Lake 1,427 none NSW144 Blue Lake (overflow) 307 none NSW145 Budtha Waterhole 124 none NSW146 Calbocaro Billabong 65 none NSW147 Camel Lake 126 none NSW148 Coona Coona Lake 75 none NSW149 Deadmans Swamp 471 none NSW150 Dick Lake 709 none NSW151 Dry Lake (2) up to 185 none NSW152 Gidgee Lake (2) up to 229 none NSW153 Gypsum Swamp 82 none NSW154 Horseshoe Lake 90 none NSW155 Horseshoe Lake (Bartons Ck) (2) 513 to 826 none NSW156 Pelora Lake 50 none NSW157 Pirillie Lake 129 none NSW158 Taylors Lake 46 none NSW160 Waitchie Lake 205 none NSW161 Wirrania Swamp 86 none NSW162 Yammaramie Swamp 3,082 none NSW164 Bottom Lila Lake 286 none NSW165 Yandaroo Lake 33 none NSW167 The Dry Lake 133 none QLD084 Balonne River Floodplain (4) 24,000 none (1) Wetland areas have been extracted from the Australian Wetlands Database and are assumed to be correct as provided from State and Territory agencies. (2) Variable according to flooding. 2 Overview of the region (3) The Macquarie Marshes Ramsar site (18,726 ha) is mainly within the Macquarie-Castlereagh region. (4) Actual area of wetlands is in the order of several hundred hectares spread out over 24,000 ha of floodplain. Note: The Paroo Overflow Lakes are fed from the Paroo River and are assessed in the report for the Paroo region. Source: A Directory of Important Wetlands in Australia (Environment Australia, 2001). 2.3 Surface water resources Rivers and storages The Barwon River flows in a south-westerly direction from the north-east of the region at the Queensland New South Wales border. The Barwon River receives flows from the Moonie, Border Rivers, Gwydir, Namoi and Macquarie- Castlereagh regions in the north-east. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 19

30 The Darling River commences downstream of Brewarrina at the confluence of the Culgoa and Barwon rivers. The Darling River receives flows from the Warrego region in the north. Flows from the Paroo River usually terminate in the Paroo River Wetlands but reach the Darling River during extreme flood events. Runoff seldom reaches the river system from large parts of the region, particularly west and south-west of Cobar, and north-east from White Cliffs. There are no major public storages in the Barwon-Darling region. However, there are large private off-river storages. The river model includes a total capacity for these storages of 284 GL (Chapter 4). They store water for irrigation obtained by either pumping during high river flow, harvesting of floodplain runoff, and/or retention of irrigation tail-water. There are also smaller farm dams in the eastern part of the region that collect runoff from their immediate catchment. The estimated total storage capacity of these dams is 94 GL (Chapter 3) Surface water management institutional arrangements 2 Overview of the region The Water Management Act 2000 in New South Wales requires implementation of ten-year plans defining water sharing arrangements between the environment and water users and amongst water user groups. The plans aim to protect rivers and aquifers and their dependent ecosystems, and provide water users with clarity and certainty regarding water access rights. Water access is based on a long-term average extraction limit. The basic rights (native title, domestic and stock) and access licences for domestic and stock use and local water utilities are volumetric and are granted highest access priority. High and general security access licences are based on shares of the water available. High security has priority over general security. Most general security access licences are expressed as a relative unit share of the available water rather than as an annual volume. Licensing continues under the Water Act 1912 in areas where water sharing plans (WSPs) are not gazetted. There is currently no WSP in place for the Barwon-Darling. The finalisation of the Murray-Darling Basin Commission Barwon-Darling Cap on surface water diversions in July 2005 (Western CMA, 2006) identified the need for the development of a WSP for the Barwon-Darling and a macro water plan for the unregulated tributaries. The WSP for the Barwon-Darling will also formalise the existing interim unregulated flow management plan for the north-west. This plan restricts water extraction following periods of low river flow to support riverine habitat and fish passage and to suppress blue-green algae growth. Current water diversions are managed by licences issued under the 1912 Water Act and agreed river flow objectives. The Murray-Darling Basin Cap on surface water diversions for the Barwon and upper and lower Darling rivers is set at 310 GL. Access rules for Barwon-Darling irrigators were altered in July 2006 (DNR, 2006) to cap use at 173 GL/year Water products and use Irrigation in the middle reaches of the Barwon-Darling River system is predominantly for cotton production. Irrigation is based on extraction during high river flow events and water storage in on-farm dams. Water use has fluctuated between 100 and 500 GL/year over the past ten years and reflects the variation in annual runoff upstream of the region. 600 Annual diversion (GL) Figure 2-3. Historical surface water diversions Note: The data in different years are not always comparable because the areas defined in each catchment changed, as did the definitions of water uses. Even where data sets should refer to the same records, data from state and Murray-Darling Basin Commission databases often vary. Source: MDBC, Water availability in the Barwon-Darling June 2008 CSIRO 2008

31 2.4 Groundwater Groundwater management units the hydrogeology and connectivity The Barwon-Darling region overlaps 14 groundwater management units (GMUs) and none are fully contained within the region. The Lower Namoi Alluvium (N01), Lower Macquarie Alluvium (N08) and Lower Lachlan Alluvium (N12) are ranked as high priority GMUs in the context of the overall project on the basis of the size of the aquifers, the level of development and the assumed degree of connectivity with the surface water system. These high priority GMUs are assessed within the Namoi, Macquarie-Castlereagh and Lachlan regional reports respectively. Similarly, the low priority St George Alluvium GMU (Q71 in Queensland) is assessed in the Condamine-Balonne regional report. The remaining ten very low and low priority GMUs that are partially located within the Barwon-Darling region are overviewed here. A generalised cross-section of the hydrogeology of the region is shown in Figure 2-4 and a map of GMUs is in Figure Overview of the region Figure 2-4. Generalised north-west south-east cross-section of the hydrogeology of the Barwon-Darling region; from Evans et al. (1994) The GAB is the major aquifer system underlying northern New South Wales and southern Queensland. It consists of two primary hydrogeological units: the deep Jurassic sandstone confined aquifers that extend beneath the western and central part of the region and outcrop in the east the Cretaceous sandstone confined aquifers and shale confining layers which lie conformably above the Jurassic aquifers. The Cretaceous confining layers separate the deeper confined aquifers from the surficial aquifers. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 21

32 These and three other hydrogeological units associated with the GAB (Intake Beds, Alluvium and Cap Rocks) are detailed within the New South Wales water management planning process and comprise four GMUs: the deep Jurassic and Cretaceous sandstone confined aquifers and Cretaceous confining layers that are administered by the Water Sharing Plan for the NSW Great Artesian Basin Groundwater Sources (N601 GAB Central Zone and N601 GAB Warrego Zone) GAB Intake Beds that occur where the Jurassic and Cretaceous sandstone aquifers outcrop (a small area in the east of the region). This GMU is also administered by the WSP (N601 GAB Southern Zone) GAB Alluvium that is a thick sequence of Cainozoic alluvium covering the GAB sequence in the north-eastern portion of the region (N63 GAB Alluvial) GAB Cap Rock aquifers that occur in the western part of the region where the GAB is not associated with another GMU. This unit can include the weathered and fractured Cretaceous rock aquifers and Cainozoic alluvium up to a depth of 60 m (N620 GAB Cap Rocks). 2 Overview of the region The deeper GAB Jurassic and Cretaceous confined sandstone aquifers are the primary groundwater source in the region. The groundwater within these aquifers is separated from aquifers closer to the surface by thick confining beds so there is little interaction. The water resources within these confined aquifers are not considered further in this assessment. This section focuses on the remaining shallow aquifer systems and GMUs. The area upstream of Bourke is dominated by alluvial sediments. The more restricted alluvial floodplain downstream between Bourke and Wilcannia contains alluvial material along the Darling River flanked by hard fractured rock aquifers on its south side and GAB confining beds to the north. The main aquifer system upstream of Bourke is associated with the GAB Alluvial GMU (N63) that is largely composed of (within increasing depth) the Narrabri, Gunnedah and Cubbaroo formations. The Narrabri Formation is composed of shallow alluvial fan sediments that cover a wide area of the eastern part of the region with groundwater contained in small discontinuous lenses of sand. The Gunnedah Formation underlies the Narrabri Formation and does not supply usable groundwater. The Cubbaroo Formation consists of medium- to coarse-grained sands and gravels further to the east and south but are probably much finer-grained in the Barwon-Darling region. Drilling in the St George-Dirranbandi area defines a deeper trough containing thick Tertiary and Quaternary sediments (greater than 100 m). Known as the Dirranbandi Trough, its south-western extension is unknown, but it has been inferred to extend to near Bourke. Groundwater contained in these sediments is often saline to very saline. The GAB Alluvial system is recharged by rainfall infiltration, flood recharge and throughflow from up-gradient sources in the east such as the Namoi region (although this throughflow is likely to be saline). Regional groundwater flow south of the Barwon-Darling River is north-west (towards the river) where it may discharge or evapotranspire. Groundwater flow north of the river is south-west (sub-parallel to the river). Low salinity groundwater of 500 to 1000 mg/l total dissolved salts (TDS) exists in the shallowest sediments along the Barwon and Darling rivers to a point upstream of Walgett and in an area south of Bourke adjacent to the Bogan River. The groundwater salinity is greater than 14,000 mg/l TDS elsewhere including at depths below the Darling River (Evans et al., 1994). Alluvial sediments (Upper Darling Alluvium, N46) are deposited along the Darling, Paroo and Warrego rivers. The Barwon-Darling region also contains a small part of Lower Darling Alluvium GMU (N45) downstream of Wilcannia. The groundwater salinity in the shallowest sediments of the Upper Darling Alluvium is 500 to 1000 mg/l TDS along and immediately adjacent to the Darling River, and generally greater than 1500 mg/l TDS along the Paroo and Warrego rivers (Evans et al., 1994). Woolley et al. (2004) confirmed supplies of low salinity groundwater within the alluvial deposits of the Darling palaeochannel at depths less than 60 m. Two town water supply bores now access this source. Woolley et al. (2004) indicate that this source of water may be significant but note that more monitoring data and investigations are required to establish an appropriate management regime. 22 Water availability in the Barwon-Darling June 2008 CSIRO 2008

33 The upland GMUs downstream of Bourke are: the Barrier Ranges to the south-west which correspond to the Kanmantoo Fold Belt GMU (N817) the Central West Highlands to the east which correspond to the Lachlan Fold Belt GMU (N811) the area west of the Darling River which corresponds with the GAB Cap Rocks GMU (N620). The Barrier Ranges and Central West Highlands upland rocks are also the basement of the GAB confined aquifers. The hydrogeology of these upland areas is dominated by rain-fed fractured rock aquifers in a range of different rock types that are usually deeply weathered. Groundwater flow in the fractured rocks is controlled by topography, but is generally to the south-west tending towards the main rivers. There are also local deposits of unconsolidated sediments within the upland areas forming aquifers in the valley floors. Thin layers of residual and aeolian sediments form aquifers that overlie the fractured rock systems. The GAB Cap Rocks GMU occurs where the Cretaceous confining layer outcrops and includes any associated minor rocks and alluvium to a depth of 60 m. The Cap Rocks contain a rain-fed unconfined to semi-confined aquifer. Groundwater salinity across these three GMUs is generally greater than 1500 mg/l TDS. Connectivity mapping has classified the lower reaches of the Barwon River as losing but are gaining between the Bokara River confluence and Walgett. The river becomes more strongly losing upstream of Walgett. The lower reaches of the Darling River are losing and the intermediate reaches are gaining. 2 Overview of the region Figure 2-5. Map of groundwater management units within the Barwon-Darling region CSIRO 2008 June 2008 Water availability in the Barwon-Darling 23

34 Table 2-3. Categorisation of groundwater management units, including extraction, entitlement and recharge details 2 Overview of the region Code Name Priority Current extraction* (2004/05) Total entitlement GL/y Long-term average extraction limit Recharge** N45 Lower Darling Alluvium very low <0.1 < N46 Upper Darling Alluvium very low N63 GAB Alluvial low N601 GAB Intake Beds very low np N604 Gunnedah Basin very low N612 Western Murray Porous Rock very low N620 GAB Cap Rocks low *** ***514.8 N811 Lachlan Fold Belt low N813 Warrambungle Tertiary Basalt very low <0.1 < N817 Kanmantoo Fold Belt very low *Current groundwater extraction for Macro Groundwater Sharing Plan areas is based on metered and estimated data provided by NSW Department of Water and Energy (DWE). Data quality is variable depending on the location of bores and the frequency of meter reading. ** This value represents only rainfall recharge in the NSW Macro Groundwater Sharing Plan areas. The volume of recharge does not account for recharge in national park areas, which is not available for consumptive use and has been effectively allocated to the environment. *** The Long Term Average Extraction Limit and rainfall recharge assigned to the GAB Cap Rock is not for the deeper GAB confined aquifers but to the GAB cap rocks and any associated minor rocks and alluvium to a depth of not more than 60 m. np not provided Water management institutional arrangements The Water Management Act 2000 in New South Wales requires the implementation of ten-year plans defining water sharing arrangements between the environment and groundwater users and amongst water user groups in a similar way to that required for surface water diversions. WSPs are prepared for the more highly developed GMUs to protect rivers and aquifers and their dependent ecosystems, and to provide water users with clarity and certainty regarding water access rights. A supplementary access volume is determined where current extraction exceeds the long-term average extraction limit. This supplementary access volume decreases to zero within ten years of commencement of the WSP. Groundwater extraction is controlled by Macro water sharing plans (Macro WSPs) outside of areas covered by WSPs. They have an extraction limit and environmental provisions. Groundwater extraction records for the Macro WSP regions are generally poor. The Macro WSPs are planned to commence in Groundwater extraction in the Barwon-Darling region, other than the GAB Intake Beds, is controlled by groundwater Macro WSPs. Extraction from the Intake Beds is controlled by the WSP for the New South Wales GAB Groundwater Sources The water planning data are summarised in Table 2-4 excluding GMUs incorporated within other WSP areas. 24 Water availability in the Barwon-Darling June 2008 CSIRO 2008

35 Table 2-4. Summary of groundwater management plans Description Great Artesian Basin (Intake Beds) Other areas Name of plan Water Sharing Plan for the NSW Great Artesian Basin Groundwater Sources Macro Water Sharing Plans Year of plan 2007 * Environmental provisions Planned share Volume required to maintain pressure levels experienced under the level of water extraction associated with water entitlements, infrastructure and management rules in place plus 30% water savings under Cap and Pipe Bores up to a maximum for each zone (for more detail refer Part 4, section 17(2) of WSP) % of rainfall recharge Adaptive provisions Refer Part 4 section 18 of WSP None Basic rights Domestic and stock rights 0.21 GL/y 9.26 GL/y Native title none none Access licences Urban 0 GL/y 1.35 GL/y Planned share 0.14 GL/y 2.54 GL/y Announced allocation Supplementary provisions *Unpublished data supplied by DWE. Macro WSPs will commence in Water products and use Groundwater extraction within the GMUs assessed in the Barwon-Darling region accounts for 0.6 percent (10 GL/year) of the MDB total. There are approximately 2911 groundwater users within the region. The region is sparsely settled and groundwater is primarily used to water livestock. There are no available records of historical groundwater extraction and the extraction rates quoted here are estimates based upon typical usage. none none 2 Overview of the region 2.5 References DWE (2007) Facts about wetlands. NSW Department of Water and Energy. Available at naturalresources.nsw.gov.au/water/wetlands_facts_international.shtml BRS (2005) 1993, 1996, 1998 and 2000 Land Use of the Murray-Darling Basin, Version 2. Resource Identifier: ID01. Online digital dataset and spatial data layer. File identifier: adl.brs.gov.au/findit/metadata_files/a_mdblur9abl_00711a00.xml. Product access: data.brs.gov.au/anrdl/a_mdblur9abl_00711a00.xml DNR (2006) A Strategy to Manage Extractions from the Barwon-Darling. Pamphlet Series Land and Water for Life. New South Wales Department of Natural Resources. Environment Australia (2001) A directory of important wetlands in Australia. Third edition. Environment Australia, Canberra. Available at: Evans WR, Hillier J and Woolley DR (1994) Hydrogeology of the Darling River Drainage Basin (1:1,000,000 scale map). Australian Geological Survey Organisation, Canberra. MDBC (2007) Water Audit Monitoring Reports (1995 to 2004). Nine reports cover the years 1994/5 to 2003/4. Murray Darling Basin Commission, Canberra. Available at: Western CMA (2006) Western Catchment Plan Draft version II. Western Catchment Management Authority. Available at Woolley DR, Williams RM and Varathan S (2004) New Water? Groundwater Supply for Wilcannia, NSW. Proceedings of the 9 th Murray-Darling Basin Groundwater Workshop, Bendigo. February 17 19, CSIRO 2008 June 2008 Water availability in the Barwon-Darling 25

36 3 Rainfall-runoff modelling This chapter includes information on the climate and rainfall-runoff modelling for the Barwon-Darling region. It has four sections: a summary an overview of the regional modelling approach a presentation and description of results a discussion of key findings. 3.1 Summary 3 Rainfall-runoff modelling Issues and observations The methods used for climate scenario and rainfall-runoff modelling across the Murray-Darling Basin (MDB) are described in Chapter 1. There are no significant differences in the methods used to model the Barwon-Darling region Key messages The annual rainfall and modelled runoff averaged over the Barwon-Darling region are 328 mm and 6 mm respectively. Rainfall is low throughout the year but highest in summer and runoff is highest in summer and early autumn. The rainfall, runoff and the fraction of rainfall that becomes runoff in the region (particularly in the west) are amongst the lowest in the MDB. The region covers 13.4 percent of the MDB but generates only about 2.8 percent of the total runoff. The average annual rainfall and runoff over the ten-year period 1997 to 2006 are 3 and 8 percent higher respectively than the long-term (1895 to 2006) averages. However, because of the inter-annual variability and the relatively short ten-year period used as the basis for comparison, the 1997 to 2006 rainfall and runoff are not significantly different to the long-term average values, even at a significance level of α = 0.2. Under the best estimate (median) 2030 climate average annual runoff would be reduced by 2 percent. The extreme estimates from the high global warming scenario range from a 22 percent reduction to a 50 percent increase in average annual runoff. The range from the low global warming scenario is an 8 percent reduction to a 12 percent increase. There are no commercial forestry plantations in the region and none are projected for the future. The total farm dam storage volume is projected to increase by 13.2 GL (14 percent) by ~2030. The projected increase in farm dams will reduce mean annual runoff by less than 0.5 percent Uncertainty Scenario A historical climate and current development The runoff for the four subcatchments in the eastern half of the region is estimated using model parameter values from gauged catchments more than 100 km away. Consequently there is less confidence in the runoff estimates in this region compared to the data-rich regions in the eastern and southern MDB. The runoff estimates for the western half of the region are relatively poor because there are no small or medium sized gauged catchments within 250 km and runoff is modelled using default parameter values. It is considerably more difficult to model runoff in the central, western and northern MDB because the region is drier, there are far fewer rainfall stations, and river flows are intermittent with most of the runoff occurring as infrequent floods. 26 Water availability in the Barwon-Darling June 2008 CSIRO 2008

37 Scenario C future climate and current development The biggest uncertainty in Scenario C modelling is in the global warming projections and the modelled implications of global warming on local rainfall. The uncertainty in the modelling of climate change impact on runoff is small compared to the climate change projections. This project takes into account the current uncertainty in climate change projections explicitly by considering results from 15 global climate models and three global warming scenarios based on the Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC, 2007). The results are then presented as a best estimate (median) and a range of the extreme estimates of climate change impact on runoff. Scenario D future climate and future development After the Scenario C climate change projections, the biggest uncertainty in Scenario D modelling is in the projections of future increases in farm dams and the impact of this development on runoff. The impact of commercial forestry plantations on runoff is not modelled because the Bureau of Rural Sciences projections indicate negligible growth in commercial forestry in the region (BRS, 2005). The increase in farm dams is estimated by considering trends in historical farm dam growth and current policy controls in New South Wales and there is uncertainty both as to how landholders will respond to these policies and how governments may set their future policies. These uncertainties have little bearing on the subsequent use of runoff in the river models because 99% of the available water in the Barwon-Darling region comes from upstream regions and only 1% from runoff within the region (see Chapter 4). 3.2 Modelling approach Rainfall-runoff modelling general approach 3 Rainfall-runoff modelling The general rainfall-runoff modelling approach is described more fully in Chapter 1 and in detail in Chiew et al. (2008). A brief summary is given below. The lumped conceptual daily rainfall-runoff model SIMHYD is used with a Muskingum routing method to estimate daily runoff at 0.05 o grids (~ 5 km x 5 km) across the entire MDB for the four climate and development scenarios. The rainfallrunoff model is calibrated against 1975 to 2006 streamflow data from about 180 small and medium size unregulated MDB catchments (50 to 2000 km 2 ). The six parameters of SIMHYD are optimised in the model calibration to maximise an objective function that incorporates the Nash-Sutcliffe efficiency of monthly runoff and daily flow duration curve. Calibration includes a constraint to ensure that the total modelled runoff over the calibration period is within 5 percent of the total recorded runoff. The runoff for a 0.05 o grid cell in an ungauged subcatchment is modelled using optimised parameter values for a calibration catchment closest to that subcatchment. The rainfall-runoff model SIMHYD is used because it is simple, has relatively few parameters and provides a consistent basis (that is automated and reproducible) for modelling historical runoff across the entire MDB and assessing the potential impacts of climate change and development on future runoff. Specific calibration of SIMHYD or more complex rainfall-runoff models in data-rich areas based on expert judgement and local knowledge (as done by some state agencies) would lead to better model calibration for the specific modelling objectives of the area Rainfall-runoff modelling for the Barwon-Darling region The rainfall-runoff modelling estimates runoff in 0.05 o grid cells in five subcatchments as defined for the river system modelling in Chapter 4 (Figure 3-1). There are no gauged catchments smaller than 2000 km 2 in the Barwon-Darling region. Optimised model parameter values from two gauged catchments are used to model runoff in the four eastern subcatchments. One of the calibration catchments is in the Gwydir region and the other is in the Macquarie-Castlereagh region. Runoff in the large western subcatchment is modelled using default parameter values (Section 3.2.3). Scenario B modelling is not done because the mean annual rainfall and modelled runoff for the ten-year period 1997 to 2006 are not significantly different (at statistical significance level of α = 0.2 with the Student-t and Rank-Sum tests) from the long-term (1895 to 2006) mean values (Section 3.3.1). CSIRO 2008 June 2008 Water availability in the Barwon-Darling 27

38 The impact of commercial forestry plantations on runoff is not modelled because the Bureau of Rural Sciences projections that take into account industry information indicate no plantations at present and no future growth. 3 Rainfall-runoff modelling The increase in farm dams in the four eastern subcatchments is estimated according to the following method. The western subcatchment ( ) is in the New South Wales Western Division where there is little runoff and farm dam growth is unlikely. The farm dam projection is dependent on three factors: current farm dam storage volume, growth rate of farm dams, and maximum harvestable right volumes in New South Wales (NSW Government, 2000). The current farm dam storage volume is estimated from the satellite imagery captured between 2004 and 2006 (Geosciences Australia, 2007). The farm dam growth rate is estimated using data from Agrecon (2005) for 1999 to A growth rate of 0.6 percent per year is used for New South Wales. The maximum harvestable right volume is estimated by multiplying the area of each land parcel by the dam capacity per unit area multiplier for that property (DNR, 2007) and then aggregating the values for all of the individual properties. The maximum harvestable right volume across rural land in the region is about 131 GL. The estimate of current farm dam storage volume is about 94 GL utilising about 31 GL of the harvestable right volume. There are farm dams capturing more than the maximum harvestable right volume that was later defined by the Water Management Act. The available harvestable right volume is therefore about 100 GL. The projected increases for each of the four subcatchments are given in Appendix A. The total increase in farm dam storage volume by ~2030 is 13.2 GL or 14 percent of the existing total volume. Figure 3-1. Map of the modelling subcatchments. The calibration catchments are not shown as they are outside the region Model calibration Figure 3-2 compares the modelled and observed monthly runoff and daily flow duration curves for the two calibration catchments. The SIMHYD calibration can reproduce the observed monthly runoff series and the daily flow duration characteristic satisfactorily for Subcatchment (Gwydir) and reasonably for Subcatchment (Macquarie- Castlereagh). The volumetric constraint used in the model calibration also ensures that the total modelled runoff is within 5 percent of the total observed runoff. 28 Water availability in the Barwon-Darling June 2008 CSIRO 2008

39 The calibration to optimise Nash-Sutcliffe efficiency means that more importance is placed on the simulation of high runoff. Therefore SIMHYD modelling of medium and high runoff is better than the simulation of low runoff. Nevertheless, an optimisation to reduce overall error variance will result in some underestimation of high runoff and overestimation of low runoff as shown in the scatter plots comparing the modelled and observed monthly runoff and the daily flow duration curves (Figure 3-2). The disagreement between the modelled and observed daily runoff characteristics is only discernable for runoff that is exceeded less than 0.1 or 1 percent of the time. This is accentuated in the plots because of the linear scale on the y-axis and normal probability scale on the x-axis. 3 Rainfall-runoff modelling Figure 3-2. Modelled and observed monthly runoff and daily flow duration curve for the calibration catchments 3.3 Modelling results Scenario A historical climate and current development Figure 3-3 shows the spatial distribution of mean annual rainfall and modelled runoff for 1895 to 2006 across the region, Figure 3-4 shows the 1895 to 2006 annual rainfall and modelled runoff series averaged over the region, and Figure 3-5 shows the mean monthly rainfall and runoff averaged over the region for 1895 to The mean annual rainfall and modelled runoff averaged over the region are 328 mm and 6 mm respectively. The mean annual rainfall varies from about 600 mm in the east to less than 250 mm in the west. The modelled mean annual runoff varies from about 20 mm in the east to less than 5 mm in the west. Rainfall is low throughout the year but highest in summer and most of the runoff occurs in summer and early autumn. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 29

40 Rainfall, runoff and the fraction of rainfall that becomes runoff in the Barwon-Darling (particularly in the west) are amongst the lowest in the MDB. The region covers about 13.4 percent of the MDB but generates only about 2.8 percent of the total runoff. Rainfall and runoff can vary considerably from year to year with long periods over several years or decades that are considerably wetter or drier than others. The coefficients of variation of annual rainfall and runoff averaged over the region are 0.36 and 1.30 respectively. They are amongst the highest in the MDB. The 10 th percentile, median and 90 th percentile values across the 18 MDB regions are 0.22, 0.26 and 0.36 respectively for rainfall and 0.54, 0.75 and 1.19 for runoff. The mean annual rainfall and modelled runoff over the ten-year period 1997 to 2006 are 3 and 8 percent higher respectively than the long-term (1895 to 2006) mean values. However, because of the inter-annual variability and the relatively short ten-year period used as the basis for comparison, the 1997 to 2006 rainfall and runoff are not statistically different to the 1895 to 1996 mean values, even at a significance level of α = 0.2 (with the Student-t and Rank-Sum tests). Potter et al. (2008) present a more detailed analysis of recent rainfall and runoff across the MDB. 3 Rainfall-runoff modelling Figure 3-3. Spatial distribution of mean annual rainfall and modelled runoff averaged over Annual rainfall (mm) Annual runoff (mm) Figure annual rainfall and modelled runoff averaged over the region (the curve shows the low frequency variability) 30 Water availability in the Barwon-Darling June 2008 CSIRO 2008

41 Mean monthly rainfall (mm) J F M A M J J A S O N D Mean monthly runoff (mm) J F M A M J J A S O N D Figure 3-5. Mean monthly rainfall and modelled runoff (averaged over for the region) Scenario C future climate and current development Figure 3-6 shows the percent change in the modelled mean annual runoff averaged over the region under Scenario C relative to Scenario A for the 45 scenarios (15 global climate models (GCMs) for each of the high, medium and low global warming scenarios). The percent change in the mean annual runoff and rainfall from the corresponding GCMs are also tabulated in Table 3-1. The figure and table indicate that the potential impact of climate change on runoff can be very significant. However, there is considerable uncertainty in the estimates, with rainfall-runoff modelling with climate change projections from about half of the GCMs showing a reduction and the other half showing an increase in mean annual runoff. These results are different to the southern MDB where nearly all of the GCMs indicate that rainfall would decrease. 3 Rainfall-runoff modelling Because of the large variation between GCM simulations and the method used to obtain the climate change scenarios (Section 1.3.3), the biggest increase and biggest decrease in runoff come from the high global warming scenario. Only results from an extreme dry, mid and extreme wet variant are used in subsequent reporting (referred to as scenarios Cdry, Cmid and Cwet). For Scenario Cdry, results from the second highest reduction in mean annual runoff from the high global warming scenario are used. For Scenario Cwet, results from the second highest increase in mean annual runoff from the high global warming scenario are used. These are shown in bold in Table 3-1, with the Cdry, Cmid and Cwet scenarios indicating a -22, -2 and +50 percent change in mean annual runoff. By comparison, the range based on the low global warming scenario is -8 to +12 percent change in mean annual runoff. Figure 3-7 shows the mean annual runoff across the region under Scenario A and under scenarios Cdry, Cmid and Cwet. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 31

42 Percent change in mean annual runoff High global warming Medium global warming Low global warming 3 Rainfall-runoff modelling -70 ipsl giss_aom cnrm iap csiro mpi gfdl inmcm mri ncar_ccsm cccma_t63 ncar_pcm cccma_t47 miroc miub Figure 3-6. Percent change in mean annual runoff under the 45 Scenario C simulations (15 global climate models and three global warming scenarios) relative to Scenario A runoff Table 3-1. Summary results under the 45 Scenario C simulations (numbers show percent change in mean annual rainfall and runoff under Scenario C relative to Scenario A) High global warming Medium global warming Low global warming GCM Rainfall Runoff GCM Rainfall Runoff GCM Rainfall Runoff ipsl ipsl ipsl -3-9 giss_aom giss_aom giss_aom -4-8 cnrm cnrm cnrm -3-7 iap iap -2-7 csiro -2-3 csiro csiro -5-7 iap -1-3 mpi -6-7 mpi -4-5 gfdl -2-3 gfdl -6-6 gfdl -4-5 mpi -2-3 inmcm -3-1 mri -3-2 mri -1-1 mri -5-1 inmcm -2-2 inmcm -1-1 ncar_ccsm 3 14 ncar_ccsm 2 8 ncar_ccsm 1 3 cccma_t cccma_t cccma_t ncar_pcm 7 23 ncar_pcm 5 14 ncar_pcm 2 6 cccma_t cccma_t cccma_t miroc miroc 8 30 miroc 4 12 miub miub 8 37 miub Water availability in the Barwon-Darling June 2008 CSIRO 2008

43 3 Rainfall-runoff modelling Figure 3-7. Mean annual rainfall and modelled runoff under scenarios A, Cdry, Cmid and Cwet CSIRO 2008 June 2008 Water availability in the Barwon-Darling 33

44 3.3.3 Summary results for all modelling scenarios Table 3-2 shows the mean annual rainfall, modelled runoff and actual evapotranspiration under Scenario A averaged over the region, and the percent changes in the rainfall, runoff and actual evapotranspiration under scenarios C and D relative to Scenario A. The Cdry, Cmid and Cwet results are based on the modelled mean annual runoff, and the rainfall changes shown in Table 3-2 are the changes in the mean annual value of the rainfall series used to obtain the Cdry, Cmid and Cwet runoff. The changes in mean annual rainfall do not necessarily translate directly to the changes in mean annual runoff because of changes in seasonal and daily rainfall distributions. 3 Rainfall-runoff modelling Figure 3-8 shows the mean monthly rainfall and modelled runoff under scenarios A, C and D averaged over 1895 to 2006 for the region. Figure 3-9 shows the daily rainfall and flow duration curves under scenarios A, C and D averaged over the region. The modelling results for all the subcatchments are summarised in Appendix A. The results show that seasonality of rainfall and runoff are not projected to change and that daily extremes of runoff could be either greater or smaller than present. The Cmid (or Cdry or Cwet) results are from rainfall-runoff modelling using climate change projections from one GCM. As Scenario Cmid is chosen based on mean annual runoff (Section 3.3.2), the comparison of monthly and daily results under Scenario Cmid relative to Scenario A in Figure 3-8 and Figure 3-9 should be interpreted cautiously. However, the C range results shown in Figure 3-8 are based on the second driest and second wettest results for each month separately from the high global warming scenario, and the C range results shown in Figure 3-9 are based on the second lowest and second highest daily rainfall and runoff results at each of the rainfall and runoff percentiles from the high global warming scenario. The lower and upper limits of C range are therefore not the same as scenarios Cdry and Cwet reported elsewhere and used in the river system and groundwater models. Scenario B modelling is not performed because the mean annual rainfall and modelled runoff for the ten-year period 1997 to 2006 are not significantly different (at statistical significance level of α = 0.2 with the Student-t and Rank-Sum tests) from the long-term (1895 to 2006) mean values. The Scenario B results would therefore be essentially the same as the Scenario A results. The modelling results indicate a best estimate of 2 percent reduction in mean annual runoff by ~2030. However, there is considerable uncertainty in the climate change impact estimate. Extreme estimates range from -22 to +50 percent. There is no projected growth in commercial forestry plantations. The total farm dam storage volume is projected to increase by 13.2 GL by ~2030. The additional impact of farm dams (over and above the best estimate climate change impact) is less than 0.5 percent, and because of rounding, does not show at the level of precision used for reporting change in Table 3-2. The best estimate of the combined impact of climate change and farm dam development is therefore a 2 percent reduction in mean annual runoff; the extreme estimates range from -23 to +49 percent. Table 3-2. Water balance over the entire region by scenario Scenario Rainfall Runoff Evapotranspiration mm A percent change from Scenario A B Cdry -13% -22% -13% Cmid -3% -2% -3% Cwet 13% 50% 12% Ddry -13% -23% -13% Dmid -3% -2% -3% Dwet 13% 49% 12% 34 Water availability in the Barwon-Darling June 2008 CSIRO 2008

45 Mean monthly rainfall (mm) Scenario C range Scenario A Scenario Cmid J F M A M J J A S O N D Mean monthly runoff (mm) Scenario C range Scenario A Scenario Cmid Scenario Dmid J F M A M J J A S O N D Figure 3-8. Mean monthly rainfall and modelled runoff under scenarios A and C averaged over across the region (C range is based on the consideration of each month separately the lower and upper limits in C range are therefore not the same as scenarios Cdry and Cwet) 3 Rainfall-runoff modelling Figure 3-9. Daily flow duration curves under scenarios A and C averaged over the region (C range is based on the consideration of each rainfall and runoff percentile separately the lower and upper limits in C range are therefore not the same as scenarios Cdry and Cwet) 3.4 Discussion of key findings The mean annual rainfall and modelled runoff averaged over the region are 328 mm and 6 mm respectively. The mean annual rainfall varies from about 600 mm in the east to less than 250 mm in the west. The modelled mean annual runoff varies from about 20 mm in the east to less than 5 mm in the west. Rainfall is low throughout the year but highest in summer, and most of the runoff occurs in summer and early autumn. Rainfall, runoff and the fraction of rainfall that becomes runoff (particularly in the west) are amongst the lowest in the MDB. The region covers about 13.4 percent of the MDB but generates only about 2.8 percent of the total runoff in the MDB. The mean annual rainfall and modelled runoff over the ten-year period 1997 to 2006 are 3 and 8 percent higher respectively than the long-term (1895 to 2006) mean values. However, because of the inter-annual variability and the relatively short ten-year period used as the basis for comparison, the 1997 to 2006 rainfall and runoff are not statistically different to the 1895 to 1996 mean values, even at a significance level of α = 0.2. The runoff for the four eastern subcatchments is estimated using model parameter values from calibration catchments more than 100 km away. Consequently, there is less confidence in the runoff estimates in this region compared to the data-rich regions in the eastern and southern MDB. The runoff estimates for the western half of the Barwon-Darling region are relatively poor because there are no small or medium sized gauged catchments within 250 km and runoff in the western subcatchment is modelled using default model parameter values. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 35

46 It is considerably more difficult to model runoff in the central, western and northern MDB because the region is drier, there are far fewer rainfall stations, and river flows are intermittent with most of the runoff occurring as infrequent floods. The best estimate would be a 2 percent reduction in mean annual runoff by ~2030. There is considerable uncertainty in the modelling results and extreme estimates range from -22 to +50 percent. These extreme estimates come from the high global warming scenario, and for comparison, the range from the low global warming scenario is a -8 to +12 percent change in mean annual runoff. The main sources of uncertainty are in the global warming projections and the global climate modelling of regional rainfall response to the global warming. The uncertainty in the modelling of climate change impact on runoff is small compared to the climate change projections. There is no projected growth in commercial forestry plantations. The total farm dam storage volume is projected to increase by 13.2 GL (14 percent) by ~2030. The best estimate of the combined impact of climate change and farm dam development is a 2 percent reduction in mean annual runoff. Extreme estimates range from -23 to +49 percent. The modelled reduction in mean annual runoff from the projected increase in farm dams alone is less than 0.5 percent. 3 Rainfall-runoff modelling There is considerable uncertainty in the projection of future increases in farm dam development and the impact of these new farm dams on runoff. The increase in farm dams is estimated by considering trends in historical farm dam growth and current policy controls and there is uncertainty both as to how landholders will respond to these policies and how governments may set policies in the future. 3.5 References Agrecon (2005) Agricultural Reconnaissance Technologies Pty Ltd Hillside Farm Dams Investigation. MDBC Project 04/4677DO. BRS (2005) 1993, 1996, 1998 and 2000 Land Use of the Murray-Darling Basin, Version 2. Resource Identifier: ID01. Online digital dataset and spatial data layer. File identifier: Product access: Chiew FHS, Vaze J, Viney N, Jordan P, Perraud J-M, Zhang L, Teng J, Pena J, Morden R, Freebairn A, Austin J, Hill P, Wiesenfeld C and Murphy R (2008) Rainfall-runoff modelling across the Murray-Darling Basin. A report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project. CSIRO, Australia. In prep. DNR (2007) Unregulated Flow Management Plan for the Northwest. New South Wales Department of Natural Resources, Sydney. Geosciences Australia (2007) Man made hydrology GIS coverage (supplied under licence to CSIRO). Australian Government, Canberra. IPCC (2007) Climate Change 2007: The Physical Basis. Contributions of Working Group 1 to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press. NSW Government (2000) Water Management Act 2000 No 92. New South Wales Parliament, December Available at Potter NJ, Chiew FHS, Frost AJ, Srikanthan R, McMahon TA, Peel MC and Austin JM (2008) Characterisation of recent rainfall and runoff across the Murray-Darling Basin. A report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project. CSIRO, Australia. 36 Water availability in the Barwon-Darling June 2008 CSIRO 2008

47 4 River system modelling This chapter includes information on the river system modelling for the Barwon-Darling region. It has four sections: a summary an overview of the regional modelling approach a presentation and description of results a discussion of key findings. The information in this chapter comes from the calibrated models (IQQM) of the Barwon-Darling River system developed and maintained by the New South Wales Department of Water and Energy (DWE) (DNR, 2006a). 4.1 Summary Issues and observations River system modelling for the Barwon-Darling region considers six modelling scenarios: Scenario O This scenario runs from 1 February 1891 to 30 June It is based on the original IQQM for the Barwon-Darling that was run from 1922 to The base case model provided for this project had changes to inflows due to recent irrigation development in the tributaries, re-calibration of tributary models, and inclusion of the current Water Sharing Plans or Resource Operation Plans. The revised model represents 2004/05 development in the Barwon-Darling region. Scenario A0 This scenario differs from Scenario O as it links with tributary river models for other regions. This scenario still represents 2004/05 development in the Barwon-Darling region and covers the extended common historical climate period for this project (1 June 1895 to 30 June 2006). It does not include the effects of current groundwater extraction at dynamic equilibrium. These are too small within the region to warrant inclusion. However, effects of these are significant in some upstream regions. Scenario A historical climate and current development This scenario incorporates Scenario A0 and the effects of current groundwater extraction at dynamic equilibrium. This is a baseline against which scenarios B, C and D are compared. Scenario P without-development This scenario is the same as Scenario A0, except that the current levels of development such as demand nodes are inactivated to represent without-development conditions. Natural water bodies, fixed diversion structures and existing catchment runoff characteristics are not adjusted. Scenario C future climate and current development Scenarios Cwet, Cmid and Cdry represent a range of future climate conditions that are derived by adjusting the historical climate and flow inputs used in Scenario A. The level of development is the same as Scenario A. For each Scenario Cwet, Cmid and Cdry, without-development model runs are also undertaken; these use Scenario C climate and Scenario P development conditions. Scenario D future climate and future development Scenarios Dwet, Dmid and Ddry incorporate Scenario C with flow inputs adjusted for 2030 projected development in farm dams, commercial forestry plantations and groundwater. The farm dam and commercial forestry plantation projections for this region are discussed in Chapter 3 while groundwater development in this region is discussed in Chapter 6. Development in upstream regions is described in the relevant reports. 4 River system modelling These scenarios may not eventuate but they encompass consequences that might arise if no management changes were made. Consequently results from this assessment highlight pressure points in the system, both now and in the CSIRO 2008 June 2008 Water availability in the Barwon-Darling 37

48 future. This assessment does not elaborate on what management actions might be taken to address any of these pressure points. The change in inflows between scenarios reported in this chapter differs from the change in runoff reported in Chapter 3 as the majority of inflows to the Barwon-Darling region are generated in the contributing upstream regions. The groundwater assessment (Chapter 6) found that estimated impacts of groundwater extraction on streamflow and projected future extraction estimates have a low level of confidence. Given these findings and the insignificance of existing impacts, groundwater was not modelled in the Barwon-Darling river model. Groundwater use and growth in the upstream regions is described in the relevant reports. The Barwon-Darling region is described by two models, the Barwon-Darling river system model and the Menindee connection model. These models: 4 River system modelling include the outputs of upstream river models (except the Paroo model, as the Paroo does not significantly contribute flow to the Darling River). The upstream river models for New South Wales generally represent the current level of water use although the definition of current differs slightly between regions. The upstream river models for Queensland represent a range of different water resource development conditions related to Queensland Government planning approaches under the current Water Resource Plans and Resource Operations Plans. The range is from the current level of water use (even where less than the currently allowable level of use) to the maximum permissible level of use under current planning arrangements. Project reports for the upstream regions provide details of the upstream models. Therefore, the simulated upstream regional inflows for current development reflect, in some cases, a level of development greater than has yet been realised represent the 2004/05 level of development of irrigators in the Barwon-Darling region. This includes farm infrastructure, irrigated areas and crop mix. The models are calibrated to represent the farm management practices. Modelled demands may not match history of use as farm development is not static over time simulate irrigation demands using a soil moisture accounting model with areas, soil depth, crop mixes, farm dams and farm infrastructure that best represents current levels of development use licence conditions to control access to the unregulated water supply and thresholds for access to floodplain harvesting include risk functions that adjust areas planted according to water availability. Consequently the models represent the change in demand as a function of available resource and climatic conditions apply a water sharing roster between major users in the Brewarrina to Louth reach. Access rules for irrigators that make provision for environmental flows in the Barwon-Darling River were established in 2000/01. The model used in this project includes those rules but differs from the model used to develop the access conditions, as it reflects irrigation development between 2000/01 and 2004/05 and includes changes to inflows time series as described under Scenario O above. Analysis of the without-development flows along the Barwon-Darling system indicates that it changes from a gaining to a losing stream (point of maximum average annual flow) at the Bourke gauge (425003). The without-development average annual flow at Bourke over the modelling period is 3485 GL/year Key messages Current average surface water availability for the entire Darling Basin (assessed at Bourke) is 3515 GL/year and 99 percent of this water is generated in regions upstream of the Barwon-Darling region. Current average surface water use in the Barwon-Darling region is 230 GL/year. Licences within the region are fully utilised. Note however, access rules for Barwon-Darling irrigators were altered in July 2006 to cap use at 173 GL/year. Current total average surface water use across the entire Darling Basin reduces streamflow at Bourke by 1365 GL/year. The relative level of use for the Darling Basin is thus 39 percent. This is a high level of development. 38 Water availability in the Barwon-Darling June 2008 CSIRO 2008

49 Water resource development across the Darling Basin has not significantly altered the seasonality of streamflow in the Barwon-Darling region but has reduced the magnitude of two-year average return interval floods by 41 percent, and the magnitude of five- and ten-year average return interval floods by around 30 percent. These reductions in flood magnitude are considerably greater than the reductions likely under the best estimate 2030 climate. Under the best estimate 2030 climate, average surface water availability (assessed at Bourke) would be reduced by 8 percent and end-of-system flows for the Barwon-Darling region would be reduced by 10 percent. Total average surface water use within the region would increase by 2 percent due to increased evaporation from on-farm storages. The impacts of climate change vary between water products: water use under Class A, B and C licences would increase by 11, 2 and less than 1 percent, respectively. Floodplain harvesting would reduce by 2 percent. The relative level of use for the entire Darling Basin would then be a very high 41 percent. Under the wet 2030 climate extreme, average surface water availability would increase by 31 percent, surface water use within the region would increase by 3 percent and end-of-system flows would increase by 47 percent. Under the dry 2030 climate extreme, average surface water availability would decrease by 27 percent, surface water use within the region would increase by 5 percent and end-of-system flows would decrease by 35 percent. Projected 2030 development (groundwater extraction and additional farm dams in upstream regions) would reduce inflows to the Barwon-Darling region (under the best estimate 2030 climate) by 3 percent or 78 GL/year on average. Additional farm dams and groundwater extraction would be responsible for about equal shares of this impact. Diversions would be reduced by 1 percent compared to current conditions and end-of-system flows would be reduced by 3 percent in addition to best estimate climate change impacts. The relative level of use for the entire Darling Basin would then be 42 percent Robustness 4 River system modelling A model robustness test was not done as the Barwon-Darling is an unregulated system, access is constrained by flow thresholds and model instability due to alternate climate inputs would not be expected. The model response to increases and decreases in inflow was reasonable with the change in end-of-system flow consistent with the change in inflow. 4.2 Modelling approach The following section provides a summary of the generic river modelling approach, a description of the Barwon-Darling and Menindee connection models and how the river models were setup for this project. Overviews of key model features and management rules are in Table 4-1 to Table 4-3. Refer to Chapter 1 for more information on the overall project methodology General River system models that describe current infrastructure, water demands, and water management and sharing rules were used to assess the implications of the changes in inflows on the reliability of water supply to users. Given the time constraints of the project, and the need to link the assessments to state water planning processes, it was necessary to use the river system models currently employed by state agencies and the Murray-Darling Basin Commission. The main models are IQQM, REALM, MSM-BigMod, WaterCress and a model of the Snowy Mountains Hydro-electric Scheme Model description The Barwon-Darling region is represented by the Barwon-Darling IQQM (up to Wilcannia) and the Menindee connection model that transfers flow to the Lower Darling/Murray River model at Menindee (Figure 4-1). CSIRO 2008 June 2008 Water availability in the Barwon-Darling 39

50 Barwon-Darling model The Barwon-Darling model is a daily IQQM V representation of the Barwon-Darling region from Mungindi gauge to two outlets: the Darling River at Wilcannia (425008) and Talyawalka Creek at Barrier Highway (425018). The Barwon-Darling model receives the majority of its inflows from seven upstream regions: 4 River system modelling Warrego region at Fords Bridge gauge (423001) Condamine-Balonne region at Culgoa River gauge (422006), Bokhara River gauge (422005) and Narran Lake overflow Moonie region at Gundablouie gauge (417001) Border Rivers region at three locations: Barwon River at Mungindi gauge (416001), Boomi River at Neeworra gauge (416028) and the Little Weir River Gwydir region at three locations: Gil Gil Creek after the return of Gingham Watercourse, Gwydir River at Collymongle gauge (418031), and Mehi River at Collarenebri gauge (418055) Namoi region at two locations: Namoi River at Goangra gauge (419026) and Pian Creek at Waminda gauge (419049). These locations are upstream of the end of the Namoi region located at the Walgett gauge (419091). These gauges are used as connection points as they are more reliable Macquarie-Castlereagh region at five locations: Macquarie at Carinda (Bells Bridge) gauge (421012), Castlereagh at Coonamble gauge (420005), Marra Creek at Billybingbone Bridge gauge (421107), Marthaguy at Carinda gauge (421011) and Bogan at Gongolgon (421023). All of these inflows are from daily IQQMs that feed directly into the Barwon-Darling model. Contributions from each upstream region as shown in Table 4-5 are not necessarily the same as the end-of-system flows reported for these regions due to factoring required for the Barwon-Darling model to achieve main streamflow calibration. Usage in each region is accounted in each of the respective regions. Inter-valley transfer of water occurs between the Barwon-Darling region and the Gwydir region; however this transfer is not interactively modelled, as a daily feedback loop does not significantly impact results. The model represents the Barwon-Darling system with 337 links and 338 nodes arranged into 57 river sections. There are no public storages. The model includes three floodplain storages in the lower system (Table 4-1). Water use is modelled by 94 irrigation nodes. The water is shared according to three licence classes: A, B and C. These licence conditions restrict access using flow thresholds associated with gauges both upstream and downstream of each irrigator node, annual entitlement, pump capacity, volumes in Menindee Lakes, Bourke water users sharing roster and intended use. Some irrigators have multiple licence classes. The major irrigators hold 49 licences two A Class licences, 34 B Class licences and 13 C Class licences. Class A licences allow water to be taken at lower river flows than the Class B and C licences. Water is extracted, when flows are sufficient, under a Class B licence in preference to a Class A licence. This enables the Class A licence to be reserved for lower flow periods. Access for Class C licences is at higher river flows than Class B licences. As described in Table 4-2 water users are classified as major irrigators, reach irrigators and floodplain harvesters. Major irrigators grow in excess of 20 ha and are required to meter diversions. Other small irrigators who do not irrigate large areas are aggregated by reaches and are referred to as reach irrigators. Floodplain harvesting is water obtained by pumping or direct inflow into on-farm storages during high river flows. As the river is unregulated, minimum flow requirements or environmental contingency allowances are controlled by the licence class access rules. As the water access rules for some irrigators in the Barwon-Darling are a function of Menindee Lakes total storage volume, an iterative solution was used between the Barwon-Darling models and the Murray model. In this project five iterations were found sufficient to enable convergence of the feedbacks between these models. 40 Water availability in the Barwon-Darling June 2008 CSIRO 2008

51 Menindee connection model The original Menindee connection model is an IQQM V representation of the Darling River from Wilcannia to the Menindee Lakes inflow at Lake Wetherell and the flow in Talyawalka Creek at the Railway Bridge. The model has two separate inflows from the Barwon-Darling model and no local inflows. The model represents the linkage between the Barwon-Darling system and modelling of the Lower Darling/Murray system. The Murray Simulation Model starts at Menindee Lakes and has one combined input from the Barwon-Darling that is split internally to separate flows for Talyawalka Creek that bypass Menindee Lakes. The model has 75 links and 76 nodes arranged into 11 river sections. There are no water users or public storages. There are three floodplain storages in the model. The original Menindee connection model was modified to be run in IQQM V for this project. It includes an inflow time series derived by the Murray-Darling Basin Commission that is used to ensure the peak inflows into Menindee Lakes match the historical gauged inflows. 4 River system modelling Figure 4-1. River system map showing subcatchments, inflow and demand nodes, links and gauge locations CSIRO 2008 June 2008 Water availability in the Barwon-Darling 41

52 Table 4-1. Storages in the river system model Storage Average annual net evaporation GL GL/y Natual water bodies Warrego storage Floodplain storages combined lakes 48.4 Lake Wongalara 223 Lake Poopelloe 928 Lakes on Talyawalka Creek 669 Region total 1850 Private storage River system modelling Number of nodes Table 4-2. Modelled water use configuration Licence entitlement Pump constraints Developed area GL/y ML/day ha Model notes Major irrigators ,694 40,075 On-farm storage at nodes Floodplain harvesters 25 na 11,920 Reach irrigators ,734 Sub-total ,162 41,810 na - not applicable Table 4-3. Model water management Minimum flow requirements Modelled and implemented as part of the licence conditions (pump thresholds) Unregulated flow plan for the north-west Water sharing Bourke Water Users Association Accounting system Annual accounting Access conditions associated with flows at both upstream and downstream river gauges for each individual irrigator Not included in the model Brewarrina to Louth shared based on individual's percentage of total available water Model reflects an annual accounting system as relevant pre-july Model setup The Barwon-Darling and Menindee connection river models and respective IQQM V and V executable code were obtained from DWE. The original Barwon-Darling model that was used to develop licence access conditions was run for the period 1 January 1922 to 30 June This model was revised by DWE to run from 1 February 1891 to 30 June It reflects changes in development and inflows from upstream tributaries. The baseline model used for this project included 2004/05 level of development in the Barwon-Darling region and was run for the period 1 May 1895 to 30 June Inputs from the upstream regions were derived from the linked model platform developed for this project. The baseline model for the Menindee connection model used outputs from the Barwon-Darling baseline model. A without-development version of the Barwon-Darling model was created by setting all irrigators as inactive and using without-development inflows from all upstream models. No modifications for the without-development scenario were required for the Menindee connection model as there is no development in the model. The Barwon-Darling system does not contain any public storages and river storage is relatively small compared to inflows. Consequently the warm-up period from 1 June 1895 to 30 June 1895 is sufficient to set the initial state of the river. Table 4-4 summarises the model setup information for the original models and baseline models used for this project. 42 Water availability in the Barwon-Darling June 2008 CSIRO 2008

53 Table 4-4. Model setup information Original model Version Start date End date Barwon-Darling IQQM /02/ /06/2002 Menindee connection IQQM /02/ /06/2002 Connection Barwon-Darling Menindee connection Baseline models Connects to Menindee model at Wilcannia (425008) and Talyawalka Creek at Barrier Highway (425018) Two outflows to the Murray model Lake Wetherell inflow and Talyawalka Creek at Railway Bridge Warm-up period 01/06/ /06/1895 Barwon-Darling IQQM /07/ /06/2006 Menindee connection IQQM /07/ /06/2006 Connection Barwon-Darling Menindee connection Modifications Model Connects to Menindee model Outflows into the Murray region model. Flow at Lake Wetherell and Talyawalka Creek at Railway Bridge combined with inflow series from Murray-Darling Basin Commission The outflow from the Wilcannia to Menindee model were combined with the inclusion of flows from Murray-Darling Basin Commission to have one inflow to Menindee Lake Data One residual inflow was extended back to 01/06/1895 Inflows Inflows to the models were automatically generated from the linked model platform 4 River system modelling 4.3 Modelling results River system water balance The mass balance table (Table 4-5) shows the net fluxes for the Barwon-Darling region. Scenario O, Scenario A0 (without groundwater at dynamic equilibrium) and Scenario A (with groundwater at dynamic equilibrium) fluxes are displayed as GL/year. All other scenarios are presented as a percentage change from Scenario A. The averaging period and inflows for Scenario O differ to the other scenarios as this scenario uses inflows supplied by DWE whereas inflows for the other scenarios came from the linked model and were for the common modelling period. The directly gauged inflows represent the contribution from each upstream region. There are two indirectly gauged inflows within the region that represent flow to achieve calibration at mainstream gauges. Diversions are listed by the different water products in the region. End-of-system flows shown are the combined outflows from the Menindee connection model. The net evaporation from the river, floodplains and lakes is included. The net evaporation from on-farm storages of irrigators is displayed below the mass balance table but is not included in the mass balance as these are indirectly included in diversions for irrigators. The model estimates the amount of rainfall harvesting and this value is displayed below the mass balance table but is not part of the water balance. Appendix B contains mass balance tables for the four subcatchments in the models. The mass balance of each of these river reaches and the overall mass balance were checked by taking the difference between total inflows and outflows of the system. In all cases the mass balance error was zero. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 43

54 Table 4-5. River system model average annual water balance under scenarios O, P, A0, A, C and D 4 River system modelling O P A0 A Cwet Cmid Cdry Dwet Dmid Ddry Model start date 01/2/ /7/ /7/ /7/ /7/ /7/ /7/ /7/ /7/ /7/1895 Model end date 30/6/ /6/ /6/ /6/ /6/ /6/ /6/ /6/ /6/ /6/2006 Storage volume GL/y percent change from Scenario A Change over period % 1% 3% -8% 1% 3% Inflows Contributions from upstream tributaries Directly gauged Warrego % -7% -27% 47% -7% -27% Condamine-Balonne % -12% -36% 23% -12% -36% Moonie % -20% -39% 43% -22% -40% Border Rivers % -12% -34% 23% -15% -37% Gwydir % -7% -29% 35% -9% -31% Namoi % -9% -41% 56% -12% -44% Macquarie-Castlereagh % -9% -29% 40% -13% -33% Sub-total (gauged) % -10% -34% 39% -13% -37% Indirectly gauged * % -7% -27% 46% -8% -28% Sub-total % -9% -33% 40% -12% -36% Diversions Major irrigators Class A % 11% 40% -10% 18% 47% Class B % 2% -3% 0% 1% -5% Class C % 0% -8% 5% -1% -10% Floodplain harvesters % -2% -28% 31% -3% -29% Reach irrigators % 2% -3% 5% 0% -4% Sub-total % 2% -5% 2% <1% -6% Outflows End-of-system flow % -10% -35% 43% -13% -38% Net evaporation** Warrego % 0% -11% 13% 0% -11% From river % 4% 7% 1% 4% 7% Floodplain lakes % -8% -42% 43% -11% -44% Total evaporation % -1% -14% 19% -2% -15% Sub-total % -9% -34% 41% -12% -36% Unattributed fluxes Total % -12% -39% 48% -16% -42% Evaporation from on-farm storages ** % 2% -6% 6% 0% -7% Rainfall harvesting % -3% -14% 7% -4% -15% * Scenario O indirectly gauged inflows and unattributed fluxes differ from the other scenarios as the Menindee connection model includes additional inflows and losses that account for differences between modelled and observed flows at Wilcannia and Menindee ** Evaporation from private licensed storages (GL/year) is not included as it is already accounted in diversions Inflows and water availability Inflows There are several ways that the total inflows into the river system can be calculated. The obvious way would be to sum all of the inflows in the model. The approach used to calibrate these inflows varies considerably between model implementations. In some cases inflows are inflated and subsequently compensated for by loss relationships. In other cases the losses are inherent in the inflows. Totalling inflows does not provide a consistent assessment of total river system inflows across different models because of the different approaches to calibration. 44 Water availability in the Barwon-Darling June 2008 CSIRO 2008

55 An alternative to simply totalling modelled inflows is to locate the point of maximum average annual flow in the river system under without-development conditions. The gauge with maximum average annual flow is a common reference across all models because they are calibrated to achieve mass balance at mainstream gauges. The without-development scenario removes the influences of upstream extractions and regulation and gives a reasonable indication of total inflows. However, the upstream models that provide without-development inputs to the Barwon-Darling model include subcatchment inflows with existing land use (farm dams and forest cover), groundwater use impacts and groundwater losses not implicitly included in calibration of the upstream river models. Thus Scenario A is not a representation of pre-european settlement conditions. All upstream contributing models and the Barwon-Darling models were run for without-development current and future climate scenarios. A comparison between scenarios for reaches along the Barwon-Darling River is presented in Figure 4-2. This shows that the maximum average annual mainstream flow occurs at the Bourke gauge (425003) with a value of 3485 GL/year for Scenario A. Average annual flow (GL) C range 4000 Cmid 3500 A EOS 4 River system modelling Figure 4-2. Transect of total river flow under scenarios A and C Water availability Water availability is a function of climate, and thus is assessed for without-development conditions for scenarios A, B and C. Total surface water availability (GL/year) is presented in Table 4-6 for each scenario. It is assessed as the without-development maximum mainstream flow (Bourke gauge) plus the reduction in subcatchment inflow that is implicitly included in the upstream tributary regional river model calibrations as a result of groundwater use. The reduction is the combination of groundwater use from the Namoi and Condamine regions, adjusted to an equivalent use at the point of maximum flow in the Barwon-Darling. No adjustments have been made for the impacts of existing farm dams or changes in forest cover in determining surface water availability under scenarios A and C. These impacts are not included as they are either insignificant, or difficult to quantify and are not relevant for guiding future policy. Table 4-6. Annual water availability under scenarios A and C (assessed for without-development conditions, which for Scenario A is synonymous with Scenario P) Water Availability A Cwet Cmid Cdry Flow contributions of upstream regions GL/y Warrego (via Norooma and Widgeegora creek inflow) CSIRO 2008 June 2008 Water availability in the Barwon-Darling 45

56 Water Availability A Cwet Cmid Cdry Condamine-Balonne Flow reductions caused by current groundwater use implicit in the model Moonie Border Rivers Gwydir Namoi Flow reductions caused by current groundwater use Flow reductions caused by current groundwater use implicit in the model Macquarie-Castlereagh Barwon-Darling flow contribution Modelled without development maximum average mainstream flow percentage change from Scenario A Change in average surface water availability 31% -8% -27% 4 River system modelling The proportional input of each upstream contributing region to the total water availability (under Scenario A) is shown in Table 4-7. This is estimated by removing each of the upstream without-development inflows one at a time and noting the change in flow at the point of maximum flow. The ratio of each contribution to the sum of all contributions is presented. Table 4-7. Contribution of each region to water availability under Scenario A Region Percent Warrego <0.1% Condamine-Balonne 15.4% Moonie 3.2% Border Rivers 20.0% Gwydir 14.2% Namoi 23.4% Macquarie-Castlereagh 22.6% Barwon-Darling 1.2% A time series of total annual surface water availability under Scenario A is shown in Figure 4-3. The lowest annual water availability was 274 GL in 1919 while the greatest annual water availability was 23,542 GL in Figure 4-4 shows the difference in total annual surface water availability under Scenario C relative to Scenario A. Annual water availability (GL) Figure 4-3. Water availability under Scenario A (assessed for without-development conditions) 46 Water availability in the Barwon-Darling June 2008 CSIRO 2008

57 Annual water availablity (GL) C range Cmid Figure 4-4. Time series of change in total surface water availability under Scenario C relative to Scenario A (assessed for without-development conditions) Consumptive water use Diversions 4 River system modelling Table 4-5 shows that total diversions in the Barwon-Darling region increase by 2 percent under Scenario Cmid despite the reduction in total inflows. The increase is attributed to increased demands associated with higher evaporation from on-farm storages under that climate scenario. Diversions under Scenario Dmid are 2 percent lower than under Scenario Cmid and are less than 1 percent lower than under Scenario A. The reduction in inflows under Scenario Dmid has reduced the frequency of times when pumping can occur. Table 4-8 shows the total average annual diversions for each subcatchment under Scenario A and the percentage change under all other scenarios compared to Scenario A. Changes in diversions are greatest for the most downstream reach (Bourke to Wilcannia). Table 4-8. Change in total diversions in each subcatchment under scenario C and D relative to Scenario A Reach A Cwet Cmid Cdry Dwet Dmid Ddry GL/y percent change from Scenario A Mungindi to Walgett % 6% 2% -4% 4% 1% Walgett to Bourke % 1% -4% 3% 0% -6% Bourke to Wilcannia % -1% -11% 7% -3% -13% Total % 2% -5% 2% 0% -6% Figure 4-5 shows total average annual diversions under scenarios A, C and D for subcatchment reaches. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 47

58 (a) (b) Average annual diversions (GL) C range Cmid A Mungindi to Walgett to Bourke to Average annual diversions (GL) D range Dmid A Mungindi to Walgett to Bourke to Walgett Bourke Wilcannia Walgett Bourke Wilcannia 4 River system modelling Figure 4-5. Total average annual diversions for subcatchments under (a) scenarios A and C and (b) scenarios A and D Figure 4-6 shows the annual time series of total diversions under Scenario A and the difference from Scenario A under scenarios C and D. The maximum and minimum diversions under Scenario A are 423 GL in 1920 and 46 GL in 1919 respectively. The 1920 use is higher as it is a wet year that follows the driest year on record. As on-farm storages are low at the start of the year the high usage reflects the filling of these storages. 48 Water availability in the Barwon-Darling June 2008 CSIRO 2008

59 (a) Scenario A 500 Annual diversions (GL) (b) Scenario Cwet (c) Scenario Dwet Annual difference (GL) Annual difference (GL) River system modelling (d) Scenario Cmid (e) Scenario Dmid Annual difference (GL) Annual difference (GL) (f) Scenario Cdry (g) Scenario Ddry Annual difference (GL) Annual difference (GL) Figure 4-6. Total diversions under (a) Scenario A and difference between total water use under (b) Scenario Cwet, (c) Scenario Dwet, (d) Scenario Cmid, (e) Scenario Dmid, (f) Scenario Cdry and (g) Scenario Ddry CSIRO 2008 June 2008 Water availability in the Barwon-Darling 49

60 Level of use The relative level of surface water use is indicated by the ratio of average surface water use to average surface water availability. It is assessed here for the entire Darling Basin and thus considers water availability and use across both the Barwon-Darling region and upstream regions. Surface water use within the Barwon-Darling region is simply the total diverted volume across the full range of water products; there is no impact on surface water due to groundwater extraction within the Barwon-Darling region. Surface water use has been reported for upstream regions in other project reports and includes: total diversions in the upstream regions, streamflow leakage induced by groundwater use in the upstream regions, inflow impacts due to groundwater use in subcatchments and inflow impacts due to additional future farm dams in subcatchments. 4 River system modelling Use values for the upstream regions were adjusted (where necessary) to provide an equivalent use at the point of maximum flow (Bourke). This adjustment was the product of (i) a transmission efficiency between the end-of-system for each upstream region and the point of maximum flow in the Barwon-Darling, and (ii) a usage factor representing the effect of usage in each upstream region on that region s end-of-system flow. Transmission efficiencies are the ratio between a region s end-of-system flow and the reduction in flow at Bourke when that region is omitted from the linked surface water modelling. This assumes that the transmission efficiency is the same between developed and without-development scenarios, and assumes that the transmission efficiency for Scenario A is applicable to all other scenarios. Transmission efficiencies should in fact differ somewhat between scenarios, however, these differences will be considerably less than the flow changes assessed between scenarios due to development and future climate change. The transmission efficiencies are also assumed to apply equally to each of the separate components for surface water use within an upstream region. The usage factor for a region is the difference in the end-of-system flow between with- and without-development scenarios. The usage factor for Scenario A is assumed to be a reasonable approximation for all other scenarios. Table 4-9 shows the relative level of surface water use for each of the scenarios. The current relative level of use for the Darling Basin is high with 39 percent of the total available surface water being diverted for use. Table 4-9. Relative level of use under scenarios A, C and D A Cwet Cmid Cdry Dwet Dmid Ddry Total surface water availability GL/y Streamflow use in region Total net diversions Usage contribution of upstream regions Warrego (via Norooma and Widgeegora creeks) Condamine-Balonne Moonie Border Rivers Gwydir Namoi Macquarie-Castlereagh Total upstream use Total use percent Relative level of use 39% 32% 41% 45% 33% 42% 46% 50 Water availability in the Barwon-Darling June 2008 CSIRO 2008

61 Use during dry periods Table 4-10 shows the average annual use for all Barwon-Darling water products, as well as the average annual use for the lowest one-, three- and five-year periods under Scenario A and the percentage change from Scenario A under each other scenario. The results for Scenario Cmid indicate the impact on water use during these dry periods is small, most probably because the low flow frequency (that is, flows below cease-to-pump thresholds) do not change substantially under Scenario C. More significant changes are seen under the extreme climate scenarios. Table Indicators of use during dry periods under scenarios A, C and D Annual diversion A Cwet Cmid Cdry Dwet Dmid Ddry GL/y percent change from Scenario A Lowest 1-year period % -3% -53% 44% -12% -58% Lowest 3-year period % 0% -17% 8% -5% -23% Lowest 5-year period % 0% -11% 5% -1% -13% Average % 2% -5% 2% 0% -6% Reliability In the Barwon-Darling region the baseline model has a total active irrigation entitlement of GL/year and an average annual diversion of GL/year. Access rules for Barwon-Darling irrigators were altered in July 2006 (DNR, 2006b) to cap use at 173 GL/year to comply with the Murray-Darling Basin Ministerial Council Cap on Surface Water Diversions. Additionally, a continuous accounting system was introduced that allows users to carry over unused entitlement to the next year. The effect of these changes is that irrigation entitlement is now aligned with long-term Cap use, and the reliability of diversions under Scenario A is effectively 100 percent (or 1.0) over the long term. This reliability under Scenario A does not imply that water users will get their full revised entitlement in every year because climatic variability will still impact on water availability as shown by Table However, over the long term water users will get their full revised entitlement. 4 River system modelling Despite being an unregulated system the reliability of water for diversions in the Barwon-Darling region is better than some of the upstream regulated systems due to the contribution of many tributaries and the major storages on them, and the ability of irrigators to store water in their on-farm storages that buffers impacts of flow variability under these scenarios. The change in total diversions for the region under climate change and development scenarios is not significant (<2 percent). However, as seen in Table 4-11 the impact on different water products varies. Under scenarios Cwet, Cdry, Dwet and Ddry the effect of diversions on each water product varies, with the largest impact on floodplain harvesting. Major irrigators in the Barwon-Darling region hold both Class A and Class B licences. Extractions are taken under Class B licence in preference to a Class A licence thus enabling the Class A licence to be reserved for lower flow periods. This behaviour is reflected in the results in Table 4-11 that show under scenarios Cwet and Dwet more extractions are taken under the Class B licence and less taken under Class A licence; conversely under scenarios Cdry and Ddry more water is taken under a Class A than under a Class B licence. Diversions for Class B licences represent the major extractions accounting for more than 70 percent of total diversions for the region. Therefore the impact on Class B diversions under climate and development scenarios is similar to impacts on total diversions (refer to Table 4-5 and Table 4-11). Access for Class C licences is at higher river flows than Class B; similarly diversions for floodplain harvesting are subject to the occurrence of large river flows. For this reason the reliability as shown in Table 4-11 has a greater change in Class C and floodplain harvesting under scenarios Cwet and Dwet compared with Class B diversions. Under these scenarios relatively less water is extracted for Class B as farm demand is supplemented by Class C diversions and floodplain CSIRO 2008 June 2008 Water availability in the Barwon-Darling 51

62 harvesting. Conversely under scenarios Cdry and Ddry there is less water taken by Class C and floodplain harvesting than Class B. Figure 4-7 shows that the frequency distribution of extractions remains similar under future scenarios, with some reduction of diversions at both ends of the distribution. Table Average reliability of water products under scenarios C and D relative to Scenario A Cwet Cmid Cdry Dwet Dmid Ddry percent change compared to diversions under Scenario A Major irrigators Class A 86% 111% 140% 90% 118% 147% 4 River system modelling Annual diversion (GL) (a) Class B 101% 102% 97% 100% 101% 95% Class C 106% 100% 92% 105% 99% 90% Floodplain harvesting 132% 98% 72% 131% 97% 71% Reach irrigators 106% 102% 97% 105% 100% 96% C range Cmid A Annual diversion (GL) (b) D range Dmid A 0 0% 20% 40% 60% 80% 100% Percent of years equal or exceeded 0 0% 20% 40% 60% 80% 100% Percent of years equal or exceeded Figure 4-7. Reliability of Class A access for irrigators under scenarios A, C and D River flow behaviour Three different indicators of flow characteristics are provided: daily flow duration curves, seasonal plot and daily event frequency. These are considered for two locations in the river: at the point of maximum flow (Bourke), and end-of-system at Menindee. The flow regime varies between the selected locations. Flow characteristics at Bourke Figure 4-8 shows the daily flow duration curves under scenarios A and P and the range of impacts under scenarios C and D at Bourke gauge. The flow duration curves show the change in frequency between scenarios for a given flow. The vertical difference between flow duration curves shows the change in mass between scenarios although care needs to be taken as the plots use a logarithmic scale that distorts the difference of lower flows. Development has increased the volume of low flows and made a more substantial reduction to mid-range flows. This pattern is predicted to continue in future scenarios. 52 Water availability in the Barwon-Darling June 2008 CSIRO 2008

63 Daily flow (ML) C range Cmid A P Daily flow (ML) D range Dmid A P Percent time flow is exceeded Percent time flow is exceeded Figure 4-8. Daily flow duration curves at Bourke gauge (425003) under scenarios P, A, C and D Figure 4-9 shows the mean monthly flow at Bourke gauge under scenarios P, A, C and D. The plots show that although there is consistently less water under all scenarios compared to Scenario P the seasonality is preserved across all months. There is a noticeable change between Scenario A and scenarios C and D in spring and early summer. Monthly flow (GL) C range Cmid A P J F M A M J J A S O N D Monthly flow (GL) D range Dmid A P J F M A M J J A S O N D 4 River system modelling Figure 4-9. Average monthly flow at Bourke gauge (425003) under scenarios P, A, C and D Table 4-12 shows the size of daily flow frequency with two-, five- and ten-year recurrence intervals under scenarios P, A, C and D. This analysis uses daily flow and not the peak flow during a day, which can be higher in river systems. Figures in the table show a 41 percent reduction in the size of two-year events from without-development to Scenario A, and a 30 percent reduction in the larger five- and ten-year return interval events. Such changes have consequences for floodplain harvesting and watering of floodplain ecosystems. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 53

64 Table Daily flow event frequency at Bourke gauge (425003) under scenarios P, A, C and D Return interval P A Cwet Cmid Cdry Dwet Dmid Ddry years ML/d percent change from Scenario A 2 42,328 24,795 33% -13% -35% 31% -15% -36% 5 70,804 49,883 50% -10% -35% 48% -11% -38% , ,711 65% -16% -43% 58% -19% -44% End-of-system flow characteristics 4 River system modelling Figure 4-10 shows the flow duration curves for the combined end-of-system flow at Menindee (the combined flow for inflow to Lake Wetherell and Talyawalka Creek at Railway Bridge). Low flow volumes are similar to without development but development has substantially reduced mid-range flows. This pattern is predicted to continue under future scenarios. Daily flow (ML) C range Cmid A P Daily flow (ML) D range Dmid A P Percent time flow is exceeded Percent time flow is exceeded Figure Daily flow duration curves for the combined end-of-system flow at Menindee under scenarios P, A, C and D Figure 4-11 gives the mean monthly flow under scenarios P, A, C and D for the Menindee end-of-system. They show that the seasonality is preserved under all of the scenarios, but with a reduction of the spring flows likely due to climate change. They also show the large change in end-of-system flows at Menindee compared to without development under all scenarios. Monthly flow (GL) C range Cmid A P Monthly flow (GL) D range Dmid A P J F M A M J J A S O N D J F M A M J J A S O N D Figure Seasonal flow curves for the combined end-of-system flows at Menindee under scenarios P, A, C and D 54 Water availability in the Barwon-Darling June 2008 CSIRO 2008

65 The percentage of time that flow occurs under these scenarios is presented in Table Cease-to-flow is when model flows are less than 1 ML/day. Table Percentage of time flow occurs at the end-of-system under scenarios P, A, C and D Outflow name P A Cwet Cmid Cdry Dwet Dmid Ddry Combined flow at Menindee 96% 99% 99% 98% 97% 99% 98% 97% Share of available resource Non-diverted water shares There are several ways of considering the relative level of impact on non-diverted water and diversions. Table 4-14 presents two indicators for relative impact on average annual non-diverted water: as a proportion of surface water availability under each scenario as a proportion of non-diverted water under Scenario A. Results presented are for the entire Darling Basin that is, they consider the surface water and the surface water use across the Barwon-Darling region and contributing upstream regions. 4 River system modelling Table Relative level of available water not diverted for use for the entire Darling Basin aggregated to Bourke under scenarios A, C and D Relative level of non-diverted water A Cwet Cmid Cdry Dwet Dmid Ddry percent Non-diverted water as a percentage of total available water 61% 68% 59% 55% 67% 58% 54% Non-diverted share relative to Scenario A non-diverted share 100% 145% 90% 65% 143% 88% 64% Combined water shares Figure 4-12 combines the results from water availability, level of development (total water diverted) and non-diverted water. The size of the bars indicates total water availability and the subdivision of the bars indicates the diverted components that include use in all contributing regions and non-diverted fractions. The graph reinforces that the non-diverted share varies more strongly than the diverted share. This analysis reflects use within the region and the contributing upstream regions. The results show that under scenarios Cdry and Ddry there are more substantial reductions in non-diverted water than changes in diverted volume as presented earlier. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 55

66 Average annual water (GL) Diverted Non-diverted P A Cwet Cmid Cdry Dwet Dmid Ddry 4 River system modelling Figure Comparison of diverted and non-diverted shares of water for the entire Darling Basin aggregated to Bourke under scenarios P, A, C and D 4.4 Discussion of key findings Model configuration The Barwon-Darling model used to establish licence access conditions by DWE was established for the period 1 January 1922 to 30 June Inflows time series for Scenario O are those currently available from DWE. These reflect recent irrigation development in the tributaries, re-calibration of tributary models, inclusion of the current Water Sharing Plans or Resource Operation Plans and the extended climatic period from 1 February 1891 to 30 June The model used in this project differs from the model used to develop the access conditions in that it reflects irrigation development on the Barwon-Darling River in 2004/05 (not in 2000/01). Scenarios Scenarios A0 to D results are based on the modelling platform specifically developed for this project which links river models across the whole of the Murray-Darling Basin. This platform is different to that used for the results for Scenario O, which uses an inflow file from individual valley models. Results cover the climatic period 1 July 1895 to 30 June The difference in inflows between Scenario O and A0 for the gauged subcatchments is 4 percent on average. This difference is due to changes in model platform, links to current tributary model results and differences in climatic period. The difference between scenarios A and A0 represents the impact of current levels of groundwater extraction at dynamic equilibrium in the upstream tributaries. For the Barwon-Darling region there is a decrease in tributary inflows of 8.1 GL/year, predominantly due to the Namoi region. Projected future development in the tributary catchments would reduce inflows to the Barwon-Darling region by 78 GL/year. This is composed of farm dam development which is estimated to cause a 36 GL/year decrease and future groundwater development which would cause a 42 GL/year decrease in inflows. The combined impact would be a 3 percent reduction in inflows. This translates to 64 GL/year less available water (36 GL/year due to groundwater development and 28 GL/year due to future farm dams) and a 3 percent reduction in end-of-system flows from the Barwon-Darling region because diversions within the region are relatively small and unchanged. Consumptive use The maximum average annual mainstream flow occurs at Bourke (gauge ) with a value of 3485 GL/year under Scenario P (without-development Scenario A). 56 Water availability in the Barwon-Darling June 2008 CSIRO 2008

67 The total surface water availability was estimated based on the maximum stream flow adjusted to account for groundwater usage and leakage implicitly incorporated in the upstream tributary models (equivalent flow of 29.9 GL/year at Bourke). The resultant water availability under Scenario A is 3515 GL/year. Under Scenario Cmid there would be an 8 percent decrease in water availability (Table 4-6). The Barwon-Darling system is not regulated. The largest diversion is for Class B irrigation (70 percent of total diversions), followed by Class C (17 percent of total diversions). There would be a 2 percent increase in total diversions under Scenario Cmid. There would be a further reduction in diversions under scenarios Dmid, Cdry and Ddry. Under Scenario Cwet more water is diverted than under the current climate. Diversions in the region are considerably lower than average during the driest one- and three-year periods, but there is negligible change under Scenario Cmid for these periods. This is likely due to irrigators being able to maintain frequency of access and on-farm irrigation infrastructure buffering the impact during dry periods. The combined use in the region and use within upstream tributaries result in an overall current level of use (ratio of water use to water availability) of 39 percent (Table 4-9). Under Scenario Cmid there would be a 2 percent increase in the level of use and there would be a 10 percent reduction in end-of-system flows. Level of use increases due to an increase in extractions and a decrease in average annual flow. The relative change in reliability under Scenario Dmid would be less than 2 percent for the main irrigation licences (Class B and Class C). This may be due to licence conditions still allowing sufficient access to enable historical areas to be grown. Diversions for floodplain harvesting would decrease under scenarios Cmid and Dmid as access for this component is dependent on large overbank flow events, which decrease in frequency for future scenarios. 4 River system modelling Flow behaviour The impact of current development on average end-of-system flows for the Barwon-Darling region at Menindee is greater than the expected impact of future climate and future development. Seasonality (as presented as mean monthly flows) is preserved at the end-of-system gauge under all scenarios, but with a reduction of spring flows under future drier scenarios. Current development has reduced the overall volume of flow, mainly through reduction of mid-range flows. 4.5 References DNR (2006a) Barwon-Darling River Valley, IQQM Implementation Calibration Summary Report. Issue 3 (Final Draft) New South Wales Department of Natural Resources, Parramatta. DNR (2006b) A Strategy to Manage Extractions from the Barwon-Darling. Pamphlet series Land and Water for Life, New South Wales Department of Natural Resources, Parramatta. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 57

68 5 Uncertainty in surface water modelling results This chapter describes the assessment of uncertainty in the surface water modelling results. It has four sections: a summary an overview of the approach a presentation and description of results a discussion of key findings. 5.1 Summary 5 Uncertainty in surface water modelling results The uncertainty that is internal to the river model (as opposed to that associated with the scenarios), and the implications that this has for confidence in the results and their appropriate use, are assessed using multiple lines of evidence. This involves comparing: (i) the river model to historical gauged main stem flows and diversions, which are its main points of reference to actual conditions; and (ii) ungauged inferred inflows and losses in the model to independent data on inflows and losses to ascertain if they can be attributed to known processes. These two aspects of model performance were then combined with some other measures to assess how well the model might predict future patterns of flow Issues and observations The Barwon-Darling surface water system is moderately well gauged. The density of gauging is less than the Murray-Darling Basin (MDB) average and concentrated in the upper reaches. Most inflows are from upstream regions and these are well gauged. Water accounts were established for seven consecutive reaches to cover the entire region. The majority of inflows and outflows are either gauged or attributable to other measurements, but considerable unattributed apparent gains and losses (of up to 40 percent) introduce large uncertainty into the modelling. Issues include the accuracy of gauging, the occurrence of gauge bypass flows, and river and wetland losses that cannot be directly gauged Key messages The assessment of uncertainty in the surface water modelling results indicates: The river model generally reproduces observed streamflow patterns well and produces estimates that agree reasonably well with water accounts. Modelled flows are greater than gauged at the most upstream and downstream gauges of the region and for the Barwon River between Danger Bridge and Brewarrina. The projected changes in flows due to future climate are greater than model uncertainty under the wet future climate scenarios but similar to model uncertainty under the dry and best estimate future climate scenarios, partly due to the model bias in some reaches. The model provides strong evidence of changes in flow pattern due to water resource development to-date, but the modelled flow changes due to projected future development are very small. While the model is well suited for the purpose of this project, it is noted that changes in low flows are not simulated well by the model. 58 Water availability in the Barwon-Darling June 2008 CSIRO 2008

69 5.2 Approach General A river model is used in Chapter 4 to analyse expected changes in water balance, flow patterns and consequent water security under climate and/or development change scenarios. Uncertainty in the analysis can be external or internal: External uncertainty is external to the model. It includes uncertainty associated with the forcing data used in the model, determined by processes outside the model such as climate processes, land use and water resources development. Internal uncertainty relates to predictive uncertainty in the river model, which is an imperfect representation of reality. It can include uncertainty associated with the conceptual model, the algorithms and software code it is expressed in, and its specific application to a region (Refsgaard and Henriksen, 2004). Full measurement of uncertainty is impossible. The analysis focuses on internal uncertainty. When scenarios take the model beyond circumstances that have been observed in the past, measurable uncertainty may only be a small part of total uncertainty (Weiss, 2003; Bredehoeft, 2005). The approach to addressing internal uncertainty involved combining quantitative analysis with qualitative interpretation of the model adequacy (similar to model pedigree, cf. Funtowicz and Ravetz, 1990; Van der Sluijs et al., 2005) using multiple lines of evidence. The lines of evidence are: the quality of the hydrological observation network the components of total estimated streamflow gains and losses that are directly gauged, or can easily be attributed using additional observations and knowledge, respectively (through water accounting) characteristics of model conceptualisation, assumptions and calibration the confidence with which the water balance can be estimated (through comparison of water balances from the baseline river model simulations and from water accounting) measures of the baseline model s performance in simulating observed streamflow patterns the projected changes in flow pattern under the scenarios compared to the performance of the model in reproducing historical flow patterns. None of these lines of evidence are conclusive in their own right. In particular: The model may be right for the wrong reasons for example, by having compensating errors. There is no absolute reference truth. All observations inherently have errors and the water accounts developed here use models and inference to attribute water balance components that were not directly measured. Adequate reproduction of historically observed patterns does not guarantee that reliable predictions about the future are produced. This is particularly so if model boundary conditions are outside historically observed conditions, such as in similar climate change studies. 5 Uncertainty in surface water modelling results Qualitative model assessment is preferably done by consulting experts (Refsgaard et al., 2006). The timing of the project prevented this. Instead a tentative assessment of model performance is reviewed by research area experts within and outside the project as well as by stakeholder representatives. The likelihood that the river model gives realistic estimates of the changes that would occur under the scenarios evaluated is assessed within the above limitations. Overall river model uncertainty is the sum of internal and external uncertainty. The range of results under different scenarios in this project provides an indication of the external uncertainty. River model improvements will reduce overall uncertainty only where internal uncertainty clearly exceeds the external uncertainty. The implication of overall uncertainty on the use of the results presented in this project depends on: (i) the magnitude of the assessed change and the level of threat that this implies, and (ii) the acceptable level of risk (Pappenberger and Beven, 2006). This is largely a subjective assessment and is not attempted herein. A possible framework for considering the implications of the assessed uncertainties is shown in Table 5-1. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 59

70 Table 5-1. Framework for considering implications of assessed uncertainties Low uncertainty Low threat Current water sharing arrangements appear sufficient for ongoing management of water resources. High threat Current water sharing arrangements are likely to be inadequate for ongoing management of water resources. A major revision of the water sharing arrangements is recommended. High uncertainty Current water sharing arrangements appear sufficient for ongoing management of water resources, but careful monitoring and adaptive management is recommended. Current water sharing arrangements may be inadequate for ongoing management of water resources. Further work to reduce the major sources of uncertainty is recommended as a basis for adjusting current water sharing arrangements. 5 Uncertainty in surface water modelling results Information sources Information on the gauging network was obtained from the Water Resources Station Catalogue ( and the Pinneena 8 database (provided on CDROM by New South Wales Department of Water and Energy (DWE)). A report that included the results of IQQM model calibration was provided (DLWC, 1998; DLWC, 2006). Time series of water balance components as modelled under Scenario A and all other scenarios were derived as described in Chapter 4. The data used in water accounting are described in the following section Water balance accounting Purpose Generic aspects of the water accounting methods are described in Chapter 1. This section includes a description of the basic purpose of the accounts: to inform the uncertainty analysis using an independent set of the different water balance components by reach and by month. The descriptions in Chapter 1 also cover the aspects of the remote sensing analyses used to estimate wetland and irrigation water use and inform calculations for attribution of apparent ungauged gains and losses. Aspects of the methods that are region specific are presented below. Framework Water accounts were established for seven successive reaches covering the entire system. The available streamflow data were adequate in most reaches for water accounting from 1990/91 to 2005/06. For the gauge at Louth between reaches 6 and 7 streamflow data were only available from 1992 onwards. The associated subcatchments are shown in Figure 5-1 and are related to water accounting reaches in Table 5-2. Table 5-2. Comparison of water accounting reaches with subcatchment codes used in the river model Water accounting reach Subcatchment code(s) Description 1, Mogil Mogil, Collarenabri , Danger Bridge (Walgett) , Brewarrina Bourke Town 6, Louth, Wilcannia 60 Water availability in the Barwon-Darling June 2008 CSIRO 2008

71 Figure 5-1. Map showing the subcatchments used in modelling, accounting reaches and contributing catchments. Black dots and red lines are nodes and links in the river model respectively Diversion data Wetland and irrigation water use 5 Uncertainty in surface water modelling results The result of the remote sensing classification (Chapter 1) is shown in Figure 5-1. Irrigation areas occur in most reaches but are concentrated in Reach 6 (near Bourke). Irrigation areas also occur in the catchment of Reach 3 but these are not supplied from the Barwon River. Diversion data were provided by DWE as annual totals. These were disaggregated to monthly totals using methods described in Chapter 1. Extensive floodplains and wetlands were identified along the entire length of the Barwon and Darling rivers. Calculation and attribution of apparent ungauged gains and losses Calculation and attribution of apparent ungauged gains and losses were undertaken according to the methods described in Chapter Model uncertainty analysis The river model results and water accounts were used to derive measures of model uncertainty. The different analyses are described below. Details on the equations used to calculate the indicators are not provided here but can be found in Van Dijk (2007). Calculations were made for each reach separately but summary indicators were compared between reaches. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 61

72 Completeness of hydrological observation network Statistics on how well all the estimated river gains and losses were gauged or, where not gauged, could be attributed based on additional observations and modelling were calculated for each reach: the volumes of water measured at gauging stations and off-takes, as a fraction of the grand totals of all estimated inflows or gains, and/or all outflows or losses, respectively the fraction of month-to-month variation in the above terms the same calculations as above, but for the sum of gauged terms plus water balance terms that could be attributed using the water accounting methods. The results of this analysis for annual totals are also presented in Appendix C. Comparison of modelled and accounted reach water balance 5 Uncertainty in surface water modelling results The water balance terms for river reaches were compared for the period of water accounting period as modelled by the river model for Scenario A and as accounted. Large divergence is likely to indicate large uncertainty in reach water fluxes and therefore uncertainty in the river model and water accounts. Climate range If the model calibration period is characterised by climate conditions that are a small subset, or atypical of the range of climate conditions that was historically observed, this increases the chance that the model will behave in unexpected ways for climate conditions outside the calibration range. The percentage of the overall climate variability range for the 111-year climate sequence used in the river modelling that was covered by the extremes in the calibration period was calculated as an indicator. Performance of the river model in explaining historical flow patterns All the indicators used in this analysis are based on the Nash-Sutcliffe model efficiency (NSME; Nash and Sutcliffe, 1970). NSME indicates the fraction of observed variability in flow patterns that is accurately reproduced by the model. In addition to NSME values for monthly and annual outflows, values were calculated for log-transformed and ranked flows, and high (highest 10 percent) and low (lowest 10 percent) monthly flows. NSME cannot be calculated for the logtransformed flows where observed monthly flows include zero values or for low flows if more than 10 percent of months have zero flow. NSME is used to calculate the efficiency of the water accounts in explaining observed outflows. This indicates the scope for model improvements to explain more of the observed variability. If NSME is much higher for the water accounts than for the model, it suggests that the model can be improved to reduce uncertainty. If similar, additional hydrological data may be required to support a better model. A visual comparison of streamflow patterns at the end-of-reach gauge with the flows predicted by the baseline river model and the outflows that could be accounted was done for monthly and annual time series and for monthly flow duration curves. Scenario change-uncertainty ratio Streamflow patterns simulated for any of the scenarios can be used as an alternative river model. If these scenario flows explain historically observed flows about as well or better than the baseline model, then it may be concluded that the modelled scenario changes are within model noise, that is, smaller or similar to model uncertainty. Conversely, if the agreement between scenario flows and historically observed flows is poor much poorer than between the baseline model and observations then the model uncertainty is smaller than the modelled change, and the modelled change can be meaningfully interpreted. The metric used to test this hypothesis is the change-uncertainty ratio (CUR). The definition was modified from Bormann (2005) and calculated as the ratio of the NSME value for the scenario model to that for the baseline (Scenario A) model. A value of around 1.0 or less suggests that the projected scenario change is not significant when compared to river model uncertainty. A ratio that is considerably greater than 1.0 indicates that the future scenario model is much poorer at producing historical observations than the baseline model, suggesting that the scenario leads to significant changes in 62 Water availability in the Barwon-Darling June 2008 CSIRO 2008

73 flow. The CUR is calculated for monthly and annual values, in case the baseline model reproduces annual patterns well but not monthly patterns. The same information was plotted as annual time series, monthly flow duration curves and a graphical comparison made of monthly and annual change-uncertainty ratios for each scenario. 5.3 Results Density of the gauging network Figure 5-2 shows the location of streamflow, rainfall and evaporation gauges in the region. Table 5-3 provides information on the measurement network. This semi-arid region is the fourth most sparsely gauged region in the MDB but most of the water comes from gauged inflows from other regions. The density of the rainfall stations is less than half the MDB average. The density of streamflow and evaporation gauging is about one-fifth of the MDB average. Streamflow gauging is concentrated upstream of Bourke. The Darling River has seven gauges. Rainfall gauges are evenly distributed. Table 5-3. Some characteristics of the gauging network of the Barwon-Darling region (142,173 km 2 ) compared with the entire Murray-Darling Basin (1,062,443 km 2 ) Gauging network characteristics Barwon-Darling MDB Number per 1000 km 2 Number per 1000 km 2 Rainfall Total stations Stations active since Average years of record Streamflow Total stations Stations active since Average years of record Evaporation Total stations Stations active since Average years of record Uncertainty in surface water modelling results CSIRO 2008 June 2008 Water availability in the Barwon-Darling 63

74 5 Uncertainty in surface water modelling results Figure 5-2. Map showing the rainfall, streamflow and evaporation observation network, along with the subcatchments used in modelling Review of model calibration and evaluation information This section summarises the previous evaluation of the Barwon-Darling IQQM. It is noted that for the current project, the model was linked to river models for upstream regions, and thus its behaviour may be different from the model described below. Model description The Barwon-Darling IQQM simulates the river system between Mungindi on the Barwon River and two gauges on the Darling River and Talyawalka Creek near Wilcannia. It was initially developed around 1993 to examine flows and irrigation diversions and became a tool for development and auditing of water policy options. It has subsequently received a series of upgrades for assessing the impacts of the proposed environmental flows and auditing of the Murray- Darling Basin Commission cap on surface water diversions. The updates included: collection of improved irrigation data; representation of individual irrigators and of on-farm water management; revision of tributary flows; inclusion of residual catchment inflows and streamflow losses; generation of daily synthetic evaporation for 100 years; re-calibration of irrigation to reflect improved crop areas, crop mix and diversion data; description of individual irrigator s planting behaviour; improved allocation of irrigated area between summer and winter crops; and inclusion of separate soil moisture modelling for each different crop type and fallow fields for each individual irrigator. Details on model concepts, assumptions, calibration and performance assessment summarised here are reported in DLWC (2006). No model report was provided for the Menindee connection model so its calibration and performance could not be evaluated. There are 15 weirs on the Barwon-Darling with total capacity in excess of 100 GL. The impact of these on routing and evaporation is simulated but any impacts on streamflow and irrigators are not modelled. Town water, stock and domestic, riparian, and industrial water uses are not explicitly modelled because of their relatively small volumes. No allowance is made in the model for groundwater usage or groundwater interaction. 64 Water availability in the Barwon-Darling June 2008 CSIRO 2008

75 Data availability Streamflow, diversion, rainfall and evaporation data are inputs to the model. Irrigation data such as licence conditions, pump details, water diversions, crops irrigated, on-farm storage details and on-farm water management practices were also used in model development. Rainfall data are used to compute contributions to on-farm storage volumes and river reaches due to direct rainfall onto water surfaces. It is also used for soil moisture accounting to calculate crop demands and rainfall-runoff harvesting. Daily data from seven rainfall stations were selected as they provided a long continuous record and stations nearby that could be used to substitute missing data and disaggregate records. Evaporation data supports estimation of rural demands and evaporation from storages and river reaches. Three of the Class A Pan evaporation stations in the region (Walgett, Bourke and Menindee) had long continuous records and were selected to represent the spatial evaporation distribution in the Collarenebri Wilcannia reaches. Evaporation data from three stations outside the region were used to represent the evaporation in the Mungindi Collarenebri reach. A factor of 0.78 was used to convert Class A Pan data to evaporation from open water. A factor of 0.7 was used for conversion to reference crop evapotranspiration. Streamflow gauges on the main river and tributaries were selected for calibration purposes and to define tributary inflows, respectively. The criteria used to select gauges for calibration of main streamflows were: a sufficient number of sites to limit the length of river reaches, sites upstream and downstream of tributary inflows or effluent outflows, sites with long, good quality records and with a minimum number of missing periods, and sites that are used to define access for irrigation. Nine main stream gauges were selected for model calibration (Table 5-4). Another 17 tributary gauges were selected. Missing data from these gauges were filled by correlation with data from nearby gauges. Irrigation infrastructure data were collected for development of the model. These data include pump and on-farm storage capacities (estimated at more than 250 GL), and crop areas and types (from surveys and remote sensing). Survey data relied on irrigators own estimates without any attempt at independent validation. There is no monitoring of water use by reach irrigators, except for irrigator estimates of crops grown from 1986/87 to 1994/95 and for 1999/00. While the number of these irrigators is large their respective irrigated areas were small (~1700 ha) and hence they were aggregated by reach. 5 Uncertainty in surface water modelling results CSIRO 2008 June 2008 Water availability in the Barwon-Darling 65

76 Table 5-4. Streamflow gauging stations for which data were used in Barwon-Darling IQQM calibration 5 Uncertainty in surface water modelling results Station Location Calibration period Main river Barwon u/s Pressbury Weir 1944 current Barwon Mogil Mogil 1944 current Barwon Collarenebri 1965 current Quality assessment of calibration Barwon Walgett 1892 current Very high for mid flow range (within 1% of daily flow), moderate for low flows due to uncertainties in water extractions Barwon Brewarrina 1947 current Darling Bourke 1954 current Model was able to replicate daily flows very accurately (errors of 3 to 7 percent) in the whole range, except for the low flows where the rating is very low Darling Louth 1995 current Darling Tilpa 1913 current Darling Wilcannia 1971 current Overall calibration was of high to very high quality, except for the low flows where it is low Tributaries Barwon Mungindi 1889 current Boomi Neewoora Gil Gil Weemelah No current Moonie Gundabluie 1945 current Gwydir Collymongle Mehi Collarenebri 1980 current Namoi Goangra 1954 current Pian Waminda 1972 current Castlereagh Coonamble 1960 current Marthaguy Carinda 1944 current Macquarie Carinda 1926 current Marra Carinda Road 1980 current Bogan Gongolgon 1942 current Bokhara Bokhara 1944 current Culgoa d/s Collerina 1944 current Warrego Ford s Bridge 1921 current Model calibration and validation procedures The flow calibration period was 1970 to 1984 as there were good flow data and minimal irrigation development. Calibration of diversion was from 1995 to 2000 as comprehensive water use data were available. Flow calibration was undertaken for three river reaches: Mungindi Walgett, Walgett Bourke, and Bourke Wilcannia. Streamflow records at these gauging stations allowed a full water balance analysis to be done. Diversion calibration used reported areas of irrigated crops, on-farm storage volumes and metered diversions. Crop water demands, rainfall harvesting, floodplain harvesting and on-farm storage losses were also considered. A calibration process was developed to proceed sequentially down the river system and progressively eliminate unknowns. Specific parameters were estimated at each step and all other parameters replaced with observed data. The steps are summarised below: Flow calibration reproduced the observed flow hydrographs at key locations given observed tributary inflows and water extractions. Routing parameters, transmission losses and ungauged inflows were calibrated during this step. The calibration period was 1970 to Diversion calibration reproduced observed irrigation extractions given crop areas, crop mix, pump capacities and on-farm storage development. Irrigation efficiency, rainfall losses, soil moisture stores, on-farm storage operation (including their reserves and rainfall and floodplain harvesting) were calibrated during this step. Calibration was achieved when simulated diversions were within 20 percent of recorded diversion in any year 66 Water availability in the Barwon-Darling June 2008 CSIRO 2008

77 and within 5 percent for the whole calibration period. The calibration period was 1995 to Irrigation diversion data was not reliable prior to the installation of time and event meters in The area planting decision step involved calibrating an irrigator s decision-making process in reproducing observed planted crop areas. Maximum and minimum area, crop mix and farmers planting decision process were calibrated. A risk function was used to simulate cropped areas because of the lack of reliable data for onfarm storage volumes on the required decision dates. The risk function is expressed as the ratio of a volume of on-farm storage water (at planting date September/October) to the area the farmer decides to plant. This step was performed for the period 1995 to The overall model calibration period (referred to as model validation period in DLWC, 2006) was the same as for planted area calibration (1995 to 2000). Model performance A quality assessment guideline was adopted with five confidence levels to assess overall model performance: very high (simulated value within 5 percent of observed value), high (5 to 10 percent), moderate (10 to 15 percent), low (15 to 20 percent) and very low (greater than 20 percent). The above limits were varied in some cases depending on the indicator and uncertainty in the measured data (DLWC, 2006). The quality indicators were adjusted on the representativeness of the climate data based on the length of the calibration period. Three main gauging stations (Walgett, Bourke and Louth) were selected to create three primary flow calibration reaches. A few sub-reaches were identified for low flow calibration. Flow calibration over the entire range was performed at each station. Tributary flows predominated during large flood events and affected the initial calibration. This was rectified in later calibrations and the results for the primary stations are summarised in Table 5-4. There was a 5 percent difference in the total volume of simulated flow and observed flow at Wilcannia, Bourke and Walgett. Differences were greater for individual flow events given data accuracy and the significant amounts of ungauged water entering the river. The quality of diversion and planted area calibrations was high to very high for most reaches (with differences in simulated and observed diversions ranging from 2 to 4 percent). The exception was in the reach between Bourke and Wilcannia where the calibration was moderate to high. The results of the model validation showed low quality flow calibration during the validation period of 1995 to The overall quality of the model calibration was very high, making the model suitable for Cap modelling and auditing and long-term analysis of system behaviour (DLWC, 2006). Sensitivity studies were undertaken to assess the likely change in the model if flows were increased or decreased by 10 percent. Streamflow is the key parameter affecting access to water and diversions in the Barwon-Darling region. Overall the percentage of time irrigators had access to flows and were able to divert water was not sensitive to the changes in flows. 5 Uncertainty in surface water modelling results A simulation was performed over the period where irrigation development was closest to Cap conditions to assess the robustness of the Cap scenario. The four irrigation seasons from 1992/93 to 1995/96 were most appropriate. The major differences between modelled and observed flow volumes were in higher flow periods. These differences are due to the factoring of tributary inflows and can only be improved with better information on high flows. Identified areas of weakness The flow calibration has not been updated since 2000 and does not include the recent drought period that could provide some important information on losses at low flows and during dry periods. A number of improvements are planned including (DLWC, 2006): a model flow calibration review with more recent and better quality streamflow data (better estimation of ungauged inflows from tributary streams will only happen if more accurate and complete streamflow data can be obtained to allow a water balance approach) a revision of model diversion calibration when new metering data is collected to overcome problems with older style time and event meters. Better measurement and monitoring of the diversion data will also provide an independent measure of on-farm storage capacities and seepage losses incorporation of the 15 on-river weirs into the Barwon-Darling IQQM. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 67

78 5.3.3 Model uncertainty analysis The calculated indicators of model uncertainty and all other water accounting results are listed by reach in Appendix C. This section provides a summary. Completeness of hydrological observation network The estimated fraction of all gains and losses that is gauged is shown for each reach in Figure 5-3. The following conclusions are drawn: 5 Uncertainty in surface water modelling results Fraction gauged Gains in the reaches with water accounts are reasonably to very well gauged (64 to 95 percent). Gauging is worst in Reach 1 where there are ungauged and mostly unattributed apparent gains. Outflows and losses are reasonably to very well gauged (65 to 93 percent).the lowest reach (Reach 7) has the least comprehensive gauging of losses associated with large floodplain and wetland losses. Overall, 77 to 85 percent of the total water balance in each reach is gauged. Attribution of gains and losses using SIMHYD estimates of local runoff, diversion data and remote sensing help to explain some of the ungauged gains and losses (to a total of 84 to 95 percent of the combined reach gains and losses). The region itself is dry and generates little runoff. The majority of streamflow derives from upstream regions and these inflows are gauged. Overall, most gains and losses are gauged or can be attributed, and therefore the water balance of the region is well understood. (a) Accounting reach inflows/gains outflows/losses total flow components Fraction attributed (b) Accounting reach Figure 5-3. The fraction of inflows/gains, outflows/losses and the total of water balance components that is (a) gauged or (b) could be attributed in the water accounts Comparison of modelled and accounted reach water balance A summary of the water balance for all reaches as simulated by the river model and derived by water accounting can be found in Appendix C. The water balances are combined in Table 5-5. The model included effluent outflows that reappeared as effluent return flows in the reach below (bypass flows) in reaches 1 and 2. These are not included in the water balance in Table 5-5. Some mass balance error occurs due to the different periods of accounting for different reaches. Overall the system shows maximum flows around Bourke (Reach 5), though individual reaches are alternately gaining and losing. 68 Water availability in the Barwon-Darling June 2008 CSIRO 2008

79 Table 5-5. Regional water balance modelled and estimated on the basis of water accounting. Water balance (Jul 1990 Jun 2006) Model (A) Accounts Difference Difference GL/y percent Inflows (gains) Main stem inflows % Tributary inflows % Local inflows % Total gains % Unattributed gains and noise % Outflows (losses) End-of-system outflows % Distributary outflows na Net diversions % River flux to groundwater na River and floodplain losses % Unspecified losses na Total losses % Unattributed losses and noise % na not applicable To aid interpretation: The model was linked to river models for upstream regions as part of this project. Therefore it is subject to the uncertainties in modelled outflows as they are reported in the respective reports for those regions. Reaches 1, 3 and 5 are gaining and the other reaches are losing. The river reaches in the region are alternately losing and gaining because of major tributary inflows in the gaining reaches and more extensive floodplain losses in the losing reaches. No attempt was made to estimate groundwater exchanges in the water accounting due to the lack of direct data. Nor were these exchanges simulated by the river model. Simulated combined main stem inflows into the Barwon River at Mungindi are 645 GL/year (63 percent) higher than accounted inflows (396 GL/year). The sum of gauged tributary inflows is estimated at 2114 GL/year by the model, which is 427 GL/year (25 percent) higher than accounted for (1688 GL/year). However, the definition between tributary and local inflows may vary between the model and the water accounting. The sum of both terms that was modelled (2144 GL/year) is 199 GL/year (10 percent) larger than the accounted sum (1945 GL/year). Simulated outflows from the Darling River below Wilcannia (1558 GL/year) are 315 GL/year (25 percent) higher than accounted (1242 GL/year). Simulated diversions for the water accounting period (215 GL/year) are 70 GL/year (48 percent) greater than those recorded (146 GL/year). Simulated combined distributary outflows, river and floodplain losses and unspecified losses (1018 GL/year) are 254 GL/year (21 percent) less than accounted river and floodplain losses (1282 GL/year), which include distributary losses. Gauged water balance terms including diversions represent 46 percent of the total water account. Another 21 percent can be attributed using SIMHYD local runoff estimates (258 GL/year) and estimates of river and floodplain losses (1282 GL/year). Unattributed gains are slightly larger than unattributed losses: for the entire accounted system combined unattributed gains (including measurement noise) represent 1480 GL/year (39 percent) of total apparent gains, whereas unattributed losses (including measurement noise) represent 1000 GL/year or 27 percent of total apparent losses. Their sum represents 40 percent of the total water balance. Overall, the system is reasonably well gauged and understood. However large ungauged and unattributed terms occur. Barwon River inflows at Mungindi, total net diversions, river and wetland losses, and end-ofsystem flows at Wilcannia all differ considerably between model and accounts (21 to 63 percent). 5 Uncertainty in surface water modelling results CSIRO 2008 June 2008 Water availability in the Barwon-Darling 69

80 Climate range The flow calibration period (1970 to 1984) best represents the overall calibration period because the region is dominated by unregulated flows. Seven years in the entire 111-year record used in modelling are drier than those included in this calibration period. Two years are wetter. The average rainfall for the calibration period (395 mm/year) is 20 percent higher than the long-term average (329 mm/year). The historical 111-year rainfall record has one year that is drier and five years that are wetter than the extremes during the period of water accounting (1990 to 2006). Overall, the period of calibration provides good representation of the longer climate record. The water accounting period also provides a good representation of long-term climate variability. Performance of the river model in explaining historical flow patterns 5 Uncertainty in surface water modelling results The better the baseline model simulates streamflow patterns, the greater the likelihood that it represents the response of river flows to changed climate, land use and regulation changes (notwithstanding the possibility that the model is right for the wrong reasons through compensating errors). Appendix C lists indicators reach by reach of the model s performance in reproducing different aspects of the patterns in historically measured monthly and annual flows (all are variants of NSME). Appendix C gives NSME values of model performance. These are also shown in Figure 5-4. Model efficiency 1.0 monthly - normal 0.8 monthly - log-norm. monthly - ranked 0.6 monthly - high flows 0.4 annual - normal 0.2 annual - log-norm. 0.0 annual - ranked Accounting reach Figure 5-4. Changes in the model efficiency (the performance of the river model in explaining observed streamflow patterns) along the length of the river (numbers refer to reach) The following observations are made: Model performance for both annual and monthly flow totals varies between reasonable and excellent (NSME= ). Performance is worst in reaches 1, 4 and 7. Performance in reproducing the 10 percent highest flows varies widely. It is good for reaches 5 and 6 but poor for reaches 1, 4 and 7, as flows are underestimated in Reach 1 and overestimated in reaches 4 and 7. Performance in reproducing the 10 percent lowest flows is poor in all cases. The model appears to consistently overestimate very low flows in all reaches (Appendix C). The simulated and observed flow-duration curves agree reasonably well except for the mentioned overestimation of low flows (Appendix C). Scenario change-uncertainty ratio A high CUR corresponds with a change in flows related to a scenario that is likely to be significant given the uncertainty, or noise, in the model. A value of around 1.0 means that the modelled change is of similar magnitude as the uncertainty in the model. CURs under without-development (P), Cmid and Dmid scenarios are shown in Figure Water availability in the Barwon-Darling June 2008 CSIRO 2008

81 (a) (b) Change-uncertainty ratio Accounting reach Change-uncertainty ratio Accounting reach P Cwet Cmid Cdry P Cwet Cmid Cdry Dwet Dmid Ddry Dwet Dmid Ddry Figure 5-5. Pattern along the river of the ratio of the projected change over the river model uncertainty under scenarios P, C and D modelled for (a) monthly and (b) annual flows The following observations are made: The CURs are generally smaller for monthly totals than for annual totals due to the greater variability in monthly flows that is harder to simulate than annual patterns. The significance of the simulated change from without-development to current flow pattern increases from being near to model uncertainty in the top reach (Reach 1) (CUR of 0.7 to 0.9) to fairly strong in the lowest reach (Reach 7) (CUR of 3.6 to 5.8). Scenario Cmid only produces detectable change in reaches 5 and 6 (CUR of 2.4 and 3.2). Other reaches experience either small changes (reaches 2 and 3) or changes are masked by model bias (reaches 1, 4 and 7). The CUR values under Scenario D are almost identical to those under Scenario C, reflecting the very small amount of additional development which is indistinguishable from model noise. From this it is concluded that: The projected changes in flow pattern due to water resource development to-date are greater than estimated model uncertainty. Scenario Cwet leads to changes in flow that are significant when compared with internal model uncertainty for most of the region. Changes under scenarios Cmid and Cdry are closer to model uncertainty. The projected impact from development under Scenario D is very small when compared to the projected impact from climate change. The significance of projected changes increases weakly from upstream to downstream across the system. 5 Uncertainty in surface water modelling results 5.4 Discussion of key findings Gauging and understanding of the hydrology of the Barwon-Darling region The hydrology of the surface water system is moderately well gauged. The density of gauging is less than the MDB average and concentrated in the upper reaches but most water comes from gauged upstream regions. Water accounts were established for seven consecutive reaches encompassing the entire region. Overall the system shows maximum flows around Bourke, though individual reaches are alternately gaining and losing. The alternating pattern is due to large tributary contributions at different points in the system and some extensive river and wetland losses along some reaches. Overall, the region is gauged and understood well enough for reliable modelling, but the considerable unattributed apparent gains and losses introduce large uncertainty into the modelling. The main issues are related to the accuracy of gauging, the occurrence of gauge bypass flows, and the importance of river and wetland losses that cannot be gauged. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 71

82 Groundwater interactions (Chapter 6) are a small term in the surface water balance and were not simulated by the river model or included in the water accounts. Surface water diversions are a relatively small component of total inflows (less than 10 percent). Uncertainty associated with changes in river regulation, irrigation and development may therefore be limited. Model performance in explaining observations and comparison to water accounts Overall model performance is reasonable to very good, although the assessment indicates that flows are not always well reproduced. In particular there appears to be a bias when comparing simulated flows to gauging data (which may be associated with bypass flows) and very low flows are not well simulated. The calibrated climate range provides a good mix of wet and dry years which increases confidence in the reliability of the model under climate change scenarios. The accounted and simulated water balance terms agree reasonably well, but differences greater than 20 percent occur in several terms. 5 Uncertainty in surface water modelling results Implications for the use of these results The model reproduces observed streamflow patterns reasonably well and produces estimates that agree reasonably well with water balance accounts. The projected changes in flows due to future climate are greater than model uncertainty under Scenario Cwet but similar to model uncertainty under scenarios Cmid and Cdry. The model provides strong evidence of changes in flow pattern due to water resource development to-date, but the changes due to projected future development under Scenario D are very small. While the model is generally well suited for the purpose of the project, low flows are not simulated well by the model. 5.5 References Bormann H (2005) Evaluation of hydrological models for scenario analyses: Signal-to-noise-ratio between scenario effects and model uncertainty. Advances in Geosciences 5, Bredehoeft J (2005) The conceptual model problem surprise. Hydrogeology Journal 13, DLWC (2006) Barwon-Darling River System, IQQM Implementation Calibration Summary Report. Surface and Groundwater Processes Unit, Centre for Natural Resources, Department of Land and Water Conservation, Sydney. DLWC (1998) IQQM User Manual, Report No. TS , NSW Department of Land and Water Conservation. Funtowicz SO and Ravetz J (1990) Uncertainty and Quality in Science for Policy. Kluwer Academic Publishers, Dordrecht. Nash JE and Sutcliffe JV (1970) River flow forecasting through conceptual models, 1: a discussion of principles. Journal of Hydrology 10, Pappenberger F and Beven KJ (2006) Ignorance is bliss: Or seven reasons not to use uncertainty analysis. Water Resources Research 42, W05302, doi /2005WR Refsgaard JC and Henriksen HJ (2004) Modelling guidelines terminology and guiding principles. Advances in Water Resources 27, Refsgaard JC, van der Sluijs JP, Brown J and van der Keur P (2006) A Framework for dealing with uncertainty due to model structure error. Advances in Water Resources 29, Van der Sluijs JP, Craye M, Funtowicz S, Kloprogge P, Ravetz J and Risbey J (2005) Combining quantitative and qualitative measures of uncertainty in model based environmental assessment: the NUSAP System. Risk Analysis 25, Van Dijk AIJM (2007) Climate variability impacts on the already stretched Murray-Darling Basin water system assessment and policy implications. In: Proceedings of the World Water Week, Stockholm, Sweden. Weiss C (2003) Expressing scientific uncertainty. Law, Probability and Risk 2, Water availability in the Barwon-Darling June 2008 CSIRO 2008

83 6 Groundwater assessment This chapter describes the groundwater assessments for the Barwon-Darling region. It has seven sections: a summary an overview of the regional assessment approach an overview of the groundwater management units a summary of trends of groundwater levels an analysis of surface groundwater connectivity an analysis of the water balance a discussion of key findings. 6.1 Summary Issues and observations The region is overlapped by 14 groundwater management units (GMUs). Ten of these are assessed as part of the Barwon-Darling region using simple water balance analyses Key messages 6 Groundwater assessment Groundwater extraction in the Barwon-Darling region for 2004/05 is estimated at 10 GL per year, being 0.6 percent of the total groundwater use in the Murray Darling Basin (MDB) (excluding confined aquifers of the Great Artesian Basin (GAB)). Groundwater development within the region is low: current extraction is less than 10 percent of rainfall recharge and 4 percent of total annual within-region water use on average. Current groundwater extraction reduces streamflow by about 1.3 GL/year. The projected 2030 climate would have little effect on groundwater with extraction remaining at less than 10 percent of rainfall recharge. Groundwater extraction is projected to increase to 240 GL/year by 2030, moving groundwater use to just over 50 percent of the total average annual water use in the region. This would reduce streamflow by 37 GL/year by Uncertainty A simple water balance approach was used to assess the region s groundwater. This was appropriate given the low priority of the GMUs in the context of the project. However, the approach would not be adequate for addressing local groundwater management issues. The estimated impacts of groundwater extraction on streamflow are assigned a low level of confidence. The estimates are sensitive to the connectivity factor used and may be over estimated given the methodology. Future extraction can not be predicted and estimates of the rate of growth are highly uncertain. 6.2 Approach The assessment approach includes an overview of the GMUs, a survey of trends in groundwater levels, an analysis of surface groundwater connectivity and an analysis of the water balance to evaluate the impact of changing rainfall recharge and extraction under each of the scenarios. The key outputs are ratios of groundwater extraction to rainfall recharge (E/R) and surface groundwater impacts for the scenarios. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 73

84 Rainfall recharge is estimated for each of the scenarios by multiplying the rainfall recharge under the historical climate by an appropriate recharge scaling factor (RSF) that is calculated using a one-dimensional Soil-Vegetation Atmosphere water transfer model (WAVES; Zhang and Dawes, 1998). The E/R ratio is used as the main indicator of groundwater development for this region: 0 to 0.3 (low), 0.3 to 0.7 (medium), 0.7 to 1.0 (high), greater than 1.0 (very high development). High to very high development would require information (for example, on flood recharge) that ensures extraction is not adversely impacting the resource or ecosystems that are dependent on groundwater. The scope does not include estimation of flood recharge. The E/R values would become smaller for GMUs such as the Upper Darling Alluvium and GAB Alluvial GMU if a flood recharge component were added. Limiting the assessment to rainfall recharge is consistent with the New South Wales macro planning approach of setting long-term average extraction limits based on rainfall recharge. The E/R ratios presented hare are a guide to macro planning decisions recognising that the approach is more conservative without flood recharge estimates. 6 Groundwater assessment The objectives of the surface groundwater connectivity mapping are to provide a catchment context for surface groundwater interactions, constrain surface water balance and constrain groundwater balances. The main output is a map of groundwater fluxes (magnitude and direction) adjacent to main streams. The approach uses Darcy s Law and hence estimates hydraulic conductivity and groundwater gradients about the streams. The method is dependent on the availability of appropriate groundwater monitoring and on previous work estimating hydraulic conductivity. A percentage estimate was made of the volume of groundwater pumped derived from streamflow (captured discharge or induced leakage) for each GMU based on detailed knowledge of the aquifer stratigraphy and geomorphology within each GMU. 6.3 Groundwater management units Location The main aquifers within the region are divided into GMUs for management purposes (Figure 6-1). These units are threedimensional. The region is overlapped by 14 GMUs and none are fully contained within the region. Lower Namoi Alluvium (N01), Lower Macquarie Alluvium (N08) and Lower Lachlan Alluvium (N12) are described within the Namoi, Macquarie-Castlereagh and Lachlan regional reports respectively. Similarly the assessment of the St George Alluvium GMU (Q71) in Queensland is reported within the Condamine-Balonne regional report. The region is underlain by the confined aquifers of the GAB. These water resources are governed by the water sharing plan for the GAB and are not considered further. This chapter provides an assessment of the remaining ten GMUs that overlap the region (Table 6-1). The hydrogeological context of the GMUs is described in Chapter Water availability in the Barwon-Darling June 2008 CSIRO 2008

85 Figure 6-1. Map of groundwater management units and key observation bores in the region 6 Groundwater assessment Table 6-1. Categorisation of groundwater management units in the region, including annual extraction, entitlement and recharge details Code Name Priority ranking Assessment ranking Current extraction (1) (2004/05) Total entitlement Long-term average extraction limit Recharge (2) N45 Lower Darling Alluvium very low simple <0.1 < N46 Upper Darling Alluvium very low simple N63 GAB Alluvial low simple N601 GAB Intake Beds very low simple N604 Gunnedah Basin very low simple N612 Western Murray Porous Rock very low simple N620 GAB Cap Rocks low simple (3) GL/y (4) np (3) N811 Lachlan Fold Belt low simple N813 Warrambungle Tertiary Basalt very low simple <0.1 < N817 Kanmantoo Fold Belt very low simple (1) Current groundwater extraction for Macro Groundwater Sharing Plan areas is based on metered and estimated data provided by New South Wales Department of Water and Energy (DWE). Data quality is variable depending on the location of bores and the frequency of meter reading. (2) This value represents only rainfall recharge in the NSW Macro Groundwater Sharing Plan areas. The volume of recharge does not account for recharge in national park areas, which is not available for consumptive use and has been effectively allocated to the environment. (3) The long-term average extraction limit and rainfall recharge assigned to the GAB Cap Rock is not for the deeper GAB confined aquifers but to the GAB Cap Rocks and any associated minor rocks and alluvium to a depth of not more than 60 m. (4) np not provided. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 75

86 6.3.2 Ranking Table 6-1 shows the GMU priority ranking and the assessment ranking for the project. Ranking criteria include the size of the aquifers, the level of development and the assumed degree of connectivity with the surface water system. The ranking helps focus project effort on aquifers that most affect MDB water resources. Groundwater assessment for GMUs can vary from minimal to very thorough, reflecting the availability of data and analysis tools and the priority ranking of the GMU. For the GMUs assessed as part of this region, a simple water balance approach including the mapping of surface groundwater connectivity is used. This is consistent with their low to very low priority ranking. While this limited assessment is appropriate within the constraints and for the terms of reference of the project, additional work may be required for local management of groundwater resources. 6.4 Groundwater levels 6 Groundwater assessment The limited available groundwater level data indicates there are no significant management issues. Bore GW and GW are located downstream of Bourke and monitor the upper and lower alluvial aquifers respectively. Water levels in the upper aquifer show a clear response to flood events (Figure 6-2). The response in the lower aquifer is more muted. Groundwater levels near Wilcannia exhibit a stable to slightly falling trend and a small response due to recharge from flood events (Figure 6-3). The similarity in responses between the upper and lower alluvium layers suggests they are well connected. Bore GW is located near the junction of the Castlereagh and Barwon rivers. This bore monitors the Narrabri Formation and displays a highly fluctuating response (Figure 6-4). This indicates that vertical leakage from flood inundation is a significant recharge mechanism for the Narrabri Formation in this area. RSWL (mahd) GW GW Figure 6-2. Hydrographs for Bores GW and GW monitoring the upper and lower aquifers respectively of the Upper Darling Alluvium near Bourke. These show a response to flood events in the upper aquifer with a lesser response in the lower aquifer 76 Water availability in the Barwon-Darling June 2008 CSIRO 2008

87 RSWL (mahd) GW GW Figure 6-3. Hydrographs for Bores GW and GW monitoring the Upper and Lower aquifers respectively of the Upper Darling Alluvium near Wilcannia. These show a slightly downward trend with a strong similarity between the upper and lower aquifers RSWL (mahd) Groundwater assessment Figure 6-4. Hydrograph for Bore GW monitoring the Narrabri Formation near the Castlereagh-Barwon Junction. This hydrograph shows a response to flood events 6.5 Surface groundwater connectivity An assessment of the surface groundwater connectivity within the region compared river and groundwater levels at a single point in time to provide a snapshot of the direction and magnitude of the exchange. The date selected for production of the flux map and associated calculations was June 2006 as this was the most recent date with a large quantity of bore and river elevation data. It represents a low flow period for the Barwon River with an average level of 2.3 m. River levels vary seasonally and range from 2.0 to 5.0 m on average. The maximum river height for 2006 of 2.4 m was well below average. For the purposes of the flux calculations, a saturated thickness of the shallow alluvial aquifer of 15 m was used for all river reaches. Hydraulic conductivity values vary between 2 and 50 m/day across the region: between 2 and 5 m/day in the west and between 10 and 50 m/day in the intermediate and lower reaches of the Barwon River. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 77

88 6 Groundwater assessment Figure 6-5. Map of surface groundwater connectivity showing losing and gaining river reaches Figure 6-5 shows the estimated fluxes to and from the Barwon-Darling rivers: The lower reaches of the Darling River are classified as low to medium losing, the intermediate reaches are gaining and reaches upstream of the Yanda Creek confluence are low to medium losing. The lower reaches of the Barwon River downstream of the Bokhara River confluence are medium to high losing and the intermediate reaches of the Barwon River between the Bokhara River confluence and Walgett are low to medium gaining. This relationship corresponds to a zone of relatively high hydraulic conductivity of between 10 and 50 m/day. The upper reaches of the Barwon River north of Walgett start to behave differently as the groundwater levels become deeper. All upper reaches are classified as low to medium losing. The river behaves as a maximum losing or disconnected system (there is an unsaturated zone between the river and the watertable) upstream of Collarenebri (reaches 8 to 13). The tributaries of the Upper Barwon River are maximum losing or disconnected. To obtain information on how these fluxes change with time, comparisons were made between river levels and adjacent groundwater levels for five reaches to represent a range of conditions. The analysis showed the same relative relationship was maintained between groundwater level and river stage for most reaches. The groundwater level is below the river along the intermediate reaches of the Darling River (Figure 6-6) and is not consistent with the flux calculations. However, as none of the gauges along this section of river have been surveyed to the Australian Height Datum (AHD) there is considerable error in the analysis. Figure 6-6 shows an elevation difference of 1 to 2 m between groundwater and river baseflow, which is beyond the resolution of the method. Groundwater levels are consistently below the river in the upper reaches of the Darling River and the lower reaches of the Barwon River and suggest that this section of the river is losing. Other studies suggest groundwater discharges to the river in this area (d Hautefeuille and Williams, 2003; Williams, 1991). There is very little difference between groundwater and surface water levels in this area (1 to 2 m) and it is difficult to determine the direction of the flux given the errors associated with the input data. 78 Water availability in the Barwon-Darling June 2008 CSIRO 2008

89 Water level (m AHD) GW (370m from river) GW (2200m from river) Darling Louth Figure 6-6. Comparison of groundwater levels and surface water height for the Lower Darling River at Louth showing low to medium losing conditions 6.6 Water balance 6 Groundwater assessment Groundwater extraction Current groundwater extraction data reported for the assessed GMUs (Table 6-2) are based on extraction estimates supplied by DWE for the Murray-Darling Basin Commission (MDBC, 2007). These estimates do not incorporate the volume set aside for basic rights and were originally provided for whole GMUs. This study divides the GMUs according to region and the extraction volumes are apportioned according to area. Groundwater extraction within the region is forecast to grow. Estimates of the likely maximum extraction were provided for each GMU by DWE (Table 6-2). The rate of growth is not determined but it is assumed in this analysis that full growth will be achieved by The likely maximum use (future extraction) is based on knowledge of the historical development of stock and domestic water supplies. All new domestic and stock water supply works were assumed drilled and constructed on separate properties. An average size for each property was calculated. The total additional stock and domestic requirement was then calculated assuming extraction rates for domestic bores of 2.25 ML/year and for stock bores of ML/year/ha. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 79

90 Table 6-2. Estimated current and future groundwater extraction for the assessed groundwater management units in the Barwon-Darling region 6 Groundwater assessment Current rainfall recharge Code GMUs Current extraction (2004/05) GL/y Future extraction (2030) N45 Lower Darling Alluvium < N46 Upper Darling Alluvium N63 GAB Alluvial N601 GAB Intake Beds N604 Gunnedah Basin N612 Western Murray Porous Rock N620 GAB Cap Rocks N811 Lachlan Fold Belt N813 Warrumbungle Tertiary Basalt < N817 Kanmantoo Fold Belt Total Rainfall recharge is a significant component of the water balance and is the focus of this assessment. Rainfall recharge to the region s aquifers was estimated by DWE using the outcropping area of the aquifers. The method used to determine rainfall is based on Hutchinson and Kesteven (1998). Rainfall stations were selected if they had at least three months with four years of rainfall data and the remaining nine months with a minimum of five years of data between January 1921 and December Rainfall distribution was discretised on 500 m grid squares. The groundwater source area and the rainfall distribution were summed from each grid square to give a volume of rainfall. The water source volume is converted to a recharge volume depending on the dominant rock type. Default recharge rates as a percentage of rainfall were provided on the basis of available porosity and the likely degree of unsaturated profile wetness. An infiltration rate of 6 percent (equivalent to porous sandstone) was assumed by DWE for the GAB Cap Rocks and Gunnedah Basin GMUs. An infiltration rate of 4 percent (representing fractured rock aquifers) was assumed for the Warrumbungle Tertiary Basalt, Western Murray Porous Rock, Kanmantoo Fold Belt and Lachlan Fold Belt GMUs. An infiltration rate of 2 percent (equivalent to inland alluvium) was assumed for the Upper Darling Alluvium, the Lower Darling Alluvium and the GAB Alluvial GMUs. The volume of rainfall recharge does not include recharge in national park areas as this is not available for consumptive use and is effectively allocated to the environment Rainfall recharge modelling The GMU recharge estimated under Scenario A is modified for the other scenarios by multiplying with a recharge scaling factor (RSF). The RSF is 1.0 for Scenario A by definition. The impacts of climate change on recharge are reported as percent changes from Scenario A (Table 6-3). The RSFs are obtained by dividing the percent change by 100 and adding to 1.0. Scenarios Cdry, Cmid and Cwet represent a range of global climate model (GCM) output selected on a ranking of mean annual runoff (Chapter 3). Groundwater recharge is not perfectly correlated with mean annual rainfall or runoff. Apart from mean rainfall, diffuse dryland recharge is sensitive to seasonal rainfall and potential evaporation and also to the extreme events or years that lead to episodic recharge. Extreme events become more important in semi-arid to sub-humid areas. A number of GCMs show an increase in extreme events, but the scenario variants are selected on mean annual runoff which is more dependent on average and seasonal rainfall. Recharge also depends on the land use and soils. These can be locally variable and reflect local spatial variation in RSFs. An estimate for a small GMU will be sensitive to these local variations, while in larger areas with a broader range of soils and land uses the estimates will be more robust. RSFs were estimated for all 15 GCMs under Scenario C. In all 80 Water availability in the Barwon-Darling June 2008 CSIRO 2008

91 cases, a one-dimensional soil-vegetation-atmosphere water transfer model (WAVES) (Zhang and Dawes, 1998) was used for selected points around the MDB for combinations of soils and vegetation. Spatial data on climate, vegetation and soils were then used to extrapolate values to regions. Figure 6-7 shows the percent change in the modelled mean annual recharge averaged over the region under Scenario C relative to Scenario A for the 45 scenarios (15 GCMs for each of the high, medium and low global warming scenarios). The percent change in the mean annual recharge and the percent change in mean annual rainfall from the corresponding GCMs are also tabulated in Table 6-3. The plots show that there is a wide range in results across GCMs and scenarios for the region with about 55 percent of the scenarios predicting less recharge and the remainder predicting more recharge. The high global warming scenario predicts both the highest and lowest change in recharge for the region. Only the scenarios Cdry, Cmid and Cwet are shown in subsequent reporting of modelling results. These variants are based on the runoff modelling and are in bold type in Table 6-3. The choice of GCMs for surface runoff is comparable to those that would be chosen if recharge formed the basis of choice with the second highest, second lowest and median in surface runoff being respectively the third highest, fourth lowest and the 20 th percentile for RSF. The large variability in RSFs is related to the large variability in rainfall produced by the various GCMs. Rainfall and RSFs are correlated, but not perfectly. Some GCMs that indicate reductions in rainfall lead to RSFs greater than 1.0. This is due to more extreme events being more frequent despite a reduction in mean rainfall. Changes in mean annual recharge for GMUs are shown in Table 6-4. The results show a small reduction in recharge under Scenario Cmid, a large increase under Scenario Cwet and small to large decreases under Scenario Cdry. Change in mean annual recharge (percent) 60% High global warming 40% Medium global warming Low global warming 20% 0% -20% -40% -60% ipsl cnrm giss_aom iap csiro mri mpi inmcm gfdl ncar_ccsm ncar_pcm cccma_t63 miub cccma_t47 miroc 6 Groundwater assessment Figure 6-7. Percent change in mean annual recharge from the 45 Scenario C simulations (15 GCMs and three global warming scenarios) relative to Scenario A recharge CSIRO 2008 June 2008 Water availability in the Barwon-Darling 81

92 Table 6-3. Summary results from the 45 Scenario C simulations. Numbers show percent change in mean annual rainfall and recharge under Scenario C relative to Scenario A. Those in bold type have been selected for further modelling 6 Groundwater assessment High global warming Medium global warming Low global warming GCM Rainfall Recharge GCM Rainfall Recharge GCM Rainfall Recharge ipsl -13% -20% cnrm -7% -15% ipsl -4% -6% cnrm -10% -19% ipsl -8% -14% cnrm -3% -6% giss_aom -13% -14% giss_aom -9% -11% giss_aom -4% -4% iap -4% -10% csiro -5% -8% csiro -2% -3% csiro -7% -9% iap -3% -7% iap -1% -2% mri -5% -9% mri -3% -7% mri -1% -2% mpi -6% -5% mpi -4% -4% inmcm -1% -1% inmcm -2% -1% inmcm -2% -2% gfdl -2% 0% gfdl -6% 0% gfdl -4% -1% ncar_ccsm 1% 0% ncar_ccsm 3% 0% ncar_ccsm 2% -1% mpi -2% 2% ncar_pcm 8% 14% ncar_pcm 5% 9% ncar_pcm 2% 4% cccma_t63 8% 17% cccma_t63 5% 10% cccma_t63 2% 5% miub 14% 29% cccma_t47 8% 16% miub 4% 7% cccma_t47 12% 32% miub 9% 17% cccma_t47 3% 7% miroc 13% 37% miroc 8% 22% miroc 4% 10% The rainfall for some GCM simulations in Table 6-3 differs very slightly (no more than 1 percent) from the analogous table in Chapter 3 due to use of an earlier version of data in the recharge modelling assessment. The timeframes of the project precluded use of the revised climate data for the recharge modelling. This inconsistency would not significantly affect the values of the estimated RSFs. Table 6-4. Change in mean annual recharge for groundwater management units in the Barwon-Darling region under Scenario C relative to Scenario A Code Name Cdry Cmid Cwet percent change relative to Scenario A N45 Lower Darling Alluvium -35% 3% 19% N46 Upper Darling Alluvium -3% -7% 41% N63 GAB Alluvial -3% 2% 32% N601 GAB Intake beds -2% -1% 32% N604 Gunnedah Basin -2% -1% 33% N612 Western Murray Porous Rock -29% -1% 21% N620 GAB Cap Rocks -2% -1% 32% N811 Lachlan Fold Belt -22% -4% 18% N817 Kanmantoo Fold Belt -7% -4% 40% Ratio of extraction to recharge Scaled recharge values under scenarios A and C are listed in Table 6-5. The ratio of current (2004/05) groundwater extraction to rainfall recharge is displayed in Table 6-6. Groundwater extraction is very low compared to recharge for these GMUs, leading to very low E/R ratios, which are maintained even under future scenarios of substantially reduced recharge. Projections of substantially higher future extraction (Table 6-7) move the Gunnedah Basin, GAB Alluvial, Western Murray Porous Rock and Kanmantoo Fold Belt GMUs to a medium level of development under Scenario Dmid. The Lower Darling Alluvium GMU would also reach a medium level of development under Scenario Ddry. 82 Water availability in the Barwon-Darling June 2008 CSIRO 2008

93 Table 6-5. Scaled recharge for assessed groundwater management units under scenarios A and C Code Name Recharge Scaled recharge A Cdry Cmid Cwet GL/y N45 Lower Darling Alluvium N46 Upper Darling Alluvium N63 GAB Alluvial N601 GAB Intake Beds np N604 Gunnedah Basin N612 Western Murray Porous Rock N620 GAB Cap Rock N811 Lachlan Fold Belt N813 Warrumbungle Tertiary Basalt N817 Kanmantoo Fold Belt Total Average scaling factor np not provided. Table 6-6. Comparison of current groundwater extraction with scaled rainfall recharge for assessed groundwater management units under scenarios A and C Code Name Current extraction (2004/05) GL/y E/R Scaled E/R A Cdry Cmid Cwet N45 Lower Darling Alluvium < N46 Upper Darling Alluvium N63 GAB Alluvial N601 GAB Intake Beds 0.2 np np np np N604 Gunnedah Basin N612 Western Murray Porous Rock N620 GAB Cap Rocks N811 Lachlan Fold Belt N813 Warrumbungle Tertiary Basalt < N817 Kanmantoo Fold Belt np not provided. ratio 6 Groundwater assessment CSIRO 2008 June 2008 Water availability in the Barwon-Darling 83

94 Table 6-7. Comparison of future groundwater extraction with scaled rainfall recharge for assessed groundwater management units under Scenario D 6 Groundwater assessment Code Name Future extraction GL/y Scaled E/R Ddry Dmid Dwet N45 Lower Darling Alluvium N46 Upper Darling Alluvium N63 GAB Alluvial N601 GAB Intake Beds 1.6 np np np N604 Gunnedah Basin N612 Western Murray Porous Rock N620 GAB Cap Rock N811 Lachlan Fold Belt N813 Warrumbungle Tertiary Basalt N817 Kanmantoo Fold Belt np not provided Impact of extraction on streamflow A percentage estimate was made by DWE of the volume of groundwater pumping that is derived from streamflow (captured discharge or induced leakage) for each GMU based on detailed knowledge of the aquifer stratigraphy and geomorphology. Extraction within the region may take some time to fully impact on rivers depending largely on the distance of groundwater extraction from the river and the characteristics of the local geology. An estimate has been made of the likely timing of the full impacts on surface water flows based on these factors using information from the Murray Darling-Basin Commission (MDBC, 2007). ratio The calculations (Table 6-8) indicate that surface groundwater impacts would increase from 1.3 GL/year currently to 37 GL/year in 2030, once equilibrium is reached. The estimated impacts of groundwater extraction on streamflow are assigned a low level of confidence. The estimates are sensitive to the connectivity factor used and may be over estimated given the methodology. For this reason, these values have not been included in river modelling. Table 6-8. Surface groundwater connectivity showing an estimate of the volumetric impact extraction has on streamflow Code Name Degree of connectivity* Impact of extraction on streamflow (2004/05) Impact of extraction on streamflow (2030) Time lag to reach full impact percent GL/y years N45 Lower Darling Alluvium 0% N46 Upper Darling Alluvium 12% N63 GAB Alluvial 17% N601 GAB Intake Beds 0% >100 N604 Gunnedah Basin 26% N612 Western Murray Porous Rock 0% >100 N620 GAB Cap Rock 0% >100 N811 Lachlan Fold Belt 30% N813 Warrumbungle Tertiary Basalt 31% < N817 Kanmantoo Fold Belt 0% >100 Total Source: MDBC (2007), except that some values have been lowered due to wide floodplain areas. *These values are the percentage of groundwater pumped that is derived from streamflow. 84 Water availability in the Barwon-Darling June 2008 CSIRO 2008

95 6.7 Discussion of key findings Groundwater use in the Barwon-Darling region in 2004/05 was 10 GL or 0.6 percent of the total groundwater use in the MDB (not including the confined aquifers of the GAB). Current extraction is a very low proportion of current rainfall recharge so one expects limited impacts on the groundwater resources of the region. Alluvial aquifers may receive additional recharge from flood flows. There is estimated to be only 1.3 GL/year reduction in streamflow from groundwater extraction. The Barwon and Darling rivers have a range of gaining and losing reaches dependent upon groundwater depth and varying hydraulic conductivity. Little is expected to change under Scenario C. Although rainfall recharge may increase or decrease substantially under particular scenarios the E/R ratio remains low in all cases. The most substantial projected changes are the result of future development, with groundwater use projected to grow to 240 GL/year, a 24-fold increase. This would produce a medium level of development for the Gunnedah Basin, the Upper Darling Alluvium, the GAB Alluvial, the Western Murray Porous Rock and the Kanmantoo Fold Belt under Scenario Dmid. The Lower Darling Alluvium would also be of medium level development under Scenario Ddry. The greater extraction rate would reduce streamflow substantially (by around 37 GL/year) once equilibrium was reached. The majority of these losses would occur in the GAB Alluvial GMU where the largest extraction increase is predicted. The estimated impacts of groundwater extraction on streamflow are assigned a low level of confidence. The estimates are sensitive to the connectivity factor used and may be over estimated given the methodology. 6.8 References d Hautefeuille F and Williams M (2003) Upper Darling salt interception scheme preliminary investigation to May Centre for Natural Resources, NSW Department of Infrastructure, Planning and Natural Resources. DNR (2007) Unregulated Flow Management Plan for the Northwest. New South Wales Department of Natural Resources, Sydney. Hutchinson MF and Kesteven JL (1998) Monthly Mean Climate Surfaces for Australia. Cres ANU (unpubl.). MDBC (2007) Updated summary of estimated impact of groundwater extraction on streamflow in the Murray-Darling Basin. Draft Report. Prepared by REM on behalf of MDBC Canberra. Williams RM (1991) Groundwater inflows to the Darling River, Mungindi to Wentworth: Run of River Study 1990/91. TS Zhang L and Dawes WE (1998) WAVES An integrated energy and water balance model. CSIRO Land and Water Technical Report No 31/98. 6 Groundwater assessment CSIRO 2008 June 2008 Water availability in the Barwon-Darling 85

96 7 Environment This chapter presents the environmental assessments undertaken for the Barwon-Darling region. It has four sections: a summary an overview of the approach a presentation of results a discussion of key findings. 7.1 Summary 7 Environment Issues and observations Assessment of the environmental implications of changes in water availability is largely beyond the terms of reference (Chapter 1) of this project. The exception is reporting against environmental water allocations and quantified environmental flow rules specified in water sharing plans. Otherwise, environmental assessments form a very small part of the project. This report deals with the main stem of the Barwon-Darling River to the Menindee Lakes. This river receives water from many tributaries that have different water sharing arrangements as reported in the relevant region. The Talyawalka Anabranch and Teryawynia Creek system contains wetlands of national importance that are just outside the region but are watered by flows from the Darling River within the region and are thus reported on here. The Barwon-Darling River itself has interim rules for water sharing that consider some environmental requirements including fish passage over Bourke Weir. The results presented here are based on river modelling for the Barwon-Darling that incorporates the outputs of upstream river models. Some Queensland upstream river models assume maximum allowable water resource development under current Water Resource Plans and Resource Operations Plans, while others essentially represent current use (Chapter 4). The simulated river flows for current development therefore reflect a level of development that is slightly greater than has actually yet been realised Key messages Current water resource development in the Barwon-Darling region and upstream contributing regions has nearly doubled the average and maximum periods between important flows to the Talyawalka Anabranch system. Individual inflow events are larger on average than under without-development conditions, but due to lower frequency of events the total volumes entering the Anabranch system are also lower. Current water resources development has more than doubled the average period between events that drown out Bourke Weir (thereby allowing fish passage) compared to without-development conditions. Individual drown-out events are slightly larger and last slightly longer. Under the best estimate 2030 climate Talyawalka Anabranch system inflow events would be slightly more frequent but smaller in volume. The average period between inflows and the individual event volume would be reduced by 8 and 18 percent, respectively. The average period between drown-outs at Bourke Weir would increase by 18 percent, however, the duration of drown-out events would be largely unaffected. Under the dry extreme 2030 climate there would be large changes to flows to the Talyawalka Anabranch system. The average period between inflows would increase by over 30 percent and the average volume of these events would be reduced by more than 30 percent. The average period between drown-outs at Bourke Weir would be increased by 66 percent and the average event volume would be reduced by 10 percent indicating shorter drown-out durations. The ecological consequences of such changes could be severe. Under the wet extreme 2030 climate Talyawalka Anabranch system inflows would be similar to without-development values. Drown-outs at Bourke Weir would be more frequent, although less than under 86 Water availability in the Barwon-Darling June 2008 CSIRO 2008

97 without-development conditions. Drown-out event volumes would be slightly smaller than under without-development conditions indicating slightly shorter drown-out event durations. Future development would have only minor additional effects on the hydrology of the Talyawalka Anabranch system and the frequency and magnitude of Bourke Weir drown-outs Uncertainty The main uncertainties involving analysis and reporting include: Aquatic and wetland ecosystems are highly complex and many factors in addition to water regime can affect ecological features and processes, such as water quality and land use practices. The indicators are based on limited hydrology parameters with no direct quantitative relationships for environmental responses. This project only makes general observations on the potential implications of changed water regimes and some related ecological responses. Considering only a few of the important environmental assets and using a limited number of indicators to represent overall aquatic ecosystem outcomes is a major simplification. Actual effects on these and other assets or localities are likely to vary. Uncertainties expressed in Chapters 3, 4 and 5 affect the hydrological information used in the environmental assessments. 7 Environment 7.2 Approach This chapter focuses on the specific rules that apply to the provision of environmental water in the Barwon-Darling region and on the assessment of hydrological indicators defined by prior studies for key environmental assets in the region. A broader description of the catchment, water resources and important environmental assets is provided in Chapter Summary of environmental flow rules There is no water sharing plan for the Barwon-Darling region. However, in late 1997, the New South Wales Government established a River Management Committee for the Barwon-Darling to recommend environmental flow rules (DLWC, 1998). The rules were not implemented until late 2000, due to delays in the establishment of five new streamflow gauging stations, which are critical to the full implementation of the rules. None of the flow rules are specifically designed to benefit the wetlands of the system, although they may provide some indirect benefit to wetlands. The rules are as follows: 1. New commence-to-pump conditions apply to all water users in the Barwon-Darling from 2000/01 onwards. (The new commence-to-pump conditions for difference licence classes and different sections of the river improve the protection of low flows). 2. The Unregulated Flow Management Plan for the Northwest (DNR, 2007) will continue to operate. The plan covers all rivers in the north-west and the Barwon-Darling River. It ensures minimum flows for the protection of basic river health as well as protecting high flows for algal suppression and fish migration. The plan targets the fish passage over Brewarrina and Bourke weirs and the control of blue-green algae Environmental assets and indicators There are many nationally important wetlands and some internationally important (Ramsar-listed) wetlands in the Barwon-Darling region including parts of the Paroo River Wetlands. Most of these are dependent on flows from the tributaries of the Barwon-Darling River. Consequently the assessments for these wetlands are reported in the reports for the relevant region; for example, the Paroo River Wetlands are reported on in CSIRO (2007). Talyawalka Anabranch and Teryawynia Creek are located downstream of the Barwon-Darling region but they are assessed and reported here because they receive water directly from the Darling River (Figure 7-1). Fish passage over the Bourke Weir along the main stem of the Darling River is also assessed. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 87

98 7 Environment Figure 7-1. Location map of environmental assets assessed in this chapter Talyawalka Anabranch and Teryawynia Creek (NSW012) The Talyawalka Anabranch and Teryawynia Creek system is a nationally important wetland (Figure 7-2) but there is limited available information as there are few scientific studies. The descriptions below are from Environment Australia (2001) unless otherwise cited. The Talyawalka Anabranch and Teryawynia Creek system is a distributary flow system that receives water from the Darling River near Wilcannia and returns water via the lower Darling River downstream of Menindee. The system consists of Teryawynia, Dry, White Water, Eucalyptus/Waterloo, Victoria, Brummeys, Dennys, Brennans, Sayers, Gum, Boolabooka, North and Ratcatchers lakes. Seddon et al. (1997) mapped the lakes of the area as part of a wider inventory, but recorded few details of the lakes characteristics. These areas receive water when the Darling River is in flood so the upstream lakes receive water before and more frequently than the downstream lakes. The main vegetation of the inundation area is Black Box (Eucalyptus largiflorens) and River Red Gum (E. camaduelensis). Dry Lake supports a large expanse of Lignum (Muehlenbeckia florulenta) and Cane Grass (Eragrostis australiasica) is prevalent along Talyawalka Creek. The lakes provide a large area of habitat for waterbirds when flooded. Kingsford et al. (1997) list the Talyawalka wetlands as being important for the Murray-Darling Basin on the basis of the area carrying more than 10,000 waterbirds during surveys. Past Aboriginal occupation of the lake area is evident. Land tenure is leasehold and the lakes are used for grazing and occasional lakebed cropping. 88 Water availability in the Barwon-Darling June 2008 CSIRO 2008

99 7 Environment Figure 7-2. Satellite image of Talyawalka Lakes, showing in yellow the areas included for this site listed in the Directory of Important Wetlands in Australia (Environment Australia, 2001) No specific hydrological or ecological studies of the Talyawalka system recommend hydrologic indicators of ecological relevance. Information from DWE identified a commence-to-flow threshold to the Talyawalka system from the Darling River of 30 GL/day as measured at the Wilcannia gauge (R. Cooke, pers. comm.). The frequency and volume of threshold events provide information about the magnitude of the inundation events (Table 7-1). Fish passage The Unregulated Flow Management Plan for the Northwest (DNR, 2007) protects flows that provide for fish passage over weirs. It also aims to protect flows that reduce the incidence of blue-green algal blooms. The Murray-Darling Basin Commission (MDBC, 2004) lists Brewarrina Weir as a priority barrier that requires a fishway (the second highest priority site in the Barwon-Darling after Menindee Weir). Cooney (1994) identified that flows in the vicinity of 14.3 GL/day are required to drown-out the weir and permit fish passage. A fishway is under construction at this weir (Webb, McKeown and Associates, 2007). Bourke Weir is the next priority for fish passage (MDBC, 2004). Flows in the vicinity of 10 GL/day drown out this weir and allow for fish passage (Cooney, 1994). As a fishway is not constructed or planned for Bourke Weir this drown-out volume is used in the assessments of this project. Both the frequency and magnitude of flows above the 10 GL/day threshold are assessed (Table 7-1). The magnitude of the events (assessed as the total volume above the 10 GL/day threshold) is an indicator of the duration of weir drown-out and indicates the relative length of time for fish passage. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 89

100 Table 7-1. Definition of environmental indicators Indicator Talyawalka Anabranch commence-to-flow Average period between events Maximum period between events Average excess volume per event Average excess volume per year Bourke Weir drown-out flow Average period between events Maximum period between events Average excess volume per event Average excess volume per year Description Average period (years) between events 30 GL/day at Wilcannia gauge Maximum period (years) between events 30 GL/day at Wilcannia gauge Average volume (GL) per event above 30 GL/day at Wilcannia gauge Average volume (GL) per year above 30 GL/day at Wilcannia gauge Average period (years) between events 10 GL/day at Bourke Maximum period (years) between events 10 GL/day at Bourke Average volume (GL) per event above 10 GL/day at Bourke Average volume (GL) per year above 10 GL/day at Bourke 7 Environment 7.3 Results The projected changes in the environmental indicators are listed for the various scenarios in Table 7-2 using the outputs from the river models used for the Barwon-Darling region (Chapter 4). Table 7-2. Environmental indicator values under scenarios P and A, and percent change (from Scenario A) in indicator values under scenarios C and D P A Cdry Cmid Cwet Ddry Dmid Dwet years percent change from Scenario A Talyawalka Anabranch commence-to-flow Average period between events % -8% -39% 32% -8% -36% Maximum period between events % 53% -37% 53% 53% -37% GL Average excess volume per event % -18% -8% -35% -21% -6% Average excess volume per year % -11% 55% -49% -14% 51% Bourke Weir drown-out flow years Average period between events % 18% -24% 70% 22% -22% Maximum period between events % 1% -10% 66% 1% -10% GL Average excess volume per event % 2% 28% -12% 2% 27% Average excess volume per year % -13% 61% -47% -16% 57% 7.4 Discussion of key findings Flows to the Talyawalka system Under current water resource development in the Barwon-Darling region and upstream contributing regions, the average period between events into the Talyawalka system has increased by over 80 percent (nearly two years) relative to without-development conditions, and the maximum period between events has doubled to almost 19 years. The inflow volumes are increased by 32 percent, but due to the reduced event frequency, the average annual inflow volumes are decreased by about 35 percent. Although the simulated level of development is slightly greater than that which has occurred to-date, overall flow regime changes are likely to have reduced waterbird breeding opportunities and the condition and extent of native vegetation within the Talyawalka system. 90 Water availability in the Barwon-Darling June 2008 CSIRO 2008

101 Under the best estimate 2030 climate the average period between inundation events would decrease by 8 percent but the maximum period between events would increase by 53 percent (about ten years). The average per event excess volume and the average annual excess volume would be reduced by 18 percent and 11 percent, respectively. Under the dry extreme 2030 climate the average period between events would be increased by 31 percent (more than one year) and the maximum period between events would be increased by 53 percent (about ten years). The average per event excess volume and the average annual excess volume would be reduced by 32 percent and 47 percent, respectively. Under the wet extreme 2030 climate the average and maximum period between events would be similar to that seen under without-development conditions. The average excess event volume would be reduced by 8 percent. However, due to the increased frequency, the average annual excess volume would be similar to the withoutdevelopment value. These changes would restore more natural hydrological and ecological conditions. Conditions between the best estimate 2030 climate and the dry extreme 2030 climate would lead to extended disconnection of the Darling River and the Talyawalka system and lower flood volumes. This would further adversely affect the condition of wetland vegetation and waterbird breeding. Future development would have very minor additional effects on the magnitude and frequency of inundation of the Talyawalka Anabranch system Bourke Weir drown-out flows 7 Environment Under current maximum allowable water resource development in the Barwon-Darling region and upstream contributing regions the average period between Bourke Weir drown-out events increases from about once every three months under without-development conditions to nearly once every seven months. The maximum period between events is doubled to 3.6 years. Webb, McKeown and Associates (2007) found nearly a 50 percent reduction in the average number of days of weir drown-out from without-development to current conditions. The average excess volume of water is increased by 4 percent (indicating a longer duration of drown-out periods) but the average annual excess volume is decreased by about 45 percent. Overall these changes have reduced the opportunities for native fish passage over Bourke Weir. Under the best estimate 2030 climate (Scenario Cmid) the average period between weir drown-outs would increase by 18 percent from Scenario A but change in the maximum period between events would be minor. The average excess volume per event would not change significantly, but the decrease in frequency would mean that the average annual excess volumes would be reduced by 13 percent. Under the dry extreme 2030 climate (Scenario Cdry) the average period between weir drown-outs would increase by 66 percent (over three months) relative to Scenario A and the maximum period between events would also be increased by 66 percent (about two years). The average excess volume per event and per year would be reduced by 10 percent and 44 percent, respectively. Conditions between the best estimate 2030 climate and the dry 2030 climate extreme would lead to extended periods without connectivity for fish migration over Bourke Weir. Under the wet 2030 climate extreme the average and maximum period between events would be reduced, but would still be considerably longer than that predicted under without-development conditions. The average excess volumes per event and per year would be increased to 90 percent of the without-development value. These changes would considerably improve fish passage conditions at Bourke. Future development would have only minor additional effects on the magnitude and frequency of weir drown-out flow at Bourke. 7.5 References Cooney T (1994) North-west weirs and fish passage report. Prepared for the Interim North-West unregulated flow management plan, Department of Water Resources, Technical Services Report TS CSIRO (2007) Water availability in the Paroo. A report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project. CSIRO, Canberra. 88pp. DLWC (1998) Building a more secure future for the Barwon-Darling. River Management Committee Report HO/25/98. Available at DNR (2007) Unregulated Flow Management Plan for the Northwest. New South Wales Department of Natural Resources, Sydney. Environment Australia (2001) A Directory of Important Wetlands in Australia. Third Edition. Environment Australia, Canberra. CSIRO 2008 June 2008 Water availability in the Barwon-Darling 91

102 Kingsford RT, Thomas RF and Wong PS (1997) Significant wetlands for waterbirds in the Murray-Darling Basin. National Parks and Wildlife Service, Hurstville. MDBC (2004) Native Fish Strategy for the Murray-Darling Basin Murray-Darling Basin Commission, Canberra ACT. ISBN Seddon J, Thornton S and Briggs S (1997) An inventory of Lakes in the Western Division of New South Wales. NSW National Parks and Wildlife Service, Lyneham ACT. Webb, McKeown and Associates (2007) State of the Darling. Interim Hydrology Report to the Murray-Darling Basin Commission. ISBN Environment 92 Water availability in the Barwon-Darling June 2008 CSIRO 2008

103 Appendix A Rainfall-runoff results for all subcatchments Table A-1. Summary of modelling results for all subcatchments under scenarios A and C Modelling catchment Area Rainfall APET Runoff Runoff coefficient Scenario A Scenario Cdry Scenario Cmid Scenario Cwet Runoff contribution Rainfall Runoff Rainfall Runoff Rainfall Runoff km 2 mm percent percent change from Scenario A % 2% -13% -11% -2% -3% 13% 43% % 6% -10% -28% -3% -3% 13% 50% % 11% -10% -31% -3% -5% 13% 53% % 24% -11% -22% -3% -2% 13% 49% % 58% -14% -21% -3% -1% 13% 49% % 100% -13% -22% -3% -2% 13% 50% Modelling catchment Table A-2. Summary of modelling results for all subcatchments under scenarios A and D A runoff Plantations increase Farm dam increase Ddry runoff Dmid runoff Dwet runoff mm ha ML ML/km 2 percent change from Scenario A % -4% 42% % -4% 49% % -7% 51% % -3% 49% % -1% 49% % -2% 49% Appendix A Rainfall-runoff results for all subcatchments CSIRO 2008 June 2008 Water availability in the Barwon-Darling 93

104 Appendix B River modelling reach mass balances Appendix B River modelling reach mass balances A0 A Cwet Cmid Cdry Dwet Dmid Ddry Model start date 01/07/ /07/ /07/ /07/ /07/ /07/ /07/1895 0/01/1900 Model end date 30/06/ /06/ /06/ /06/ /06/ /06/ /06/2006 0/01/1900 GL/y percent change from Scenario A Inflows Subcatchments Directly gauged % -11% -38% 42% -14% -41% Indirectly gauged % -7% -28% 46% -9% -29% Sub-total % -10% -36% 43% -13% -38% Outflows End-of-catchment flows % -10% -35% 43% -13% -38% Net evaporation* From river % 3% 3% 3% 2% 3% Sub-total % -10% -35% 43% -13% -37% Unattributed fluxes River unattributed loss % -12% -41% 45% -15% -43% Mass balance Mass balance error (%) 0% 0% 0% 0% 0% 0% 0% 0% * Evaporation from private licensed storages (GL/y) is not included as it is already accounted in diversions 94 Water availability in the Barwon-Darling June 2008 CSIRO 2008

105 A0 A Cwet Cmid Cdry Dwet Dmid Ddry Model start date 01/02/ /07/ /07/ /07/ /07/ /07/ /07/1895 0/01/1900 Model end date 30/06/ /06/ /06/ /06/ /06/ /06/ /06/2006 0/01/1900 GL/y percent change from Scenario A Inflows Subcatchments Directly gauged (Walgett) % -10% -38% 43% -14% -41% Macquarie/Castlereagh % -9% -29% 40% -13% -33% Condamine/Balonne % -12% -36% 23% -12% -36% Indirectly gauged % -2% -22% 49% -2% -22% Effluent return % -14% -48% 66% -18% -51% Sub-total % -11% -37% 44% -14% -40% Diversions Major irrigators Class A % 11% 40% -10% 18% 47% Class B % 2% -2% 1% 1% -4% Class C % 0% -8% 5% -1% -10% Floodplain harvest % 3% -25% 24% 1% -28% Reach irrigators % 1% -3% 5% 0% -4% Sub-total % 1% -4% 3% 0% -6% Outflows Subcatchment effluent % -14% -48% 66% -18% -51% End-of-catchment flows % -11% -37% 43% -14% -40% Net evaporation* From river % 5% 9% -1% 5% 9% Sub-total % -11% -39% 46% -14% -41% Unattributed fluxes River unattributed loss % -12% -39% 34% -20% -47% Mass balance Mass balance error (%) 0% 0% 0% 0% 0% 0% 0% 0% Evaporation from on farm storages * % 2% -4% 5% 1% -6% Rainfall harvest % -3% -13% 7% -4% -15% * Evaporation from private licensed storages (GL/y) is not included as it is already accounted in diversions Appendix B River modelling reach mass balances CSIRO 2008 June 2008 Water availability in the Barwon-Darling 95

106 Appendix B River modelling reach mass balances A0 A Cwet Cmid Cdry Dwet Dmid Ddry Model start date 01/07/ /07/ /07/ /07/ /07/ /07/ /07/1895 0/01/1900 Model end date 30/06/ /06/ /06/ /06/ /06/ /06/ /06/2006 0/01/1900 GL/y percent change from Scenario A Inflows Subcatchments Directly gauged Border Rivers % -12% -34% 23% -15% -37% Gwydir % -7% -29% 35% -9% -31% Moonie % -20% -39% 43% -22% -40% Namoi % -9% -41% 56% -12% -44% Sub-total % -10% -37% 41% -13% -39% Diversions Major irrigators Class A Class B % 6% 3% -4% 5% 2% Class C % 2% -8% 5% 2% -6% Floodplain harvest % -15% -38% 15% -17% -40% Reach irrigators % 5% 3% 1% 4% 2% Sub-total % 6% 2% -4% 4% 1% Outflows End-of-catchment flows % -10% -38% 43% -14% -41% Net evaporation* From river % 6% 9% -3% 6% 9% Sub-total % -10% -38% 43% -14% -41% Unattributed fluxes River unattributed loss % -11% -30% 31% -14% -33% Mass balance Mass balance error (%) 0% 0% 0% 0% 0% 0% 0% 0% Evaporation from OFS * % 3% -4% 5% 1% -5% Rainfall harvest % -1% -3% 2% -1% -4% * Evaporation from private licensed storages (GL/y) is not included as it is already accounted in diversions 96 Water availability in the Barwon-Darling June 2008 CSIRO 2008

107 A0 A Cwet Cmid Cdry Dwet Dmid Ddry Model start date 01/07/ /07/ /07/ /07/ /07/ /07/ /07/1895 0/01/1900 Model end date 30/06/ /06/ /06/ /06/ /06/ /06/ /06/2006 0/01/1900 GL/y percent change from Scenario A Storage volume Change over period % 1% 3% -8% 1% 3% Inflows Subcatchments Directly gauged (Bourke) % -11% -37% 43% -14% -40% Warrego % -7% -27% 47% -7% -27% Sub-total % -11% -37% 43% -13% -39% Diversions Major irrigators Class A Class B % 0% -9% 3% -2% -11% Class C Floodplain harvest % -4% -29% 40% -4% -29% Reach irrigators % -2% -14% 12% -4% -16% Sub-total % -1% -11% 7% -3% -13% Outflows End-of-catchment flows % -11% -38% 42% -14% -41% Net evaporation* Warrego storage % 0% -11% 13% 0% -11% From river % -4% -27% 30% -6% -28% Evaporation % -3% -22% 25% -4% -23% Sub-total % -11% -37% 41% -13% -40% Unattributed fluxes River unattributed loss % -12% -38% 60% -15% -41% Mass balance Mass balance error (%) 0% 0% 0% 0% 0% 0% 0% 0% Evaporation from OFS * % 0% -11% 10% -2% -13% Rainfall harvest % -4% -16% 7% -5% -17% * Evaporation from private licensed storages (GL/y) is not included as it is already accounted in diversions Appendix B River modelling reach mass balances CSIRO 2008 June 2008 Water availability in the Barwon-Darling 97

108 Appendix C River system model uncertainty assessment by reach This Appendix contains the results of river reach water accounting for this region, as well as an assessment of the magnitude of the projected change under each scenario compared to the uncertainty associated with the river model. Each page provides information for a river reach that is bounded by a gauging station on the upstream and downstream side, and for which modelling results are available. Table C-1 provides a brief explanation for each component of the results page. Appendix C River system model uncertainty assessment by reach Table Land use Gauging data Correlation with ungauged gains/losses Water balance Description Table C-1. Explanation of components of the uncertainty assessments Information on the extent of dryland, irrigation and wetland areas. Land use areas are based on remote sensing classification involving BRS land use mapping, water resources infrastructure and remote sensing-based estimates of actual evapotranspiration. Information on how well the river reach water balance is measured or, where not measured, can be inferred from observations and modelling. The volumes of water measured at gauging stations and off-takes is compared to the grand totals of all inflows or gains, and/or all outflows or losses, respectively. The fraction of total refers to calculations performed on average annual flow components over the period of analysis. The fraction of variance refers to the fraction of month-tomonth variation that is measured. Also listed are the same calculations but for the sum of gauged terms plus water balance terms that could be attributed to the components listed in the Water balance table with some degree of confidence. The same terms are also summed to water years and shown in the diagram next to this table. Information on the likely nature of ungauged components of the reach water balance. Listed are the coefficients of correlation between ungauged apparent monthly gains or losses on one hand, and measured components of the water balance on the other hand. Both the normal (parametric) and the ranked (or non-parametric) coefficient of correlation are provided. High coefficients are highlighted. Positive correlations imply that the apparent gain or loss is large when the measured water balance component is large, whereas negative correlation implies that the apparent gain or loss is largest when the measured water balance component is small. In the diagram below this table, the monthly flows measured at the gauge at the end of the reach are compared with the flows predicted by the baseline river model, and the outflows that could be accounted for (i.e., the net result of all measured or estimated water balance components other than main stem outflow which ideally should equal main stem outflows in order to achieve mass balance). Information on how well the modelled and the best estimate river reach water balances agree, and what the nature of any unspecified losses in the river model is likely to be. The river reach water balance terms are provided as modelled by the baseline river model (Scenario A) over the period of water accounting. The accounted terms are based on gauging data, diversion records, and (adjusted) estimates derived from SIMHYD rainfall-runoff modelling, remote sensing of water use and simulation of temporary storage effects. Neither should be considered as absolutely correct, but large divergences point to large uncertainty in river modelling. Model efficiency Information on the performance of the river model in explaining historic flow patterns at the reach downstream gauge, and the scope to improve on this performance. All indicators are based on the Nash-Sutcliffe model efficiency (NSME) indicator. In addition to the conventional NSME calculated for monthly and annual outflows, it has also been calculated after log-transformation or ranking of the original data, as well as having been calculated for the 10% of months with highest and lowest observed flows, respectively. Using the same formulas, the model efficiency of the water accounts in explaining observed outflows is calculated. This provides an indication of the scope for improving the model to explain more of the observed flow patterns: if NSME is much higher for the water accounts than for the model, than this suggests that the model can be improved upon and model uncertainty reduced. Conversely, if both are of similar magnitude, then it is less likely that a better model can be derived without additional observation infrastructure. 98 Water availability in the Barwon-Darling May 2008 CSIRO 2008

109 Table Changeuncertainty ratios Description Information on the significance of the projected changes under different scenarios, considering the performance of the river model in explaining observed flow patterns at the end of the reach. In this table, the projected change is compared to the river model uncertainty by testing the hypothesis that the scenario model is about as good or better in explaining observed historic flows than the baseline model. The metric to test this hypothesis is the change-uncertainty ratio, which is calculated as the ratio of Nash-Sutcliffe Model Efficiency indicators for the scenario model and for the baseline (scenario A) model, respectively. A value of around 1.0 or less suggests that is likely that the projected scenario change is not significant when compared to river model uncertainty. Conversely, a ratio that is considerably greater than 1.0 implies that the scenario model is much worse in reproducing historic observations than the baseline model, which provides greater confidence that the scenario indeed leads to a significant change in flow patterns. The change-uncertainty ratio is calculated for monthly as well as annual values, to account for the possibility that the baseline model may reproduce annual patterns well but not monthly. Below this table on the left, the same information is provided in a diagram. Below the table on the right, the observed annual flows at the end of the reach is compared to those simulated by the baseline model and in the various scenarios. To the right of this table, the flow-duration curves are shown for all scenarios. Appendix C River system model uncertainty assessment by reach CSIRO 2008 May 2008 Water availability in the Barwon-Darling 99

110 Downstream gauge Barwon Mogil Mogil Reach 1 Upstream gauge Barwon Mungindi Reach length (km) 59 Area (km 2 ) 1545 Outflow/inflow ratio 2.17 Net gaining reach Land use ha % Dryland 141, Irrigable area - - Open water* - - River and wetlands 13,282 9 Open water* - - * averages for This is a strongly gaining reach. Most inflows are from the Barwon River and originate from the Border Rovers region, but tributary inflows from the Moonie region and local inflows are also considerable. Only two-thirds of total inflows are gauged. Estimated local runoff explains about half of the ungauged gains (adjustment did not improve this). There are some unregulated diversions but they are small, and other losses are also estimated to be very small. Baseline model performance is modest; high flows appear severely underestimated and low flows overestimated. The projected changes are all considered insignificant when compared to model uncertainty. Appendix C River system model uncertainty assessment by reach Gauging data Inflows Outflows Overall and gains and losses Fraction of total Gauged Attributed Fraction of variance Gauged Attributed Correlation with ungauged Gains Losses Linear adjustment normal ranked normal ranked Main gauge inflows Tributary inflows Main gauge outflows Distributary outflows Recorded diversions Estimated local runoff Monthly streamflow (GL/mo) gauged accounted model 0.01 Jun-90 Jun-91 Jun-92 Jun-93 Jun-94 Jun-95 Jun-96 Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Jun-03 Jun-04 Jun-05 Water balance Model (A) Accounts Difference Model efficiency Model (A) Accounts Jul 1990 Jun 2006 Monthly Gains GL/y GL/y GL/y Normal Main stem inflows Log-normalised Tributary inflows Ranked Local inflows Low flows only <0 <0 Unattributed gains and noise High flows only Losses GL/y GL/y GL/y Annual Main stem outflows Normal Distributary outflows Log-normalised Net diversions Ranked River flux to groundwater River and floodplain losses Definitions: Unspecified losses low flows (flows<10% percentile ) : 0.3 GL/mo Unattributed losses and noise high flows (flows>90% percentile) : GL/mo 0.1 Change-uncertainty ratios P B Cwet Cmid Cdry Dwet Dmid Ddry Annual streamflow Monthly streamflow Reach gains and losses (GL/y) /91 91/92 92/93 93/94 94/95 Monthly streamflow (GL/mo). 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/ /04 04/05 05/06 Pecentage of months flow is exceeded unattributed gains ungauged gains gauged gains unattributed losses ungauged losses gauged losses Monthly Change-Uncertainty Ratio P C D + wet O mid dry Annual Change-Uncertainty Ratio Annual streamflow (GL/y) /91 91/92 92/93 93/94 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/06 gauged A P Cwet Cmid Cdry Dwet Dmid Ddry 100 Water availability in the Barwon-Darling May 2008 CSIRO 2008

111 Downstream gauge Barwon Collarenebri Reach 2 Upstream gauge Barwon Mogil Mogil Reach length (km) 36 Area (km 2 ) 463 Outflow/inflow ratio 0.83 Net losing reach Land use ha % Dryland 42, Irrigable area - - Open water* - - River and wetlands 3,985 9 Open water* - - * averages for This is a losing reach. Inflows are dominated by main stem inflows. Most of the inflows are gauged. Estimated local runoff does not explain some of the ungauged gains. There are some recorded diversions. Baseline model performance is good. Accounting explains monthly flows extremely well, when large river and floodplain losses are considered. The projected changes are rather small compared to model uncertainty because of the tendency of the model to underestimate high fows and overestimate low flows. Gauging data Inflows Outflows Overall and gains and losses Fraction of total Gauged Attributed Fraction of variance Gauged Attributed Correlation with ungauged Gains Losses Linear adjustment normal ranked normal ranked Main gauge inflows Tributary inflows Main gauge outflows Distributary outflows Recorded diversions Estimated local runoff Adjusted % Monthly streamflow (GL/mo) gauged accounted model Jun-90 Jun-91 Jun-92 Jun-93 Jun-94 Jun-95 Jun-96 Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Jun-03 Jun-04 Jun-05 Water balance Model (A) Accounts Difference Model efficiency Model (A) Accounts Jul 1990 Jun 2006 Monthly Gains GL/y GL/y GL/y Normal Main stem inflows Log-normalised - - Tributary inflows Ranked Local inflows Low flows only <0 <0 100 Unattributed gains and noise High flows only Losses GL/y GL/y GL/y Annual 10 Main stem outflows Normal Distributary outflows Log-normalised Net diversions Ranked River flux to groundwater 0-0 River and floodplain losses Definitions: 0.1 Unspecified losses low flows (flows<10% percentile ) : 0.4 GL/mo Unattributed losses and noise high flows (flows>90% percentile) : GL/mo Change-uncertainty ratios P B Cwet Cmid Cdry Dwet Dmid Ddry Annual streamflow Monthly streamflow Reach gains and losses (GL/y) /91 91/92 92/93 93/94 94/95 Monthly streamflow (GL/mo). 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/ /04 04/05 05/06 Pecentage of months flow is exceeded unattributed gains ungauged gains gauged gains unattributed losses ungauged losses gauged losses Appendix C River system model uncertainty assessment by reach Monthly Change-Uncertainty Ratio P B C D + wet O mid dry Annual Change-Uncertainty Ratio Annual streamflow (GL/y) /91 91/92 92/93 93/94 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/06 gauged A P Cwet Cmid Cdry Dwet Dmid Ddry CSIRO 2008 May 2008 Water availability in the Barwon-Darling 101

112 Downstream gauge Barwon Danger Bridge Reach 3 Upstream gauge Barwon Collarenebri Reach length (km) 83 Area (km 2 ) 5765 Outflow/inflow ratio 2.06 Net gaining reach Land use ha % Dryland 550, Irrigable area - - Open water* - - River and wetlands 25,815 4 Open water* - - * averages for This is a strongly gaining reach. Inflows are derived equally from upstream and from the Namoi River. About three-quarters of inflows are gauged. Estimated local runoff explains some of the ungauged gains after adjustment, but unattributed agains and losses remain. There are some diversions. Baseline model performance is good. Accounting explains monthly flows very well. Some of the projected changes are greater than model uncertainty. Appendix C River system model uncertainty assessment by reach Gauging data Inflows Outflows Overall and gains and losses Fraction of total Gauged Attributed Fraction of variance Gauged Attributed Correlation with ungauged Gains Losses Linear adjustment normal ranked normal ranked Main gauge inflows Tributary inflows Main gauge outflows Distributary outflows Recorded diversions Estimated local runoff Adjusted -39.6% Monthly streamflow (GL/mo) gauged accounted model Jun-90 Jun-91 Jun-92 Jun-93 Jun-94 Jun-95 Jun-96 Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Jun-03 Jun-04 Jun-05 Water balance Model (A) Accounts Difference Model efficiency Model (A) Accounts Jul 1990 Jun 2006 Monthly Gains GL/y GL/y GL/y Normal Main stem inflows Log-normalised - - Tributary inflows Ranked Local inflows Low flows only <0 <0 100 Unattributed gains and noise High flows only Losses GL/y GL/y GL/y Annual 10 Main stem outflows Normal Distributary outflows Log-normalised Net diversions Ranked River flux to groundwater 0-0 River and floodplain losses Definitions: 0.1 Unspecified losses low flows (flows<10% percentile ) : 1.2 GL/mo Unattributed losses and noise high flows (flows>90% percentile) : GL/mo 0.01 Change-uncertainty ratios P B Cwet Cmid Cdry Dwet Dmid Ddry Annual streamflow Monthly streamflow Reach gains and losses (GL/y) /91 91/92 92/93 93/94 Monthly streamflow (GL/mo). 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/ /04 04/05 05/06 Pecentage of months flow is exceeded unattributed gains ungauged gains gauged gains unattributed losses ungauged losses gauged losses Monthly Change-Uncertainty Ratio P B C D + wet O mid dry Annual Change-Uncertainty Ratio Annual streamflow (GL/y) /91 91/92 92/93 93/94 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/06 gauged A P Cwet Cmid Cdry Dwet Dmid Ddry 102 Water availability in the Barwon-Darling May 2008 CSIRO 2008

113 Downstream gauge Barwon Brewarrina Reach 4 Upstream gauge Barwon Danger Bridge Reach length (km) 249 Area (km 2 ) Outflow/inflow ratio 0.91 Net losing reach Land use ha % Dryland 2,206, Irrigable area - - Open water* - - River and wetlands 572, Open water* - - * averages for This is a slightly losing reach. Inflows are dominated by inflows from usptream, but tributary inflows are contributed by the Macquarie- Castlereagh region. Most of the inflows are gauged. Estimated local runoff does not explain ungauged gains. There are recorded diversions, and losses to wetlands are considerable. Baseline model performance is modest. Accounting explains monthly flows very well. Due to bias in the model, only the changes projected under pre-development and wet climate change scenarios appear stronger than model uncertainty. Gauging data Inflows Outflows Overall and gains and losses Fraction of total Gauged Attributed Fraction of variance Gauged Attributed Correlation with ungauged Gains Losses Linear adjustment normal ranked normal ranked Main gauge inflows Tributary inflows Main gauge outflows Distributary outflows Recorded diversions Estimated local runoff Adjusted % Monthly streamflow (GL/mo) gauged accounted model 0.1 Jun-90 Jun-91 Jun-92 Jun-93 Jun-94 Jun-95 Jun-96 Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Jun-03 Jun-04 Jun-05 Water balance Model (A) Accounts Difference Model efficiency Model (A) Accounts Jul 1990 Jun 2006 Monthly Gains GL/y GL/y GL/y Normal Main stem inflows Log-normalised Tributary inflows Ranked Local inflows Low flows only <0 <0 Unattributed gains and noise High flows only < Losses GL/y GL/y GL/y Annual 100 Main stem outflows Normal Distributary outflows Log-normalised Net diversions Ranked River flux to groundwater 0-0 River and floodplain losses Definitions: Unspecified losses low flows (flows<10% percentile ) : 1.0 GL/mo 1 Unattributed losses and noise high flows (flows>90% percentile) : GL/mo Change-uncertainty ratios P B Cwet Cmid Cdry Dwet Dmid Ddry Annual streamflow Monthly streamflow Reach gains and losses (GL/y) /91 91/92 92/93 93/94 Monthly streamflow (GL/mo). 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/ /05 05/06 Pecentage of months flow is exceeded unattributed gains ungauged gains gauged gains unattributed losses ungauged losses gauged losses Appendix C River system model uncertainty assessment by reach Monthly Change-Uncertainty Ratio P B C D + wet O mid dry Annual Change-Uncertainty Ratio Annual streamflow (GL/y) /91 91/92 92/93 93/94 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/06 gauged A P Cwet Cmid Cdry Dwet Dmid Ddry CSIRO 2008 May 2008 Water availability in the Barwon-Darling 103

114 Downstream gauge Darling Bourke Town Reach 5 Upstream gauge Barwon Brewarrina Reach length (km) 220 Area (km 2 ) 6255 Outflow/inflow ratio 1.54 Net gaining reach Land use ha % Dryland 489, Irrigable area - - Open water* - - River and wetlands 136, Open water* - - * averages for This is a gaining reach. Inflows are dominated by mains tem inflows. About three-quarters of inflows are gauged. Estimated local runoff explains little of the ungauged gains. There are some recorded diversions and river losses. Baseline model performance is good. Accounting also explains monthly flows well. The projected changes are all greater than the uncertainty in modelling. Appendix C River system model uncertainty assessment by reach Gauging data Inflows Outflows Overall and gains and losses Fraction of total Gauged Attributed Fraction of variance Gauged Attributed Correlation with ungauged Gains Losses Linear adjustment normal ranked normal ranked Main gauge inflows Tributary inflows Main gauge outflows Distributary outflows Recorded diversions Estimated local runoff Monthly streamflow (GL/mo) gauged accounted model 0.1 Jun-90 Jun-91 Jun-92 Jun-93 Jun-94 Jun-95 Jun-96 Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Jun-03 Jun-04 Jun-05 Water balance Model (A) Accounts Difference Model efficiency Model (A) Accounts Jul 1990 Jun 2006 Monthly Gains GL/y GL/y GL/y Normal Main stem inflows Log-normalised Tributary inflows Ranked Local inflows Low flows only <0 <0 Unattributed gains and noise High flows only Losses GL/y GL/y GL/y Annual 100 Main stem outflows Normal Distributary outflows Log-normalised Net diversions Ranked River flux to groundwater 0-0 River and floodplain losses Definitions: Unspecified losses low flows (flows<10% percentile ) : 0.9 GL/mo 1 Unattributed losses and noise high flows (flows>90% percentile) : GL/mo Change-uncertainty ratios P B Cwet Cmid Cdry Dwet Dmid Ddry Annual streamflow Monthly streamflow Reach gains and losses (GL/y) /91 91/92 92/93 93/94 Monthly streamflow (GL/mo). 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/ /05 05/06 Pecentage of months flow is exceeded unattributed gains ungauged gains gauged gains unattributed losses ungauged losses gauged losses Monthly Change-Uncertainty Ratio P B C D + wet O mid dry Annual Change-Uncertainty Ratio Annual streamflow (GL/y) /91 91/92 92/93 93/94 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/06 gauged A P Cwet Cmid Cdry Dwet Dmid Ddry 104 Water availability in the Barwon-Darling May 2008 CSIRO 2008

115 Downstream gauge Darling Louth Reach 6 Upstream gauge Darling Bourke Town Reach length (km) 453 Area (km 2 ) Outflow/inflow ratio 0.95 Net losing reach Land use ha % Dryland 2,200, Irrigable area - - Open water* - - River and wetlands 308, Open water* - - * averages for This is a slightly losing reach. Inflows are dominated by main stem inflows. Most of the inflows are gauged. Estimated local runoff does not explain the ungauged gains. There are diversions and losses to wetlands are expected. Baseline model performance was good. Accounting explains monthly flows very well. The projected changes are greater than model uncertainty. Gauging data Inflows Outflows Overall and gains and losses Fraction of total Gauged Attributed Fraction of variance Gauged Attributed Correlation with ungauged Gains Losses Linear adjustment normal ranked normal ranked Main gauge inflows Tributary inflows Main gauge outflows Distributary outflows Recorded diversions Estimated local runoff Adjusted % Monthly streamflow (GL/mo) gauged accounted model 0.01 Jun-90 Jun-91 Jun-92 Jun-93 Jun-94 Jun-95 Jun-96 Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Jun-03 Jun-04 Jun-05 Water balance Model (A) Accounts Difference Model efficiency Model (A) Accounts Jul 1990 Jun 2006 Monthly Gains GL/y GL/y GL/y Normal Main stem inflows Log-normalised Tributary inflows Ranked Local inflows Low flows only <0 <0 Unattributed gains and noise High flows only Losses GL/y GL/y GL/y Annual Main stem outflows Normal Distributary outflows Log-normalised Net diversions Ranked River flux to groundwater River and floodplain losses Definitions: Unspecified losses low flows (flows<10% percentile ) : 0.8 GL/mo Unattributed losses and noise high flows (flows>90% percentile) : GL/mo Change-uncertainty ratios P B Cwet Cmid Cdry Dwet Dmid Ddry Annual streamflow Monthly streamflow Reach gains and losses (GL/y) /93 93/94 94/95 95/96 Monthly streamflow (GL/mo). 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/ /05 05/06 Pecentage of months flow is exceeded unattributed gains ungauged gains gauged gains unattributed losses ungauged losses gauged losses Appendix C River system model uncertainty assessment by reach Monthly Change-Uncertainty Ratio P B C D + wet O mid dry Annual streamflow (GL/y) gauged A P Cwet Cmid Cdry Dwet Dmid Ddry Annual Change-Uncertainty Ratio 92/93 93/94 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/06 CSIRO 2008 May 2008 Water availability in the Barwon-Darling 105

116 Downstream gauge Darling Wilcannia Main Channel Reach 7 Upstream gauge Darling Louth Reach length (km) 334 Area (km 2 ) Outflow/inflow ratio 0.70 Net losing reach Land use ha % Dryland 6,600, Irrigable area - - Open water* - - River and wetlands 926, Open water* - - * averages for This is a slighty losing reach. About two-thirds of the inflows are gauged. Estimated local runoff did not explain ungauged gains. There are few recorded diversions. River and wetland losses are conisderable. Baseline model performance was reasonable. Accounting also explained monthly flows well. Only the projected changes under the pre-development and wet cimate change scenarios are greater than model uncertainty due to bias in the model. Appendix C River system model uncertainty assessment by reach Gauging data Inflows Outflows Overall and gains and losses Fraction of total Gauged Attributed Fraction of variance Gauged Attributed Correlation with ungauged Gains Losses Linear adjustment normal ranked normal ranked Main gauge inflows Tributary inflows Main gauge outflows Distributary outflows Recorded diversions Estimated local runoff Adjusted % Monthly streamflow (GL/mo) gauged accounted model 0.01 Jun-90 Jun-91 Jun-92 Jun-93 Jun-94 Jun-95 Jun-96 Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Jun-03 Jun-04 Jun-05 Water balance Model (A) Accounts Difference Model efficiency Model (A) Accounts Jul 1990 Jun 2006 Monthly Gains GL/y GL/y GL/y Normal Main stem inflows Log-normalised Tributary inflows Ranked Local inflows Low flows only <0 <0 Unattributed gains and noise High flows only < Losses GL/y GL/y GL/y Annual Main stem outflows Normal Distributary outflows Log-normalised Net diversions Ranked River flux to groundwater River and floodplain losses Definitions: Unspecified losses low flows (flows<10% percentile ) : 0.1 GL/mo Unattributed losses and noise high flows (flows>90% percentile) : GL/mo Change-uncertainty ratios P B Cwet Cmid Cdry Dwet Dmid Ddry Annual streamflow Monthly streamflow Reach gains and losses (GL/y) /93 93/94 94/95 95/96 Monthly streamflow (GL/mo). 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/ /05 05/06 Pecentage of months flow is exceeded unattributed gains ungauged gains gauged gains unattributed losses ungauged losses gauged losses Monthly Change-Uncertainty Ratio P B C D + wet O mid dry Annual streamflow (GL/y) gauged A P Cwet Cmid Cdry Dwet Dmid Ddry Annual Change-Uncertainty Ratio 92/93 93/94 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/ Water availability in the Barwon-Darling May 2008 CSIRO 2008