Water Availability in the Eastern Mount Lofty Ranges a Report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project

Water Availability in the Eastern Mount Lofty Ranges A report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project

October 2007 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 (2007). Water availability in the Eastern Mount Lofty Ranges. A report to the Australian Government from the CSIRO Murray- Darling Basin Sustainable Yields Project. CSIRO, Australia. 104 pp.

Publication Details

Published by CSIRO © 2007 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 1835-095X

Cover Photo courtesy of the Department of Water, Land and Biodiversity Conservation (South Australia)

Director’s Foreword

Following the November 2006 Summit on the Southern Murray-Darling Basin, the 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 surface- subsurface 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 2025. 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

Executive Summary

Background

The CSIRO Murray-Darling Sustainable Yields Project is providing governments with a robust, basin-wide estimate of water availability on an individual catchment and aquifer basis, taking into account climate change and other risks. This report describes the assessments undertaken for the Eastern Mount Lofty Ranges (EMLR) 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 Eastern Mount Lofty Ranges (EMLR) region in the far south west of the Murray-Darling Basin represents less than xctv Summar Executive one percent of the total area of the Basin and is based around the Marne, Bremer and Finniss Rivers. These and numerous other rivers and streams drain into the River Murray and into Lake Alexandrina. The regional population is about 52,000 or slightly less than three percent of the Basin total, including the towns of Murray Bridge, Mount Barker and Strathalbyn. The predominant land uses are dryland cropping and grazing. Irrigated cropping – for hay production, vines and horticulture – accounts for less than three percent of the total area and there are only small areas of commercial plantation forestry. Current surface water diversions for irrigation make up less than 0.1 percent of the Basin total, and groundwater use is less than two percent of the Basin total. y About 80 percent of surface water used in the region is from the Murray River and Lake Alexandrina, and is considered in the reporting for the Murray region. About 8000 farm dams in the region capture runoff for stock, domestic and irrigation use; water use from these dams has been modelled. Relatively small volumes of groundwater are extracted in the region for stock and domestic use and for irrigation. Wetlands of national significance in the region include the Tookayerta and Finniss Catchments, the Marne River Mouth and Ambersun – West Swamp. Some of the rivers of the region discharge into The Coorong and Lakes Alexandrina and Albert Ramsar wetland site.

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.

Historical climate and current development (Scenario A)

The average annual rainfall for the entire EMLR region is 463 mm and modelled average annual runoff is 30 mm. Rainfall is significantly higher in the winter half of the year and most of the runoff occurs in winter and early spring. The region generates about 0.5 percent of the total runoff of the Murray-Darling Basin, commensurate with its area. Average annual groundwater recharge is 64 GL/year.

Average annual surface water availability is 120 GL/year across the Marne, Bremer, Angas, Finniss, Tookayerta and Currency catchments. Current average net diversions from farm dams (including net evaporation) across these catchments are around 11 GL/year – or about 9 percent of the average available water. The level of use varies however, across the six catchments considered, from about 5 percent to 18 percent.

Current groundwater use is about 19 GL/year with over 80 percent of this occurring in the EMLR groundwater management unit (GMU) – the largest of the three GMUs in the region. Current extraction in the EMLR GMU is a quarter of rainfall recharge – this is a low level of use. Use is two-thirds of rainfall recharge in the Marne Saunders GMU but rainfall recharge here is only approximately 40 percent of the total recharge. As much as 60 percent of the recharge here is flood recharge, but this amount is poorly quantified. Extraction from the Angas-Bremer GMU is from the confined aquifer which is not directly recharged by rainfall. The main concern at current extraction levels is increasing groundwater salinity in the Angas-Bremer GMU and the Currency Creek region.

The total reduction in stream flow across the region due to current groundwater extraction is estimated to increase to 7 GL/year over the next 50 years; proportionally, these losses would be highest in Tookayerta Creek, with smaller impacts in the Angas, Bremer, Marne and Finniss rivers.

Environmental water planning has only recently commenced in the EMLR region and at this time there are no formal environmental water allocations or environmental flow rules for the region. However, the regionally important Fleurieu Swamps may have been affected by reductions in low flows in the Finniss River and Currency Creek caused by water resource development. The creeks and rivers of the region flow into the River Murray (an “Icon Site” under the Living Murray Initiative) and The Coorong and Lakes Alexandrina and Albert Ramsar wetland site. These environmental assets are analysed in the reporting for the Murray region.

Recent climate and current development (Scenario B)

Rainfall for the period 1997 to 2006 was 7 percent lower than the historical average, surface water availability was 26 percent lower and groundwater recharge was 24 percent lower. Long-term continuation of conditions similar to these would reduce average surface water availability by 28 GL/year and reduce average groundwater recharge by 15 GL/year. These changes would in turn reduce net diversions from farm dams by 6 percent, but due to the large reduction in water

y avilability would effectively increasing the level of surface water use to 11 percent. Similarly, under these climate conditions, groundwater use (at current levels) would represent a greater fraction of rainfall recharge: 8 percent greater for the EMLR GMU and 12 percent greater for the Marne Saunders GMU.

Future climate and current development (Scenario C)

Climate change is likely to reduce water availability by 2030. Under the best estimate (median) 2030 climate water availability would be reduced by about 18 percent, that is, about 22 GL/year less water available. This would reduce net

Executive Summar diversions from farm dams by 5 percent, but due to the large reduction in water avilability would effectively increase the level of surface water use to 10 percent. This 2030 climate would have very little impact on the fraction of rainfall recharge that is extracted from groundwater for use. However, the frequency with which flow ceases at end-of-system locations would increase for the ephemeral Angas, Marne and Bremer Rivers.

The results from nearly all 15 global climate models used indicate a decrease in runoff by 2030. However, there is considerable uncertainty in the climate predictions for 2030 (different climate models and different global warming scenarios), meaning reductions in average water availability could range from 3 percent to 52 percent. Thus in the future ‘worst case scenario’ water availability would be less than half the current average level and the level of surface water use would be 15 percent of this. The extreme estimates come from a high global warming scenario; the range from a low global warming scenario is only one third as wide. For groundwater, this uncertainty range between the dry and wet future climate extremes means extraction could be between 22 and 39 percent of rainfall recharge for the EMLR GMU (best estimate 26 percent) and between 59 and 101 percent for the Marne Saunders GMU (best estimate 69 percent).

Future climate and future development (Scenario D)

Future development could further reduce water availability. The combined impact of a potential 16 percent increase in total farm dam volume (an additional 3400 ML) and of a potential doubling of the area of commercial plantation forestry (an additional 2000 ha), both of which could occur by 2030, would be a 3 percent reduction in water availability in addition to the climate change impacts. The best estimate for combined reduction in water availability from climate change and future development would be 21 percent, with climate uncertainty indicating a range from 6 percent to 54 percent less water.

These changes would actually increase net diversions considerably, as more farm dams provide more opportunity for water use. Future forestry growth would only be likely in the wetter Tookayerta and Currency catchments, for which the modelled reductions in runoff are 17 and 7 percent respectively. The level of surface water use with these developments for the best estimate 2030 climate would be 12 percent.

Without intervention, groundwater extraction from all GMUs is likely to increase in the future. For the best estimates of future extraction (a 60 percent increase) and 2030 climate, extraction would be over 40 percent of rainfall recharge for the EMLR GMU and would exceed rainfall recharge for the Marne Saunders GMU. This increase in extraction would reduce streamflow by an about additional 4 GL/year – a total average impact of about 11 GL/year; the largest proportional impact would be on Tookayerta Creek. Current groundwater allocation planning may however, prevent these levels of extraction being realised.

Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

The flow changes expected due to any future commercial plantation forestry development in the Tookayerta and Currency catchments and as a result of climate change may further affect the ecology of the regionally important Fleurieu Swamps. However detailed assessments of wetland hydrology have not been made.

Uncertainty

The runoff modelling for the EMLR region is comparatively good because there are four calibration catchments in the higher runoff producing areas in the western and south-western parts of the region from which to estimate the model parameter values. The river model is conceptually sound, reproducing monthly and inter-annual flow patterns well. The model is considered suitable for assessing relative changes in average flows. However, the model overestimates total volumes by 14 percent and 21 percent respectively in the two rivers independently evaluated. Projections of absolute current and longer-term flow volumes are thus uncertain. This may have implications for the use of these results to assess local in-stream environmental values and availability of water for users. Summar Executive

A simple water balance approach has been used for groundwater modelling in the EMLR. This is appropriate given the low priority rating of the GMUs in this region in the context of the overall project. However, this approach would be inappropriate for addressing local groundwater management issues. The importance of flood recharge in the Angas Bremer and Marne Saunders GMUs implies that a more complex analysis may be warranted to evaluate the implications of variations in flood recharge under each of the climate change scenarios. The estimated impacts of groundwater extraction on streamflow are considered to have a low level of confidence but this process exerts a relatively low

pressure on water availability. y

The largest sources of uncertainty for Scenario C results are the climate change projections (global warming level) and the modelled implications of global warming on local rainfall. A wide range of the best available climate modelling was used but there is considerable scope for improvement in those global models at predicting regional rainfall. The future scenarios include very large reductions to water availability suggesting that improvements in the ability to predict the hydrological consequences of climate change would have substantial benefits for water management.

There are considerable uncertainties associated with the future development projections. There are multiple drivers for commercial plantation forestry and farm dams, many of which have not been considered, and the ways in which land holders will respond to the current policies is uncertain. Future development could be very different should governments impose different policy controls on these activities. Future groundwater use has been assessed by considering changes in demand, and hence these are simply best guesses. Current groundwater allocation planning in EMLR is likely to impose new limits on groundwater extraction, such that the levels of future extraction considered herein may never actually be reached.

1 Introduction ...... 1 1.1 Background ...... 1 1.2 Project methodological framework ...... 3 1.3 Climate and development scenarios ...... 4 1.4 Rainfall-runoff modelling ...... 6 1.5 River system modelling...... 7 1.6 Monthly water accounts ...... 9 1.7 Groundwater modelling...... 11 1.8 Environmental assessment...... 11 1.9 References...... 12 2 Overview of the region...... 14 2.1 The region ...... 14 2.2 Environmental description...... 17 2.3 Surface water resources ...... 18 2.4 Groundwater ...... 20 2.5 References...... 24 3 Rainfall-runoff modelling ...... 25 3.1 Summary...... 25 3.2 Modelling approach...... 26 3.3 Modelling results ...... 28 3.4 Discussion of key findings...... 34 3.5 References...... 35 4 River system modelling ...... 36 4.1 Summary...... 36 4.2 Modelling approach...... 38 4.3 Modelling results ...... 42 4.4 Discussion of key findings...... 59 4.5 References...... 63 5 Uncertainty in surface water modelling results ...... 65 5.1 Summary...... 65 5.2 Approach...... 66 5.3 Results ...... 70 5.4 Discussion of key findings...... 74 5.5 References...... 75 6 Groundwater assessment...... 77 6.1 Summary...... 77 6.2 Groundwater management units in EMLR...... 78 6.3 Hydrogeological context...... 80 6.4 Trends in groundwater levels and salinity...... 80 6.5 Surface water-groundwater connectivity...... 82 6.6 Water balances for lower priority GMUs ...... 83 6.7 Discussion...... 89 6.8 References...... 90 7 Environment...... 91 7.1 Summary...... 91 Appendix A Rainfall-runoff results for subcatchments ...... 93 Appendix B River water modelling reach mass balances ...... 95 Appendix C River system model uncertainty assessment by reach...... 101

Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007 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 Eastern Mount Lofty Ranges region ...... 15 Table 2-2. Wetlands of national significance located within or adjacent to the Eastern Mount Lofty Ranges region ...... 18 Table 2-3. Summary of farm dams in the Eastern Mount Lofty Ranges region...... 19 Table 2-4. Categorisation of GMUs, including annual extraction, entitlement and recharge details ...... 21 Table 2-5. Summary of groundwater management plans...... 23 Table 2-6. Historic installation of bores in the EMLR GMU...... 24 Table 3-1. Summary results from the 45 Scenario C simulations (numbers show percentage change in mean annual rainfall and runoff under Scenario C relative to Scenario A) ...... 31 Table 3-2. Water balance over the entire region by scenario ...... 33 Table 4-1. Model setup information ...... 41 Table 4-2. Rainfall, evaporation and flow factors for model robustness test ...... 42 Table 4-3. River system model average annual water balance under scenarios P, A, B, C and D...... 43 Table 4-4. Mean and median flows (ML/day) under Scenario A and for other scenarios relative to Scenario A...... 44 Table 4-5. Change in annual water availability under scenarios B, C and D relative to Scenario A...... 44 Table 4-6. Details of combined farm dam behaviour ...... 46 Table 4-7. Change in net diversions in each sub-catchment relative to Scenario A...... 47 Table 4-8. Relative level of use under scenarios A, B, C and D...... 49 Table 4-9. Indicators of net use including net evaporation during dry periods under scenarios A, B, C and D...... 49 Table 4-10. Summary of average reliability under scenarios A, B, C and D...... 51 Table 4-11. Daily flow event frequency and magnitude for scenarios P, A, B, C and D...... 58 Table 4-12. Percent of time stream continues to flow under scenarios A, B, C and D ...... 58 Table 4-13. Relative level of available water not diverted for use under scenarios A, B, C and D...... 59 Table 4-14. End-of-catchment outflows for WaterCRESS and SIMHYD under each scenario ...... 61 Table 4-15. Percentage reduction in end-of-catchment outflow for WaterCRESS and SIMHYD under each scenario ...... 62 Table 4-16. Percentage change in runoff due to commercial plantation forestry in WaterCRESS and SIMHYD under best estimate 2030 climate scenario ...... 62 Table 5-1. Possible framework for considering implications of assessed uncertainties ...... 66 Table 5-2. Some characteristics of the gauging network of the Eastern Mount Lofty Ranges region (4693 km2) compared with the entire Murray-Darling Basin (1,062,443 km2)...... 70 Table 5-3. Details of calibration and validation periods, number of years between 1895–2006 (112 years) with annual rainfall less than the driest and more than the wettest year in the calibration period, respectively, and prior assessment of the models performance...... 72 Table 5-4. Regional water balance modelled and estimated on the basis of water accounting ...... 73 Table 6-1. Description of GMUs, including annual extraction, entitlement and recharge details...... 79 Table 6-2. Subcatchment partitioning of streamflow into run-off and baseflow for selected catchments ...... 82 Table 6-3. Estimated rainfall recharge for subcatchments within the EMLR region ...... 84 Table 6-4. Summary results from the 45 Scenario C simulations. Numbers show percentage change in mean annual rainfall and recharge under Scenario C relative to Scenario A. Those in bold type have been selected for further modelling...... 86 Table 6-5. Summary results of the scenarios for modelling for each GMU in the EMLR region. Numbers show percentage change in mean annual recharge under scenario relative to Scenario A...... 86 Table 6-6. Scaled recharge under scenarios A, B and C ...... 87 Table 6-7. Comparison of current and projected future groundwater extractions with scaled rainfall recharge ...... 87 Table 6-8. Surface water-groundwater connectivity showing an estimate of the volumetric impact extraction has on stream flow..89

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 2-1. 1895–2006 annual and monthly rainfall averaged over the reporting region. The curve on the annual graph shows the low frequency variability...... 15 Figure 2-2. Map of dominant land uses of the Eastern Mount Lofty Ranges region with inset showing the region’s location within the Murray-Darling Basin ...... 17 Figure 2-3. History of groundwater extractions including total diversion limits for the Angas Bremer GMU. Source: RMCWMB (2001), MDBC (2007) and DWLBC (2004) ...... 20 Figure 2-4. Map of groundwater management units within the Eastern Mount Lofty Ranges region ...... 22 Figure 3-1. Map of modelling subcatchments and calibration catchments ...... 27 Figure 3-2. Modelled and observed monthly runoff and daily flow duration curves for the calibration catchments...... 28 Figure 3-3. Spatial distribution of mean annual rainfall and modelled runoff averaged over 1895–2006...... 29 Figure 3-4. 1895–2006 annual rainfall and modelled runoff series averaged over the region. The curve shows the low frequency variability...... 29 Figure 3-5. Mean monthly rainfall and modelled runoff (averaged over 1895–2006 for the region)...... 30 Figure 3-6. Percentage change in mean annual runoff from the 45 Scenario C simulations (15 GCMs and three global warming scenarios) relative to Scenario A runoff...... 31 Figure 3-7. Mean annual rainfall and modelled runoff under scenarios A, Cdry, Cmid and Cwet ...... 32 Figure 3-8. Mean monthly rainfall and modelled runoff under scenarios A, C and D averaged over 1895-2006 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)...... 34 Figure 3-9. Daily flow duration curves under scenarios A, C and D 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)...... 34 Figure 4-1. River system map showing rivers, gauges and model sub-catchments...... 40 Figure 4-2. Annual water availability under Scenario A ...... 45 Figure 4-3. Annual water availability relate to Scenario A under (a) Scenarios B, (b) Scenario C and (c) Scenario D ...... 45 Figure 4-4. Behaviour of the water storages over the maximum days between spills under (a) scenarios A and B, (b) scenarios A and C, (c) scenarios A and D...... 46 Figure 4-5. Total average annual diversions for sub-catchments under (a) scenarios A and C and (b) scenarios A and D...... 47 Figure 4-6. Total diversions for (a) Scenario A and difference between total water use for Scenario A and (b) Scenario B (c) Scenario Cwet; (d) Scenario Cmid; (e) Scenario Cdry; (f) Scenario Dwet; (g) Scenario Dmid; and (h) Scenario Ddry ...... 48 Figure 4-7. Comparison of allocated and used water under (a) Scenario A and Scenario C, and (b) Scenario A and Scenario D ..50 Figure 4-8. Diversion reliability under (a) scenarios A and B; (b) Scenario C and (c) Scenario D ...... 50 Figure 4-9. Daily flow duration curves for modelled subcatchments under scenarios P, A and B...... 52 Figure 4-10. Daily flow duration curves for modelled subcatchments under scenarios P, A and C ...... 53 Figure 4-11. Daily flow duration curves for modelled sub-catchments under scenarios P, A and D ...... 54 Figure 4-12. Seasonal flow duration curves under scenarios P, A and B at (a) Marne River; (b) Bremer River, (c) Angus River, (d) Finniss River, (e) Tookayerta Creek, and (f) Currency Creek ...... 55 Figure 4-13. Seasonal flow duration curves under scenarios P, A and C at (a) Marne River; (b) Bremer River, (c) Angus River, (d) Finniss River, (e) Tookayerta Creek, and (f) Currency Creek ...... 56 Figure 4-14. Seasonal flow duration curves under scenarios P, A and D at (a) Marne River; (b) Bremer River, (c) Angus River, (d) Finniss River, (e) Tookayerta Creek, and (f) Currency Creek ...... 57 Figure 4-15. Comparison of diverted and non-diverted shares of water under scenarios A, B, C and D ...... 59 Figure 4-16. Comparison between the mean annual flows for each catchment under scenarios A, B, C and D ...... 63 Figure 5-1. Map showing the subcatchments used in modelling, with the reaches for which river water accounts were developed (‘accounting reach’) and contributing head water catchments with gauged inflows (‘contributing catchment’). Black dots and red lines are nodes and links in the river model respectively...... 68 Figure 5-2. Map showing the rainfall, streamflow and evaporation observation network, along with the subcatchments used in modelling...... 71 Figure 6-1. Map of groundwater management units within the Eastern Mount Lofty Ranges region ...... 79 Figure 6-2. Hydrograph for bore MOR220 showing a declining trend in groundwater level with the lowest point in 2004 associated with below average rainfall...... 80 Figure 6-3. Hydrograph for bore MCF009 showing a declining trend in groundwater level throughout the 1990s with a rising trend after 2000...... 81 Figure 6-4. Hydrograph for bore NGK006 displaying neither a rising nor falling trend ...... 81 Figure 6-5. Groundwater levels and salinity variations for the Langhorne Creek town water supply over the last 15 years...... 82 Figure 6-6. Percentage change in mean annual recharge from the 45 Scenario C simulations relative to Scenario A recharge.....85

Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

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 resources 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 Prime Minister of Australia met with the First Ministers of Victoria, New South Wales, South 1 Australia and Queensland at a water summit focussed primarily on the future of the Murray-Darling Basin (MDB). As an Introduction 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

• 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; and • compare the estimated current and future water availability to that required to meet the current levels of extractive use.

The MDB Sustainable Yields Project is reporting progressively on each of 18 contiguous regions that comprise the entire Murray-Darling Basin. 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 south-westerly 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).

1 Introduction 1 Introduction

Figure 1-1. Region by region map of the Murray-Darling Basin

The MDB Sustainable Yields Project will be the most comprehensive Basin-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 Basin, • the hydrologic subcatchments required for detailed modelling have been precisely defined across the entire Basin, • 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),

2 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

• 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, and • 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 cross- check on the performance of river system models).

The successful completion of these outcomes, among many others, relies heavily on a focussed collaborative and team- oriented 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 Murray-Darling Basin to the project. 1

1.2 Project methodological framework Introduct

The methodological framework for the project is shown in the diagram below (Figure 1-2). This also indicates in which chapters of this report the different aspects of the project assessments and results are presented. ion

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.

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 1895–2006 climate:

1. historical climate and current development

2. recent climate and current development

1 Introduction 1 Introduction 3. future climate and current development

4. future climate and future development.

These scenarios are described in some detail below with full details provided in Chiew et al. (2007a).

1.3.1 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 0.05o x 0.05o (5 km x 5 km) grid across the continent (Jeffrey et al., 2001; and www.nrm.qld.gov.au/silo). Areal potential evapotranspiration (PET) data are calculated from the SILO climate surface using Morton’s wet environment evapotranspiration algorithms (www.bom.gov.au/climate/averages; 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 license 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 license entitlements. Surface–groundwater exchanges in the river and groundwater models represent an equilibrium condition for the above levels of surface and groundwater development.

1.3.2 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.

1.3.3 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 2030. Three global warming scenarios are analysed in 15 global climate models (GCM) to

4 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

provide a spectrum of 45 climate variants for the 2030. 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 Murray-Darling Basin. 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 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 2030.

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 1 changes in the daily rainfall distribution is important because many GCMs indicate that extreme rainfall in an enhanced Introduction 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.

1.3.4 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.

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 licenses 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 Victoria DSE, 2004) and current policy controls (Queensland Water Act, 2000; New South Wales Water Management Act, 2000; Victoria Water Act, 1989; South Australia Natural Resources Management Act, 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 north-eastern 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, and • 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 0.05o grids (~ 5 km x 5 km) across the entire MDB for the four scenarios.

The model is calibrated against 1975 to 2006 streamflow data from about 200 unregulated catchments of 50 km2 to 2000 km2 across the MDB (calibration catchments). Although unregulated, streamflow in these catchments for the calibration period may reflect 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 five percent of the total recorded runoff. The resulting optimised model parameters are therefore identical for all cells within a calibration catchment.

The runoff for non-calibration catchments is modelled using optimised parameter values from the geographically closest

1 Introduction 1 Introduction 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 point the parameter values are used. The default parameter values are taken from the whole-of-Basin modelling run (identical parameters across the entire Basin 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 km2 to 2000 km2 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 north-western 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 good 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 CO2 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. (2007b).

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., 2007b).

For the 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) and www.toolkit.net.au/fcfc.

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 Murray-Darling Basin 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. (2007b) 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 Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

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 1

river system models (separately for each season). The Scenarios B, C and D inflow series for the river system modelling Introduction 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

IQQM Integrated Quantity-Quality Model: hydrologic modelling tool Paroo, Warrego, Condamine-Balonne (Upper, Mid, developed by the NSW Government for use in planning and Lower), Nebine, Moonie, Border Rivers, Gwydir, Peel, evaluating water resource management policies. Namoi, Castlereagh, Macquarie, Marthaguy, Bogan, Lachlan, Murrumbidgee, Barwon-Darling

REALM Resource Allocation Model: water supply system simulation Ovens (Upper, Lower), Goulburn, Wimmera, Avoca, tool package for modelling water supply systems configured ACT water supply. as a network of nodes and carriers representing reservoirs, demand centres, waterways, pipes, etc.

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 Eastern Mt Lofty Ranges (six separate catchments) 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.

SMHS Snowy Mountains Hydro-electric Scheme model: purpose Snowy Mountains Hydro-electric Scheme built by Snowy Hydro Ltd to guide the planning and operation of the SMHS.

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 Basin 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

• 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.

1.5.1 Surface-groundwater interactions

The project is explicitly considering and quantifying the water exchanges between rivers and groundwater systems. The approaches used are described below.

The river models used by State agencies have in turn 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 1 Introduction 1 Introduction river system. That is, the calibration period is often one of changing hydrologic relationships, a period where 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

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,

8 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

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 1

about the performance and assumptions in the river model, as for example associated with recent water Introduction 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.

Despite these advantages, it is emphasised that the water accounting methodology invokes models and indirect estimates of water balance components where direct measurements are not available. Because of this, these water accounts are not an absolute point of truth. Rather, 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 our findings. The methods for water accounting are based on existing methods and those used by Kirby et al. (2006) and Van Dijk et al. (2007) and are described in detail in Kirby et al. (2007).

1.6.1 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., 2007). The two methods are constrained using direct on-ground AET measurements at seven study sites and catchment stream flow observations from more than 200 catchments across Australia. Both methods provide AET estimates at 1 km resolution.

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 2000/2001 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, and • LANDSAT satellite imagery for the years 1998 to 2004.

The reach-by-reach estimates of net water use from irrigation areas and from floodplains and wetlands are subject to the following limitations:

• 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

1 Introduction 1 Introduction 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 are used internally to conceptual water balance models of wetland and irrigation water use that include a simulated storage as considered appropriate based on ancillary information.

1.6.2 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 comparison of apparent ungauged gains and losses and a comparison with a time series of estimated water balance components.

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.

10 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

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 will be described in greater detail in Kirby et al. (2007) 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 1 Introduction 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.

1.7 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 Basin. A range of methods are used appropriate to the size and importance of different GMUs. There are over 100 GMUs in the Murray-Darling Basin, 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. 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.

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 commence- to-fill levels) which relate to ecological responses such as bird breeding events.

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 Murray- Darling Basin. 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 Basin 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 1 Introduction 1 Introduction 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 2022. Available at: www.abs.gov.au. 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, 831–850. Brown AE, Podger PM, 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 et al. (2007a) 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 et al. (2007b) 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. Paper prepared for 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, 1–11. Chiew FHS, Peel MC and Western AW (2002) Application and testing of the simple rainfall-runoff model SIMHYD. In: Singh VP and Frevert DK (Ed.s), Mathematical Models of Small Watershed Hydrology and Application. Littleton, Colorado, pp335–367. DSE (2004) Victoria in Future 2004 – Population projections. Department of Sustainability and Environment, Victoria. Available at: www.dse.vic.gov.au. Environment Australia (2001) A Directory of Important Wetlands in Australia. Available at: http://www.environment.gov.au/water/publications/environmental/wetlands/pubs/directory.pdf 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, 309–330.

12 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

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, 16-20 October 2006. Jakarta: UNESCO. 67-77 . Kirby J et al. (2007) Uncertainty assessments for scenario modelling. 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, 199–212. Nash JE and Sutcliffe JV (1970) River flow forecasting through conceptual models 1: A discussion of principles. Journal of Hydrology 10, 282–290. 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, 1–12. New South Wales Government (2000) Water Management Act 2000 No 92. Queensland Government (2000) Water Act 2000. South Australia Government (2004) Natural Resources Management Act 2004. 1 Tan KS, Chiew FHS, Grayson RB, Scanlon PJ and Siriwardena L (2005) Calibration of a daily rainfall-runoff model to estimate high daily flows. Paper prepared for the Congress on Modelling and Simulation (MODSIM 2005), December 2005. Melbourne, Australia. Introduction pp2960–2966. Van Dijk A et al. (2007) Reach-level water accounting for 1990-2006 across the Murray-Darling Basin. A report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project. CSIRO, Australia. In prep. Victoria Government (1989) Water Act 1989, Act Number 80/1989.

2 Overview of the region

The Eastern Mount Lofty Ranges region is based around the Marne, Bremer and Finniss Rivers and includes the towns of Murray Bridge, Mount Barker and Strathalbyn. The region covers less than one percent of the Murray-Darling Basin. It has slightly less than three percent of the MDB’s population, uses less than 0.1 percent of the surface water diverted for irrigation and uses less than two percent of the Basin’s groundwater resource. Around 60 percent of the total surface and groundwater used within the region is sourced from the River Murray system which is outside of the reporting region.

There are three wetlands of national significance within the region and an adjacent internationally listed Ramsar site The Coorong and Lakes Alexandrina and Albert. There is no direct drainage to The Coorong from the Eastern Mount Lofty Ranges, although numerous rivers and streams drain from the ranges into the Murray River, the EMLR-confined wetlands, and the adjacent lakes Alexandrina and Albert.

The predominant land use within the region is dryland cropping and grazing. The majority of the water diverted for irrigation use in the region is extracted from the Murray River and Lake Alexandrina. Irrigated cropping makes up less than three percent of the total land area. Pasture and lucerne hay production and vines fruits and other horticulture are the major irrigated land uses. There are a large number of (approximately 8000) farm dams in the region which capture runoff in the hills for stock and domestic use and irrigation. Relatively small volumes of groundwater are extracted for irrigation use on the plains. There are only small areas of commercial plantation forestry.

The following sections summarise 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

2 Overview of the region presents key features of the surface and groundwater resources of the region including historic water use.

2.1 The region Located at the lower end of the Murray-Darling River system in South Australia, the Eastern Mount Lofty Ranges (EMLR) region covers 4693 km2 or 0.4 percent of the MDB. It is bounded to the east by the River Murray and forms the lower western edge of the MDB. The major water sources within the EMLR region are the Marne, Bremer, and Finniss Rivers, alluvial aquifers and private on-farm water storages. The region also uses surface water diverted from the Murray River and Lake Alexandrina however the impacts of these diversions on the Murray River system are not considered as part of the analysis for this region, nor are the environmental assets adjacent to or within the Murray River channel. For information on these adjacent regional impacts, refer to ‘Water Availability in the Murray’ (CSIRO, 2007).

The eastern slopes of the Mount Lofty Ranges are steep and fall away to the broad Murray Plains and eventually to the Murray River and the lower lakes of Lake Alexandrina and Lake Albert. Mean annual rainfall is 463 mm varying from 900 mm in the west and south-west to 300 mm in the north. Most of the rainfall and runoff occur in winter and early spring. The region’s average annual rainfall has remained relatively consistent over the past 111 years and has a reasonably low coefficient of variation of 0.21. The mean annual rainfall over 1997 to 2006 of 431 mm has been 7 percent lower than the long-term mean.

14 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

800 80

600 60

400 40

200 20 Annual rainfall (mm) rainfall Annual

0 Mean monthly rainfall (mm) 0 1895 1915 1935 1955 1975 1995 JFMAMJJASOND

Figure 2-1. 1895–2006 annual and monthly rainfall averaged over the reporting region. The curve on the annual graph shows the low 2 frequency variability. Overview

The EMLR region contributes about 0.5 percent of the total annual runoff in the MDB. The mean annual modelled runoff averaged over the region for the 111 year period is 30 mm and is significantly higher in the winter and early spring period. of The mean annual modelled runoff for the ten year period 1997 to 2006 is around 36 percent lower than the long-term the mean. The runoff estimates for the region are considered to be relatively good due to streamflow gauging in four upper catchments. Most of these stations were established in the late 1960s for monitoring flows from the high yielding areas of region the EMLR.

The regional population is approximately 52,000 which is 2.6 percent of the MDB. The major towns are Murray Bridge, Mount Barker and Strathalbyn. The predominant land use within the region is dryland grazing and cropping. Irrigated cropping makes up less than three percent of the total land use. Pasture and lucerne hay production and vine fruits and other horticulture are the major irrigated land uses.

Table 2-1. Summary of land use in the year 2000 within the Eastern Mount Lofty Ranges region

Land use Area Area percent ha Dryland crops 27.2% 127,700 Dryland pasture 66.6% 311,100 Irrigated crops 2.8% 13,200 Cereals 2.9% 400 Horticulture 10.5% 1,400 Orchards 1.5% 200 Pasture and hay 40.2% 5,300 Vine fruits 44.9% 5,900 Native vegetation 1.0% 4,500 Plantation Forests 0.5% 2,100 Urban 1.1% 5,000 Water 0.8% 3,800 Total 100.0% 467,400 Source: BRS (2000)

Water and land resources are managed under the South Australia Natural Resource Management Act (2004). The Act provides the framework for the integrated use and management of the state’s natural resources.

A Natural Resource Management Board is responsible for providing an integrated response to water, soil, biodiversity, pest plant and animal control. A key function of the Natural Resource Management (NRM) Board is the development of plans that will assist in the development of better management and conservation of the region’s natural resources. Two types of plans are prepared, a Natural Resource Management Plan (NRM Plan) and Water Allocation Plans (WAPs).

The NRM Plan is a ten-year strategic plan and includes a three-year business plan detailing the resources needed to implement the plan. It is required to include:

• information on (i) the natural resources and their state and condition; (ii) environmental, social, economic and practical considerations relating to their use, management, conservation, protection, improvement and where relevant their rehabilitation; and, (iii) management of pest species of animals and plants; and • information about the issues surrounding the management of natural resources at the regional and local level and specifically about methods for improvement of natural resources and their conservation, use or management, actions plans for proper storm water management and flood mitigation, arrangements for management of wetlands, estuaries and marine resources (SAMDBNRMB, 2006).

The NRM Board has developed a plan involving ten program areas. These programs include:

• Salinity – the South Australian River Murray Salinity Strategy commits the Government to support the development and implementation of Land and Water Management Plans to meet salinity management goals. • Flow management – development of water allocation policy that considers the water users of the region and the needs of the environment, including provision for, and management of, environmental water requirements for

2 Overview of the region the Eastern Mount Lofty Ranges. • Water quality – current works to control erosion, and minimise sediment and nutrient input to watercourses needs to continue. • Water Use – Improving water use efficiency is considered a key issue for the South Australia MDB region.

16 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

2 Overview of the region

Figure 2-2. Map of dominant land uses of the Eastern Mount Lofty Ranges region with inset showing the region’s location within the Murray-Darling Basin

2.2 Environmental description There are 14 catchments in the Eastern Mount Lofty Ranges which extend from Currency Creek in the south to the Marne River in the north. The highly variable rainfall and the presence of groundwater that occasionally resurfaces as springs near the mouths of the tributaries creates a variety of important habitats for native plants and animals within these catchments. The eastern outwash slopes of the Mount Lofty Ranges, along the western boundary of the region, are composed mainly of sandy loam soils. The northern, eastern and southern areas of the region are dune-swale plains.

There are a number of wetlands within the region listed as either having national or regional importance (Table 2-2).

Table 2-2. Wetlands of national significance located within or adjacent to the Eastern Mount Lofty Ranges region

SITE CODE DIWA name DIWA area RAMSAR sites ha SA034 Tookayerta & Finniss Catchments 236.40 SA045 Marne River Mouth 18.84 SA063 The Coorong, Lake Alexandrina & Lake Albert* 139317.80 The Coorong, and Lakes Alexandrina and Albert Wetland*

SA084 Ambersun - West Swamp 5.68 * RAMSAR site (area 139318 ha) is adjacent to the EMLR region.

The Coorong and Lakes Alexandrina and Albert Wetland is a Ramsar listed wetland of international importance adjacent to the EMLR region but supplied with water from Lake Alexandrina and Lake Albert. Collectively this general area is known as the Coorong – Lower Lakes region of South Australia. Lake Alexandrina receives flows from the Eastern Mount Lofty Range region’s tributaries and from the Murray River which also receives flows from EMLR streams. The lower parts of the Finniss River and Tookayerta and Currency creek catchments are listed as part of the Coorong, Lake Alexandrina and Lake Albert Wetland of International Importance. The lower Murray River and the lakes are listed as one of the six Icon sites under the Living Murray Initiative.

The Fleurieu Swamps are located in the Currency Creek, Tookayerta Creek and Finniss River catchments. These swamps are considered a critically endangered ecological community under the Environment Protection and Biodiversity Conservation (EPBC) Act. Several water dependant species found in the EMLR region are also listed under the EPBC

2 Overview of the region Act including Yarra Pigmy Perch, Murray Hardyhead and the Mount Lofty Ranges Southern Emu Wren which is dependant on the Fleurieu swamps. A range of water dependant plants and animals found in the EMLR region are listed as rare, vulnerable, endangered or critically endangered under the South Australia National Parks and Wildlife threatened species schedules.

2.3 Surface water resources

2.3.1 Rivers and storages

The major tributaries include the Angas, Bremer, Marne and Finniss rivers as well as Tookayerta and Currency creeks. The streams of the Eastern Mount Lofty Ranges gain water from catchment runoff in the hills. However, on the plains the streams lose water to groundwater. Around 80 percent of the stream flow is generated from the hills subcatchments of the Angas River, Bremer River, Finniss River and the Tookayerta Creek catchments.

Most of the EMLR streams have been ephemeral with the exception of Tookayerta Creek and the Finniss River. (RMCWMB, 2003). There are approximately 8000 farm dams in the EMLR with an estimated storage capacity of about 22 GL. Data on irrigation dams is currently being compiled as part of the Land and Water Use surveys being undertaken within the water allocation planning process detailed below. The number and capacity of dams used for stock and domestic purposes are not being surveyed and is considered an estimate. On the plains, surface water is pumped primarily from the Murray River and Lake Alexandrina for irrigation.

Farm dam construction and direct pumping from watercourses in the EMLR have significantly changed the stream flow patterns, particularly during the drier months. The water resources of the EMLR have recently been prescribed and the region is now required to develop water allocation policy that considers the water users of the region and the needs of the environment.

2.3.2 Surface water management institutional arrangements

Water resources in South Australia are managed under the Natural Resources Management Act 2004. This Act requires the South Australian Murray-Darling Basin NRM Board to prepare a water allocation plan (WAP) for each of the prescribed water resources in its area.

18 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

A WAP is a statutory instrument that is used for various purposes in the administration of the Act; in particular, to guide the granting of licences to take water, the transfer of a licence and/or water allocation, and long-term sustainable management of water resources. Implementation of the WAP for the Murray River must support the State in meeting its obligations under the MDB Ministerial Council Cap on diversions, and the MDBC Salinity and Drainage Strategy.

The River Murray Water Allocation Plan covers the prescribed watercourse of the Murray River from the Victorian border to the edge of Lake Alexandrina and Lake Albert and portions of Currency Creek, the Finniss River, Angas River and Bremer River. WAPs are currently being developed for the water resources of the Marne River and Saunders Creek, and the Eastern Mount Lofty Ranges.

Due to pressures on available water of the Mount Lofty Ranges, the Government of South Australia ‘prescribed’ the region’s water resources in September 2005. The prescription requires that all persons wishing to extract surface or groundwater require a licence to do so. The purpose is to ensure that the resources are well managed to meet the present and future water needs of the region, the surface water, watercourse and groundwater resources of the EMLR. 2

The NRM Board is now preparing a WAP for the surface and groundwater resources of the Marne River – Saunders Overview Creek, and the Eastern Mount Lofty Ranges Prescribed Water Resources Areas. As the Angas Bremer Prescribed Water Area now falls within the boundaries of the EMLR prescribed watercourse and surface water area, a single WAP will be prepared to cover both areas. The combination of these areas will be referred to as the Eastern Mount Lofty Ranges

Prescribed Water Resources Area. of

As no plan currently exists for the EMLR Prescribed Water Resources Area, the presiding water management plan is the the

River Murray Catchment Water Management Plan (2003). It is noted that this is not a water-sharing plan as such, but region has scope to place limits on the future development of the surface water resource in relation to farm dams.

The main mechanism is to limit farm dam development volume to 30 percent of the mean winter (May to November) runoff. Assessment of these limits at a subcatchment scale was underway as of September 2007, as part of the water allocation process.

2.3.3 Water products and use

Farm dams provide the primary source of surface water for stock and domestic use and for irrigation in the upper part of the region. Farm dams are broadly grouped into two categories:

• Stock and domestic dams, which are not required to be licensed under the proposed Water Allocation Plan and • Irrigation dams, which are required to be licensed.

The total net diversions for consumptive use of surface water sourced from within the EMLR region are defined as the water extracted from farm dams. The net diversions include the total amount of water extracted from the farm dams for consumptive, as well as stock and domestic use plus the total net evaporative loss from farm dams. This is estimated to be 10.7 GL which is less than 0.1 percent of the total surface water diversions within the Murray-Darling Basin (estimated by MDBC (2002)).

Table 2-3. Summary of farm dams in the Eastern Mount Lofty Ranges region

Dam type Total number of Total volume Assumed usage* dams GL GL/y** Stock and domestic 7300 15.6 4.68 Irrigation 700 6 3 Total 8000 21.6 7.68 *Excluding net evaporation **Assumption of 30% max volume use for stock and domestic; 50% for irrigation dams Source: Zulfic & Barnett (2007)

In addition there is 27 GL of regulated water supply allocated to irrigation and town use allocated from the River Murray system in the Angas Bremer irrigation area and a further 9.3 GL allocated for use in the lower Murray zone, also from the River Murray system. There is slightly over 14 GL of water currently sourced from the Murray River and Lake Alexandrina in the Angas Bremer. This volume of water use has increased from 3 GL in 1990 as a result of substitution of surface water for groundwater. During the same period the annual groundwater use in the Angas Bremer has reduced from 12 GL to a current annual use of 1.3 GL (Figure 2-3).

Water sourced from the Murray River and Lake Alexandrina has a high security rating similar to all regulated river allocations on the Murray River system in South Australia. The water sourced from the tributary streams is unregulated flow.

Pasture and lucerne production and horticulture, including vines, are the major irrigated industries produced within the Eastern Mount Lofty Ranges region. The areas of irrigated production are small (13,200 ha) compared to many other regions within the Murray-Darling Basin. Almost 60 percent of the irrigated land use is horticulture production.

10 2 Overview of the region Annual waterAnnual use (GL/y) 0 1980/1 1985/6 1990/1 1995/6 2000/1 2005/6

Figure 2-3. History of groundwater extractions including total diversion limits for the Angas Bremer GMU. Source: RMCWMB (2001), MDBC (2007) and DWLBC (2004)

2.4 Groundwater

2.4.1 Groundwater management units – the hydrogeology and connectivity

The three GMUs in the EMLR region have been assigned a low priority ranking 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.

There are some areas within the EMLR region that are not assigned to GMUs. These unassigned areas lie within the Plains region east of the Marne Saunders GMU, south of Mannum and south of Murray Bridge. The areas south of Mannum and south of Murray Bridge are excluded from this assessment since the salinity of groundwater in these areas is generally greater than 3000 mg/L.

20 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

Table 2-4. Categorisation of GMUs, including annual extraction, entitlement and recharge details

Code Priority GMU Current groundwater Total entitlement Long-term average Recharge ranking extraction extraction limit (from rainfall) GL/y GL GL/y S14 Low Eastern Mount Lofty Ranges 15.6 46* 46* 61.38 S18 Low Angas Bremer 1.2 6.5 6.5 *** S23 Low Marne Saunders 2.0 4** 4** 3 Unincorporated area (east of Not available Not licensed Not licensed Marne Saunders) * A water allocation plan for the EMLR GMU is currently being prepared. The DWLBC recommended interim allocation will be confirmed after environmental investigations are completed. ** A water allocation plan for the Marne Saunders is currently being prepared. The annual allocation volume is an interim value. *** The confined aquifer receives minimal recharge from rainfall or streamflow.

The EMLR region is topographically separated into a highland (Hills) zone and a lowland (Plains) zone. The Hills zone is Overview coincident with fractured basement rock aquifers of the Mount Lofty Ranges and the Plains Zone with sedimentary aquifers of the Murray Basin. Groundwater is also contained within the alluvial fill within highland valleys.

The fractured rock aquifers in the highland region consist of a number of different geological units which include the of Barossa Complex, the Kanmantoo Group, the Adelaidean, and the Normanville Group. the Recharge to the Barossa Complex and Kanmantoo Group aquifers is limited due to the impermeable nature of the rocks region and the limited fractures and joints in which groundwater is stored and transmitted.

These aquifers generally have poor yields (<3 L/s) and relatively higher salinity (2000 mg/L to 3000 mg/L). The consolidated sedimentary aquifers of the Adelaidean have jointing and fractures, allowing greater recharge, resulting in relatively high yields of up to 10 L/s to 15 L/s, and low salinity groundwater (<1500 mg/L). In addition, the Normanville Group has secondary porosity thereby increasing its potential yield.

Within the southern hills regions of the EMLR GMU there are unconsolidated Permian Sand aquifers in broad glacially carved valleys, such as Ashbourne and Tookayerta, which vary in productivity. Where sands are present, yields are high (30 L/s) and groundwater salinities are low (<500 mg/L), elsewhere the presence of more clay-natured sediments may have led to lower yields and higher salinities.

Unconsolidated sedimentary aquifers of the Plains include the Renmark Group which is not generally considered as a productive aquifer due to the presence of better quality water in the overlying Murray Group Limestone. This unit is highly fossiliferous and of sandy limestone containing groundwater of varying salinity. Significant recharge occurred to this aquifer in the Marne Saunders and Angas Bremer GMUs several thousand years ago during a much wetter climate and now supplies groundwater for irrigation, stock and domestic uses. However, there is very little current recharge in the Angas Bremer GMU with current extractions gradually depleting the lower salinity groundwater resource and resulting in salinity increases (Zulfic and Barnett, 2003). The Marne Saunders is recharged through rainfall infiltration and flood recharge.

The groundwater flow direction across the Plains is generally eastwards from the Eastern Mount Lofty Ranges towards the River Murray valley or Lake Alexandrina, which are among the lowest points in the landscape.

2 Overview of the region

Figure 2-4. Map of groundwater management units within the Eastern Mount Lofty Ranges region

In the Hills zone, groundwater moves from the higher points in the landscape to the lowest where discharge occurs along streams, providing baseflow in this zone. Groundwater also feeds permanent pools along main drainage lines. Where the streams flow out of the hills onto the plains, the streams change from gaining to losing and begin to recharge the underlying sedimentary aquifers. This form of recharge may be locally and seasonally important. For example recharge during floods in the Marne Saunders GMU is the main source of recharge to the productive aquifer (Barnett et al., 2001).

2.4.2 Water management institutional arrangements

Water Allocation Plans (WAPs) for the Marne Saunders and the EMLR Prescribed Water Resources Areas (PWRAs) are currently being prepared and both areas are subject to a moratorium on the extraction of groundwater. The Angas Bremer WAP has been operational since 2001, but will be incorporated into the WAP for the EMLR PWRA, with a review of the sustainable yield currently being carried out. Allocations are proposed to be made according to the type of water use activity, with different volumes allocated depending on the size and nature of existing uses. Users with existing surface water access will be required to nominate a ratio between their existing surface water use and groundwater use (DWLBC, 2004).

The Angas Bremer WAP identifies a number of groundwater dependent ecosystems and there are general provisions provided, usually related to maintaining current groundwater levels. There is no explicit allocation of groundwater for the environment.

22 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

Allocation volumes in the Angas Bremer WAP are fixed but adaptive provisions are provided. For example, if there is an unused portion of water on a licence at the end of a water-use year, up to 30 percent of the licensed allocation volume may be taken and used after the end of that water-use year and at any time in the subsequent three water-use years alongside an allocation for a subsequent water-use year. Ongoing monitoring and periodic evaluation of the capacity of the groundwater resource to meet demand will occur allowing for modification of WAP policy as new plans are developed. Currently, the allocation volumes are fixed for the life of the plan.

Table 2-5. Summary of groundwater management plans

Name of plan Angas Bremer Water Allocation Plan Year of plan 2001 Basic rights 2 Domestic and stock rights Not licensed, but taken into account in the water allocation planning process Overview Native title Not licensed Access licences Urban Not licensed, but taken into account in the water allocation planning process of

Planned share 6.5 GL/y the Announced allocation Not applicable region Environmental provisions It is assumed that the needs of the environment will be met if current groundwater levels are generally maintained Supplementary provisions Not applicable NB. Water allocation planning arrangements have not been completed for the EMLR and Marne Saunders GMUs

2.4.3 Water products and use

Groundwater extraction within the Eastern Mount Lofty Ranges region accounts for around 1.3 percent of the total groundwater extraction throughout the Murray-Darling Basin (MDBC, 2007). There are approximately 2440 groundwater users within the region (Zulfic and Barnett, 2003). Around 60 percent of operational wells are used for stock and domestic purposes and around 40 percent of wells were primarily installed for irrigation. The irrigation wells are concentrated in the areas around Mount Barker, Meadows, Macclesfield, Ashbourne, Springton, between Langhorne Creek and Lake Alexandrina, and between Currency Creek and Goolwa.

There is very little metered extraction within the EMLR GMU. As such, there is no comprehensive historical record of groundwater extraction. Drilling history can be used as a partial indicator of trends in groundwater development (Table 2-6). The number of wells constructed between 1995 and the completion of the study by Zulfic and Barnett (2003) is around three-quarters of the number of wells constructed over the previous 20 years. This suggests a high level of groundwater development and use has occurred in recent times.

Table 2-6. Historic installation of bores in the EMLR GMU

Irrigation Period No. of bores drilled: All subcatchments <1976 845 1976–1995 1039 >1995 721 Source: Zulfic and Barnett et al. (2003)

Prior to the prescription of the Angas Bremer GMU there were extractions within the Murray Group Limestone aquifer of up to 27 GL/year (Barnett, 2007). The use of surface water from Lake Alexandrina in place of groundwater has allowed a reduction in the rate of groundwater extraction to current levels of around 1 GL/year to 2 GL/year. Groundwater extraction in the Marne Saunders GMU has fallen by around 30 percent between 2002/03 and 2005/06 from around 2 GL/year to slightly over 1 GL/year.

2.5 References

Barnett SR (2007) Currency Limestone Management Area – Status Report 2007. South Australia. Department of Water, Land and Biodiversity Conservation. DWLBC Technical Note 2007/10. Barnett SR, Zulfic D and Yan W (2001) Marne River Catchment Groundwater Assessment. Department for Water Resources, Groundwater Assessment Resource Assessment Division. Report DWR 2001/009. BRS (2000) Land use data. Available at: http://adl.brs.gov.au/mapserv/landuse/

2 Overview of the region CSIRO (2007) Water Availability in the Murray. A report to the Australian Government from the CSIRO Murray-Darling basin Sustainable Yields Project. CSIRO, Australia. In prep. DWLBC (2004) Marne-Saunders Prescribed Water Resources Area Discussion Paper. Department of Water, Land and Biodiversity Conservation. MDBC (2002) Water audit monitoring report 2000/01. Report of the Murray Darling Basin Commission on the Cap on Diversions. MDBC Canberra. MDBC (2007) Updated summary of estimated impact of groundwater extraction on stream flow in the Murray-Darling Basin. Draft Report. MDBC Canberra. River Murray Catchment Water Management Board (RMCWMB) (2003) Catchment Water Management Plan for the River Murray in South Australia. RMCWMB . South Australian Murray-Darling Basin Natural Resources Management Board (SAMDBNRMB) (2006) Natural Resources management Plan for the South Australian Murray-Darling basin natural resources Management Region 2006-2007. Government of South Australia. Zulfic D and Barnett SR (2003) Eastern Mount Lofty Ranges groundwater assessment. South Australia. Department of Water, Land and Biodiversity Conservation. DWLBC Report 2003/25. Zulfic D and Barnett SR (2007) Angas Bremer Prescribed Wells Area – Groundwater Status Report 2007. South Australia. Department of Water, Land and Biodiversity Conservation. DWLBC Technical Note 2007/09.

24 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

3 Rainfall-runoff modelling

This chapter includes information on the climate and rainfall-runoff modelling for the Eastern Mount Lofty Ranges (EMLR) region. It has four sections:

• a summary • an overview of the regional modelling approach • a presentation and description of results, and • a discussion of key findings.

3.1 Summary Rifl-uofmodelling 3 Rainfall-runoff 3.1.1 Issues and observations The methods used for climate scenario and rainfall-runoff modelling across the Murray-Darling Basin are described in Chapter One. There are no significant differences in the methods used to model the EMLR region. However, for this region the results presented in this chapter are not used in the river system modelling presented in Chapter Four (see Chapter Four for details).

3.1.2 Key messages • The mean annual rainfall and modelled runoff averaged over the EMLR region are 463 mm and 30 mm respectively. Rainfall is significantly higher in the winter half of the year and most of the runoff occurs in winter and early spring. The EMLR region covers 0.4 percent of the Murray-Darling Basin and contributes about 0.5 percent of the total runoff in the Murray-Darling Basin. • The mean annual rainfall and runoff over the past ten years (1997 to 2006) averaged over the entire region are 7 percent and 36 percent lower respectively than the 1895 to 2006 long-term means. However, because of the inter-annual rainfall variability and the relatively short ten-year period used as the basis for comparison, the 1997 to 2006 rainfall is not statistically different to the 1895 to 1996 rainfall even at a significance level of α = 0.2. The 1997 to 2006 runoff is statistically different to the 1895 to 1996 runoff, but only at a significance level of α = 0.1. • Rainfall-runoff modelling with climate change projections from global climate models suggests that future runoff in the EMLR will decrease significantly. Nearly all the modelling results with different global climate models show a decrease in runoff. The best estimate (median) is a 15 percent reduction in mean annual runoff by ~2030 relative to ~1990. The extreme estimates, which come from the high global warming scenario, range from a 44 percent reduction to no change in mean annual runoff. By comparison, the range from the low global warming scenario is from a 15 percent reduction to no change in mean annual runoff. • Projected increases in commercial forestry plantations and farm dam storage volume in the EMLR region by ~2030 of 2000 ha and 3400 ML (or an increase of 16 percent of current farm dam storage volume) respectively are used in the modelling. These projected increases would reduce mean annual runoff by about three percent in addition to the median climate change impact on runoff. The median estimate of the combined impact of climate change and increases in plantations and farm dams is an 18 percent reduction in mean annual runoff, with extreme estimates ranging from 3 percent to 46 percent reductions.

3.1.3 Uncertainty • Scenario A – historical climate and current development The runoff estimates for the EMLR region are relatively good because there are four calibration catchments in the higher runoff producing areas in the western and south-western parts of the region from which to estimate the model parameter values. Rainfall-runoff model verification analyses for the Murray-Darling Basin indicate that the mean annual runoffs estimated for ungauged catchments using optimised parameter values from a nearby catchment have an error of less than 20 percent in more than half the catchments and less than 50 percent in almost all the catchments.

• 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 rainfall-runoff 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 Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPEC, 2007). The results are then presented as a median estimate of climate change impact on runoff and as the range of the extreme estimates. • 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 commercial forestry plantations and farm dam development and the impact of these developments on runoff. The increase in commercial forestry is based on Bureau of Rural Sciences projections that take into account industry information. The increase in farm dams is estimated by considering trends in historical farm dam growth and future policy controls that are likely to be implemented in the EMLR. With regards to both plantations and farm dams in the EMLR, there is uncertainty both as to how land holders will respond to these policies and how governments may set their future policies.

3.2 Modelling approach

3.2.1 Rainfall-runoff modelling – general approach

The general rainfall-runoff modelling approach is described more fully in Chapter One and in detail in Chiew et al.

3 Rainfall-runoff modelling (2007a). A brief summary is given below.

The lumped conceptual daily rainfall-runoff model, SIMHYD, with a Muskingum routing method is used to estimate daily runoff at 0.05o grids (~ 5 km x 5 km) across the entire Murray-Darling Basin for the four scenarios. The rainfall-runoff model is calibrated against 1975 to 2006 streamflow from about 180 small and medium size unregulated catchments (50 km2 to 2000 km2). In the model calibration, the six parameters of SIMHYD are optimised to maximise an objective function that incorporates the Nash-Sutcliffe efficiency 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 five percent of the total recorded runoff. The runoff for a 0.05o 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 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 Murray-Darling Basin 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. (2007a) 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.

3.2.2 Rainfall-runoff modelling for the Eastern Mount Lofty Ranges The rainfall-runoff modelling is carried out to estimate runoff in 0.05o grid cells in 32 subcatchments as defined for the river system modelling in Chapter 4 for the EMLR region (Figure 3-1). Optimised parameter values from four calibration catchments are used.

For Scenario D modelling, projected increases in commercial forestry plantations and farm dam storage volumes in the entire EMLR of 2000 ha and 3400 ML respectively by ~2030 are used. The ~2030 projection of commercial forestry comes from the Bureau of Rural Sciences which takes into account industry information. The additional 2000 ha of commercial forestry plantations is distributed to areas adjacent to existing plantations (which are not natural forest land use) with the highest biomass productivity, all of which are in two subcatchments in south-west EMLR (see Appendix A).

26 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

To estimate farm dam growth, it was assumed that the combined reduction of mean annual runoff from forestry plantations and farm dams in each management subcatchment is limited to a maximum of 30 percent of the estimated natural May through November runoff volume. This is inferred from the South Australia Natural Resources Management Act (2004), under which the EMLR is proclaimed as a prescribed area, and guidance from the South Australia Department of Water, Land and Biodiversity Conservation as to the future policy that may apply once the catchment water management plan for the EMLR is completed. It should be noted that these are initial estimates as data on irrigation farm dams are currently being compiled as part of the Land and Water Use surveys. The increase in farm dams in each subcatchment is then estimated as the lower of the estimated 30 percent of May through November runoff, after allowing for the reduction in runoff resulting from existing and projected new commercial forestry plantations, and the projected additional storage volume based on extrapolation of the historical farm dam growth rate. The projected increase in farm dam storage volume over the entire EMLR of 3400 ML is a 16 percent increase from the existing volume. The projected increases in farm dam storage volume by ~2030 for each subcatchment are given in Appendix A. Rifl-uofmodelling 3 Rainfall-runoff

Figure 3-1. Map of modelling subcatchments and calibration catchments

3.2.3 Model calibration Figure 3-2 compares the modelled and observed monthly runoff and the modelled and observed daily flow duration curves for the four calibration catchments. The results indicate that the SIMHYD calibration can reproduce satisfactorily the observed monthly runoff series (Nash-Sutcliffe E values greater than 0.8) and the daily flow duration characteristic (Nash Sutcliffe E values greater than 0.8). The volumetric constraint used in the model calibration also ensures that the total modelled runoff is within five percent of the total observed runoff.

The calibration to optimise Nash-Sutcliffe E means that more importance is placed on the simulation of high runoff, and therefore SIMHYD modelling of the medium and high runoff are considerably 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. This is evident in some of the scatter plots comparing the modelled and observed monthly runoff and clearly seen in the daily flow duration curves. The discernible disagreement between the modelled and observed daily runoff characteristics only occurs for runoff that is exceeded less than 0.1 or one 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. In any case, the volumetric constraint used in the model calibration ensures that the total modelled runoff is always within five percent of the total observed runoff.

The runoff estimates for the EMLR region are relatively good because there are four calibration catchments in the higher runoff producing areas in the western and south-western parts of EMLR from which to estimate the model parameter values. The EMLR region is also a lot smaller than the other reporting regions in the Murray-Darling Basin Sustainable Yields Project, allowing more meaningful use of parameter values from a neighbouring catchment. The rainfall-runoff model verification analyses for the Murray-Darling Basin with data from about 180 catchments indicate that the mean annual runoffs for ungauged catchments are under or over estimated, when using optimised parameter values from a nearby catchment, by less than 20 percent in more than half the catchments and by less than 50 percent in almost all the catchments (see Chiew et al. (2007a) for more detail).

3 Rainfall-runoff modelling

Figure 3-2. Modelled and observed monthly runoff and daily flow duration curves for the calibration catchments

3.3 Modelling results

3.3.1 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 EMLR 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 2006.

The mean annual rainfall and modelled runoff averaged over the EMLR region are 463 mm and 30 mm respectively. The mean annual rainfall varies from about 900 mm in the west and south-west to less than 300 mm in the north-east. The modelled mean annual runoff varies from about 200 mm in the south-west to under five mm in the north-east (Figure 3-3). Rainfall is significantly higher in the winter half of the year and most of the runoff occurs in winter and early spring (Figure 3-5). The EMLR region covers 0.4 percent of the Murray-Darling Basin and contributes about 0.5 percent of the total runoff in the Murray-Darling Basin.

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 (Figure 3-4). The coefficients of variation of annual rainfall and runoff averaged

28 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

over the EMLR are 0.21 and 0.68 respectively, amongst the least variable of the 18 Murray-Darling Basin reporting regions. The tenth percentile, median and ninetieth percentile values across the 18 reporting 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 past ten years (1997 to 2006) are 7 percent and 36 percent lower respectively than the 1895 to 2006 long-term means. However, because of the inter-annual rainfall variability and the relatively short ten-year period used as the basis for comparison, the 1997 to 2006 rainfall is not statistically different to the 1895 to 1996 rainfall even at a significance level α = 0.2 (with the Student-t and Rank Sum tests). The 1997 to 2006 runoff is statistically different to the 1895 to 1996 runoff, but only at a significance level of α = 0.1. Because the 1997 to 2006 runoff is statistically different to the long-term mean, Scenario B modelling is undertaken. The Scenario B is a stochastic replicate selected such that its long term mean annual runoff matches the mean annual runoff of the last ten years. Potter et al. (2007) present a more detailed analysis of recent rainfall and runoff across the Murray-Darling Basin.

Rifl-uofmodelling 3 Rainfall-runoff

Figure 3-3. Spatial distribution of mean annual rainfall and modelled runoff averaged over 1895–2006

800 120

100 600 80

400 60

40 200 20 Annual runoff (mm) runoff Annual Annual rainfall (mm) rainfall Annual 0 0 1895 1915 1935 1955 1975 1995 1895 1915 1935 1955 1975 1995

Figure 3-4. 1895–2006 annual rainfall and modelled runoff series averaged over the region. The curve shows the low frequency variability.

80 10

8 60 6 40 4

20 2 Mean monthly runoff (mm)

Mean monthly rainfall (mm) 0 0 JFMAMJJASOND JFMAMJJASOND

Figure 3-5. Mean monthly rainfall and modelled runoff (averaged over 1895–2006 for the region)

3.3.2 Scenario C – future climate and current development Figure 3-6 shows the percentage change in the modelled mean annual runoff averaged over the EMLR region for Scenario C relative to Scenario A for the 45 scenarios (15 GCMs for each of the high, medium and low global warming scenarios). The percentage change in the mean annual runoff and the percentage change in mean annual rainfall from the corresponding GCMs are also tabulated in Table 3-1 (see Chiew et al. (2007b) for description of the GCMs and detailed discussion of method used to obtain Scenario C climate series).

The plot and table indicate that climate change would significantly reduce runoff in the EMLR with practically all the

3 Rainfall-runoff modelling modelling results showing a decrease in runoff.

Because of the large variation between GCM simulations and the method used to obtain the climate change scenarios (see Section 1.3.3), the biggest increase and biggest decrease in runoff come from the high global warming scenario. For the high global warming scenario, rainfall-runoff modelling with climate change projections from two-thirds of the GCMs indicates a decrease in mean annual runoff greater than 10 percent, and none of the modelling results show an increase in mean annual runoff greater than one percent.

In subsequent reporting here and in other Sections, only results from an extreme ‘dry’, ‘mid’ and extreme ‘wet’ variant are shown (referred to as Cdry, Cmid and Cwet). For the Cdry scenario, results from the second highest reduction in mean annual runoff from the high global warming scenario are used. For the Cwet scenario, results from the second highest increase in mean annual runoff from the high global warming scenario are used. For the Cmid scenario, the median mean annual runoff results from the medium global warming scenario are used. These are shown in bold in Table 3-1, with the Cdry, Cmid and Cwet scenarios indicating a -44, -15 and 0 percent change in mean annual runoff. By comparison, the range based on the low global warming scenario is -15 to 0 percent change in mean annual runoff.

Figure 3-7 shows the mean annual runoff across the EMLR region for Scenario A and for the Cdry, Cmid and Cwet scenarios.

30 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

60 High global w arming

40 Medium global w arming

Low global w arming 20

-20

-40 % change in mean annualinrunoff mean % change Rifl-uofmodelling 3 Rainfall-runoff -60 iap mri mpi ipsl gfdl miub cnrm csiro miroc inmcm giss_aom ncar_pcm cccma_t63 cccma_t47 ncar_ccsm

Figure 3-6. Percentage change in mean annual runoff from the 45 Scenario C simulations (15 GCMs and three global warming scenarios) relative to Scenario A runoff

Table 3-1. Summary results from the 45 Scenario C simulations (numbers show percentage 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 giss_aom -20 -49 giss_aom -13 -34 giss_aom -6 -16 ipsl -19 -44 ipsl -12 -31 ipsl -5 -15 cnrm -14 -41 gfdl -12 -29 gfdl -5 -14 gfdl -18 -40 cnrm -9 -28 cnrm -4 -13 csiro -14 -39 csiro -9 -27 csiro -4 -13 inmcm -10 -31 inmcm -7 -22 inmcm -3 -10 mri -10 -27 mri -6 -20 mri -3 -9 miub -8 -22 miub -5 -15 miub -2 -7 ncar_ccsm -4 -17 ncar_ccsm -3 -12 ncar_ccsm -1 -5 iap -4 -13 iap -3 -8 iap -1 -4 mpi -4 -5 mpi -2 -6 mpi -1 -3 miroc 1 -4 miroc 1 -3 miroc 0 -1 ncar_pcm 2 -1 ncar_pcm 1 -1 ncar_pcm 0 0 cccma_t63 1 0 cccma_t63 1 0 cccma_t63 0 0 cccma_t47 -2 1 cccma_t47 -1 1 cccma_t47 0 0

3 Rainfall-runoff modelling

Figure 3-7. Mean annual rainfall and modelled runoff under scenarios A, Cdry, Cmid and Cwet

32 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

3.3.3 Summary results for all modelling scenarios Table 3-2 shows the mean annual rainfall, modelled runoff and actual evapotranspiration (rainfall minus runoff) for Scenario A (averaged over the EMLR region), and the percentage changes in the rainfall, runoff and actual evapotranspiration in Scenarios B, C and D relative to Scenario A. Figure 3-8 shows the mean monthly rainfall and modelled runoff for Scenarios A, C and D averaged over 1895 to 2006 for the region. Figure 3-9 shows the daily rainfall and flow duration curves for Scenarios A, C and D averaged over the region. The modelling results for all the subcatchments in the EMLR region are summarised in Appendix A.

It should be noted that the Cmid (or Cdry or Cwet) results are from rainfall-runoff modelling using climate change projections from one GCM. As the Cmid scenario is chosen based on mean annual runoff (see Section 3.3.2), the comparison of monthly and daily results in Scenario Cmid relative to Scenario A in Figure 3-8 and Figure 3-9 should be interpreted cautiously. However, the Crange 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 Crange results shown in Rifl-uofmodelling 3 Rainfall-runoff 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 Crange are therefore not the same as the Cdry and Cwet scenarios reported elsewhere and used in the river system and groundwater models.

Figure 3-8 indicates that the GCM projections show a bigger decrease in the winter-half rainfall compared to summer-half rainfall and this translates to an even bigger percent runoff reduction in the winter half when most of the runoff in the EMLR occurs. Although almost all the GCMs show a reduction in mean annual rainfall, more than half of the GCMs indicate that the extreme rainfall that is exceeded 0.1 percent of the time will be more intense (see also Figure 3-9).

The mean annual runoff over the past ten years (1997 to 2006) is 36 percent lower than the 1895 to 2006 long-term mean. For Scenario B modelling, 100 replicates of 112-year daily climate sequences are generated using the annual rainfall characteristics over 1997 to 2006 (see Chiew et al. (2007b) for more details). The replicate that reproduced the 1997 to 2006 mean annual runoff is used to obtain catchment inflows for the river system modelling described in Chapter 4. The mean annual rainfall for the replicate chosen is 14 percent lower than the mean annual rainfall in Scenario A, compared to the 1997 to 2006 rainfall being only 7 percent lower than the 1895 to 2006 long-term mean (Table 3-2).

The modelling results indicate a median estimate of -15 percent change in mean annual runoff by ~2030 (Scenario C). However, there is considerable uncertainty in the climate change impact estimate with extreme estimates ranging from -44 percent no change in mean annual runoff.

Table 3-2. Water balance over the entire region by scenario

Scenario Rainfall Runoff Evapotranspiration mm A 463 30 433 percent change from Scenario A B -14% -36% -5% Cdry -19% -44% -18% Cmid -5% -15% -4% Cwet 1% 0% 1% Ddry -19% -46% -17% Dmid -5% -18% -4% Dwet 1% -3% 1%

(a) (b)

80 10 Scenario C range Scenario C range Scenario A 8 Scenario A 60 Scenario Cmid Scenario Cmid 6 Scenario Dmid 40 4

20 2

Mean monthly rainfall (mm) 0 Mean monthly runoff (mm) 0 JFMAMJJASOND JFMAMJJASOND

Figure 3-8. Mean monthly rainfall and modelled runoff under scenarios A, C and D averaged over 1895-2006 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)

(a) (b) 3 Rainfall-runoff modelling

Figure 3-9. Daily flow duration curves under scenarios A, C and D 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 EMLR region are 463 mm and 30 mm respectively. The mean annual rainfall varies from about 900 mm in the west and south-west to less than 300 mm in the north-east. The modelled mean annual runoff varies from about 200 mm in the south-west to under 5 mm in the north-east. Rainfall is significantly higher in the winter half of the year and most of the runoff occurs in winter and early spring. The EMLR reporting region covers about 0.5 percent of the Murray-Darling Basin and contributes about 0.5 percent of the total runoff in the Murray-Darling Basin.

The mean annual rainfall and modelled runoff over the past ten years (1997–2006) are 7 percent and 36 percent lower respectively than the 1895 to 2006 long-term means. However, because of the inter-annual rainfall variability and the relatively short ten-year period used as the basis for comparison, the 1997 to 2006 rainfall is not statistically different to the 1895 to 1996 rainfall even at a significance level of α = 0.2. The 1997 to 2006 runoff is statistically different to the 1895 to1996 runoff, but only at a significance level of α = 0.1.

Although the rainfall over the past ten years (1997–2006) is 7 percent lower than the 1895 to 2006 long-term mean, the runoff is 36 percent lower than the long-term mean. The likely reasons for this include: rainfall-runoff is a nonlinear process and the changes in rainfall are amplified more in runoff in a drier climate; subsurface water storage is low after a long dry period and significant amount of rainfall is required to fill the storage before runoff can occur; and changes in the daily and seasonal rainfall distribution and sequencing of rainfall events could have amplified the reduction in runoff.

34 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

The mean annual runoff over the ten year period is almost as low as the projected decrease in mean annual runoff in the extreme dry climate change scenario. However, because it is based on a relatively short ten years of data, it is not sufficient evidence that the hydroclimate has shifted to a new regime. Nevertheless, if the hydroclimate has shifted to a new regime (like in the extreme dry climate change scenario), the dry conditions over the past ten years will occur more frequently.

Rainfall-runoff modelling with climate change projections from global climate models indicates that future runoff in the EMLR would decrease significantly. Practically all the modelling results with different global climate models show a decrease in runoff. Most of the global climate models show a greater reduction in winter-half rainfall and this translates to Rifl-uofmodelling 3 Rainfall-runoff an even bigger percent reduction in winter-half runoff, when most of the runoff in the EMLR occurs. However, although the projections indicate a decrease in mean annual rainfall and runoff, more than half of the results also indicate that the extreme rainfall events will be more intense.

The median estimate is a 15 percent reduction in mean annual runoff by ~2030 relative to ~1990. However, there is considerable uncertainty in the modelling results with the extreme estimates ranging from -44 percent to no change in mean annual runoff. These extreme estimates come from the high global warming scenario, and for comparison the range from the low global warming scenario is -15 percent to no change in mean annual runoff. The main sources of uncertainty are in the global warming projections and the global climate modelling of local rainfall response to the global warming. The uncertainty in the rainfall-runoff modelling of climate change impact on runoff is small compared to the climate change projections.

For Scenario D modelling, projected increases in commercial forestry plantations and farm dam storage volumes in the EMLR region of 2000 ha and 3400 ML respectively by ~2030 are used. The median estimate of the combined impact of climate change and farm dam development is an 18 precent reduction in mean annual runoff (with extreme estimates ranging from -46 percent to -3 percent). The modelled reduction in mean annual runoff from these projected increases in commercial forestry plantations and farm dams alone is therefore about 3 percent, compared to the 15 percent reduction in the median climate change projection.

There is considerable uncertainty in the projection of future increase in commercial forestry and farm dam development and the impact of these developments on runoff. The increase in commercial forestry is based on Bureau of Rural Sciences projections that take into account industry information. The increase in farm dams is estimated by considering trends in historical farm dam growth and future policy controls that are likely to be implemented in the EMLR. With regards to both plantations and farm dams in the EMLR, there is uncertainty both as to how land holders will respond to these policies and how governments may set their future policies.

3.5 References

Chiew et al. (2007a) 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 et al. (2007b) 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. 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. Potter NJ, Chiew FHS, Frost AJ, Srikanthan R, McMahon TA, Peel MC and Austin JM (2007) 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. In prep. South Australia Government (2004) Natural Resources Management Act 2004.

4 River system modelling

This chapter includes information on the river system modelling for the Eastern Mount Lofty Ranges (EMLR) region. It has four sections:

• a summary of major issues and observations, key messages and the robustness of the results • an overview of the regional modelling approach • a presentation and description of results • a discussion of key findings.

The information in this chapter is derived from the Department of Water, Land and Biodiversity Conservation’s (DWLBC)

six calibrated models of the Marne, Bremer, Angas, Finniss, Tookayerta and Currency catchments developed in the g WaterCRESS modelling framework. The modelling and data presented in this report describe the region as presented in Figure 4-1. Water availability in the EMLR should only be taken to represent the region which is described by the models.

4.1 Summary stem modellin y s

r 4.1.1 Issues and observations

River system modelling for the EMLR region considered eight modelling scenarios:

4 Rive 1. Scenario A Scenario A is the baseline model developed and used by DWLBC and is run for the common historic climate period (1 June1895 to 30 June 2006) and represents current levels of development. This scenario is the baseline scenario that all other scenarios are compared against.

2. Scenario B Scenario B represents the climatic conditions of the last ten years applied over the common historic modelling period. The level of development is the same as Scenario A (Current level of development). 3. Scenarios Cwet, Cmid and Cdry The C scenarios represent a range of future climate conditions that are derived by adjusting the historic climate and flow inputs used in Scenario A which is described in Chapter 3. The level of development is the same as Scenario A, i.e. current level of development. 4. Scenarios Dwet, Dmid and Ddry The D scenarios are based on the C scenarios and use the climate inputs used in the respective C scenarios. Flow inputs are adjusted for 2030 estimates of development in farm dams and commercial plantation forestry. The farm dam and forestry projections are discussed in Chapter 3 while groundwater development is discussed in Chapter 6. For the river system modelling, increases in commercial plantation forestry were modelled by an areal catchment reduction proportional to the expected decrease in runoff. Only two catchments (Tookayerta Creek and Currency Creek) were affected by expected increases in commercial plantation forestry and had the areal reduction applied.

The change in inflows between scenarios reported in this chapter differs from the changes in runoff reported in Chapter 3. These differences are due to the difference in areas that are considered to contribute runoff to the surface water model. In Chapter 3 the entire region is considered while a sub-set of this area is considered in this chapter.

The modelling approach used in the EMLR differs from the approach used in other regions as EMLR uses existing calibrated rainfall-runoff models to adjust inflows. For other regions the results from SIMHYD models are used to adjust the inflow time series used by surface water models. In the EMLR climate inputs for the various scenarios are input into the WaterCRESS model and a new time series of inflows is generated by the model. Despite the different modelling approach when comparing the total average annual end of system flow for current development and historic climate the results are similar with WaterCRESS generating 110 GL/year and SIMHYD generating 109 GL/year.

36 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

These scenarios are not of conditions that will eventuate but are of 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 future. This assessment does not elaborate on what management actions might be taken to address any of these pressure points.

The river system modelling and reporting for the EMLR considers water availability for irrigation use, stock and domestic use from farm dams, and water for the environment by modelling the rainfall-runoff process and water storage behaviour to determine catchment water balances.

The rainfall-runoff modelling is similar to that reported in Chapter 3. However, it is carried out at a finer spatial resolution. The results derived for the different scenarios use the same climate inputs; although the results in this chapter contain information about both subcatchment inflows and water storage and supply behaviour. Comparison between the modelling approaches suggests that the implicit modelling farm dams in SIMHYD may underestimate farm dam impacts in extreme dry climate conditions as increased evaporative losses under new climate conditions are not explicitly 4 River system modelling represented. However, for all but the extremely dry climate scenarios the results are very similar suggesting for the level of detail required in this report, the implicit modelling of farm dams through the SIMHYD approach was found to be sufficiently similar to the WaterCRESS modelling to provide confidence in the implicit modelling approach.

The river systems of the EMLR are mainly ephemeral in nature. The effect of pronounced reductions in rainfall and increases in evaporation on the catchment is twofold; with firstly a reduced amount of soil moisture (decreased rainfall), and then a more rapid depletion of the soil moisture store via evapotranspiration (increased evaporation).

End-of-system flows referred to in this section are taken at the end of the gaining reach of each modelled catchment, as no stream losses are modelled across the plains regions of the EMLR by WaterCRESS.

4.1.2 Key messages

• Current long-term average water availability across the Marne, Bremer, Angas, Finniss, Tookayerta and Currency catchments of EMLR is 120 GL/yr. Current average net diversions from farm dams (including net evaporation) across these catchments are around 11 GL/year – or about 9 percent of the average available water. The level of use varies however, across the six catchments considered, from about 5 to 18 percent. • A continuation of the climate of the last ten years (1997 to 2006) would reduce water availability by 23 percent – or 28 GL/year less water available on average. This would lead to a 6 percent reduction in net diversions from farm dams on average; however, the lowest 1-year diversions would decrease by 21 percent. These changes effectively increase the level of use to 11 percent for these climate conditions. The average period between spilling for farm dams across the region would increase by 33 percent increase. • Under the best estimate (median) 2030 climate an 18 percent reduction in water availability (about 22 GL/year) would be expected. However, because the level of water use is low, the reduction in average net diversions from farm dams would be only 5 percent, effectively increasing the level of surface water use to 10 percent. • The climate extremes for 2030 indicate: o Under the wet extreme, a three percent decrease in water availability but a one percent increase in average net diversions from farm dams; and o Under the dry extreme, a 52 percent decrease in water availability and a 19 percent reduction in average net diversions from farm dams. Thus in the worst future case water availability would less than half the current average level and the level of surface water use would be 15 percent of this. • The combined impact of a potential 16 percent increase in total farm dam volume (an additional 3400 ML) and of a potential doubling of the area of commercial plantation forestry (an additional 2000 ha), both of which could occur by 2030, would be a 3 percent reduction in water availability in addition to the climate change impacts. These changes would actually increase net diversions considerably, as more farm dams provide more opportunity for water use. Future commercial plantation forestry growth would only be likely in the wetter Tookayerta and Currency catchments, for which the modelled reductions in runoff are 17 and 7 percent respectively. The level of surface water use with these developments for the best estimate 2030 climate would be 12 percent.

4.1.3 Robustness

One WaterCRESS model was run for an extreme climate scenario to assess how robustly it would behave. Typically the physical processes in the model, such as routing and storage behaviour, work through a full range of flow and storage conditions. However management rules in the model are closely tied to the historic data set that was used to develop them. When the historic data set is changed to much drier conditions there is no guarantee that models will behave robustly. As there are no water management models it was expected that all EMLR models would behave robustly.

The model response to increases and decreases in inflow was reasonable with the change in diversions and end of systems flows consistent with the change in inflow. Mass balance over the modelling period was maintained with 0.1 percent for all scenarios (Appendix B).

The future development scenario uses growth projections for farm dams and commercial plantation forestry. The uncertainties associated with the projection used in this project (including in the WaterCRESS modelling) are articulated

g in Chapter 3.

4.2 Modelling approach

The following section provides a summary of the generic river modelling approach, a description of the EMLR catchment stem modellin y models and how the river models were developed. Refer to Chapter 1 for more details on the overall project methodology. s r

4.2.1 General 4 Rive River system models that encapsulate descriptions of current infrastructure, water demands, and water management and sharing rules are used to assess the implications of the changes in inflows described above in this chapter 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 and the Murray-Darling Basin Commission. The main models in use are IQQM, REALM, MSM-Bigmod, WaterCRESS and a model of the Snowy Mountains Hydro-electric Scheme.

A few areas of the Basin 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. A key example is the Pride model used to estimate irrigation for Victorian REALM models.

1. Model configuration

2. 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

3. Using scenario climate and inflow time series, run river model for all scenarios (historical, recent and future climate as well as future development)

4. Where relevant, extract initial estimates of surface-groundwater exchanges and provide to groundwater model

5. Where relevant, use revised estimates of surface-groundwater exchanges from groundwater models and re-run river model for all scenarios.

For river systems that receive inflows or transfer from upstream or adjacent river systems, model inputs for each scenario are taken from the ‘upstream’ model(s). In a few cases several iterations are 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.

38 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

4.2.2 Model description

The EMLR reporting region is described by six separate water balance models developed using the WaterCRESS (Version 2, March 2007) modelling platform. In this chapter results are presented for each the six modelled catchments (Marne, Bremer, Angas, Finniss, Tookayerta and Currency) with five of the catchments being divided into two subcatchments. This division was based on the location of the most downstream gauging station and any residual ungauged area within the catchment.

Three of the modelled catchments have reaches in excess of 15 km on the plains downstream of the hills where most of the runoff is produced. These reaches are thought to lose significant proportions of flow. Estimation of loss has not been attempted for these reaches, although they are the subject of ongoing investigations by DLWBC, and as such actual catchment outflows to the River Murray may be overestimated.

WaterCRESS is a freely available modelling platform incorporating some of Australia’s more widely used hydrologic 4 River system modelling models such as WC-1, SIMHYD, HYDROLOG, and AWBM. Runoff for five of the catchments (Bremer River, Angas River, Finniss River, Tookayerta Creek, and Currency Creek) was generated using WC-1 (Cresswell, 2002) – a water balance model developed specifically for the ephemeral systems of the Mount Lofty Ranges. AWBM was used for the Marne catchment.

The size of modelled subcatchments varies over the EMLR depending on the level of farm dam development, topography and land-use information. For the combined models the EMLR is described by 1083 rural catchment nodes and 1010 farm dam nodes representing around 6800 farm dams within the modelled catchments. The total farm dam capacity is 18,129 ML.

The WaterCRESS platform allows the incorporation of various components to generate and control the runoff within the model. These components include:

• water storage nodes used to describe both reservoirs and on and off-stream farm dams;

• transfer components including weirs and routing nodes;

• catchment components including rural and urban catchment nodes; and

• inflow components used to describe catchment inflows such as imported water and effluent.

Of the above components the water storage and catchment components make up the vast majority of the nodes used to model the EMLR. Catchments are subdivided into rural catchment node components that form the network of streams. Due to the extensive development of farm dams throughout the rural catchments a farm dam node is usually incorporated at the downstream end of each rural catchment node. In the EMLR there are around 8000 farm dams with 6800 falling within modelled catchments. This large number necessitated the aggregating of farm dams into single nodes representing all of the storages within the respective rural catchment node.

The EMLR is predominantly rural in nature and urban runoff nodes were applied only in the models for the Bremer and Angas catchments. The urban model used was an initial-loss continuing-loss model with the area of urban development being divided into three components; roof area, paved area and roads. Determination of the ratio of these three components was done by analysing selected study areas from aerial photography.

Catchment inflows, via treated effluent, were applied only in the model of the Bremer catchment. For this study population was assumed to be similar under all modelling scenarios.

The EMLR models were initially developed to estimate the impact of farm dams on streamflow in each of the catchments for the purpose of informing the water allocation planning process which is currently underway in the region. These investigations are the subject of published reports and ongoing work carried out by DLWBC (Savadamuthu, 2002, 2003, 2004, 2006; Alcorn, 2006, 2007). Full descriptions of model set-up and calibration results can be found in these reports.

DWLBC projections for farm dam growth were a total of 2948 ML that was distributed to the Marne (478), Bremer (427), Angas (964), Finniss (680) and Tookayerta (399). No farm dams were added to the Currency Creek catchment. The additional farm dams projected for 2030 are represented explicitly in the WaterCRESS models, aggregated to rural nodes. Note these projections differ slightly from those applied in Chapter 3 (3400 ML of additional farm dams).

DWLBC projections for future growth in commercial plantation forestry were 2000 ha added to the Tookayerta Creek and Currency Creek catchments. The approach to model commercial plantation forestry was to reduce the runoff from those modelled subcatchments (represented by rural nodes) where growth was projected. Studies on runoff reduction due to forestry in the Mount Lofty Ranges have shown up to an 85 percent decrease in annual runoff (Greenwood & Cresswell, 2007). That maximum reduction in runoff was assumed for this study. Using a spatial dataset describing the locations of future commercial plantation forestry, the areas identified were reduced in area by 85 percent. Consequently, 2000 ha reduced the catchment areas of Tookayerta Creek and Currency Creek by 1700 ha.

g stem modellin y s r 4 Rive

Figure 4-1. River system map showing rivers, gauges and model sub-catchments

4.2.3 Model setup

Model setup involved the following:

• The modelling period used in the pre-calibrated models was extended to cover the common modelling period used in this study; • Determination of the initial state of all storages within the model; and • Check the model for robust operation in an example extreme dry climate scenario.

As the original modelling period was 1974 to 2003 climate inputs had to be extended to encompass the full 1895 to 2006 modelling period. Models were extended using climate data from the SILO Point-Patched Dataset (Jeffrey et al., 2001; and Bureau of Meteorology, 2007).

40 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

Whilst the EMLR system contains a large amount of farm dam storage, it contains no active reservoirs for public water supply. The large volume of water in the dams is the cumulative effect of around 6800 small farm dams. Due to the distributed nature of the model and relatively small individual volumes, the initial storage volume has little impact on the results of the model, with the storage state of most dams resetting each year. To account for any initial storage effects, the model was started with a six month warm up period 1 January 1895 to 30 June 1895 (Table 4-1).

Table 4-1. Model setup information

Model Version Start date End date

Marne River WaterCRESS (AWBM) 2.0 1/10/1974 31/12/2003

Bremer River WaterCRESS (WC-1) 2.0 1/10/1974 31/12/2003

Angas River WaterCRESS (WC-1) 2.0 1/10/1974 31/12/2003 4 River system modelling

Finniss River WaterCRESS (WC-1) 2.0 1/10/1974 31/12/2003

Tookayerta Creek WaterCRESS (WC-1) 2.0 1/10/1974 31/12/2003

Currency Creek WaterCRESS (WC-1) 2.0 1/10/1974 31/12/2003

Connection

No connection to Murray

Baseline models

Warm up period 01/01/1895 30/06/1895

Marne River WaterCRESS (AWBM) 2.0 01/08/1895 30/06/3006

Bremer River WaterCRESS (WC-1) 2.0 01/08/1895 30/06/3006

Angas River WaterCRESS (WC-1) 2.0 01/08/1895 30/06/3006

Finniss River WaterCRESS (WC-1) 2.0 01/08/1895 30/06/3006

Tookayerta Creek WaterCRESS (WC-1) 2.0 01/08/1895 30/06/3006

Currency Creek WaterCRESS (WC-1) 2.0 01/08/1895 30/06/3006

Farm dam capacity (ML) 18,129 6800 storages

Connection

No connection to Murray

EMLR modifications

Data Extend to 30/06/2006

Inflows No adjustment required

Initial storage volume Resets in warm up period

To determine model robustness, a test was made on one of the models (Currency Creek). The model was configured for an extreme dry climate scenario by applying seasonal factors to rainfall and evaporation (Table 4-2). The model was run and behaved as expected.

Table 4-2. Rainfall, evaporation and flow factors for model robustness test

Season Rainfall Evaporation DJF 0.938 1.048 MAM 0.888 1.047 JJA 0.843 1.044 SON 0.885 1.044

4.3 Modelling results

4.3.1 River system water balance g Appendix B contains mass balance tables for eight subcatchments in the six 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 less than 0.1 percent. Table 4-3 shows the net fluxes for the EMLR region. Fluxes for Scenario A are given in GL/year with other scenario results given as percentage changes from Scenario A. stem modellin y The following points summarise the entries in the table: s r Inflows:

• Imported water: mean annual inflows to the catchment via effluent flows (fixed supply) 4 Rive • Urban runoff: mean annual runoff generated from urban areas • Subcatchment flows: mean annual runoff generated from catchment areas

Diversion:

• Farm dam usage: total average use from farm dams (fixed annual demand)

Outflows:

• End-of-system outflows: mean annual flow at end-of-system nodes for each model • Net evaporation: mean annual net evaporative loss from all farm dam nodes • River loss: mean loss from sink components used to reduce flows • The change in storage between 30 June 1895 and 30 June 2006 averaged over the 111-year period is also included.

42 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

Table 4-3. River system model average annual water balance under scenarios P, A, B, C and D

P A B Cwet Cmid Cdry Dwet Dmid Ddry

Model start date 01/01/1889 01/07/1895

Model end date 31/12/1999 30/6/2006

GL/y Percent change from Scenario A

Storage volume

Change over period 0.0 0.1 -15% -4% -13% -49% 14% 3% -38%

Inflows

Imported water 0.4 0.4 0% 0% 0% 0% 0% 0% 0%

Urban runoff 1.2 1.2 -9% 0% -6% -23% 0% -6% -23%

Sub-catchment flows 120.6 120.6 -22% -3% -18% -51% -6% -20% -53% 4 River system modelling

Sub-total 122.2 122.2 -22% -3% -18% -51% -6% -20% -53%

Diversions

Total estimated use from farm dams 0.0 6.3 -2% -1% -2% -12% 12% 10% -2%

Sub-total 0.0 6.3 -2% -1% -2% -12% 12% 10% -2%

Outflows

End-of-system outflows

Marne River 10.8 8.9 -13% -6% -24% -63% -7% -25% -64%

Bremer River 22.0 19.8 -23% -6% -27% -69% -6% -28% -70%

Angas River 17.1 15.1 -37% -4% -20% -56% -8% -23% -59%

Finniss River 45.9 42.7 -36% -4% -20% -55% -5% -20% -56%

Tookayerta Creek 17.4 16.7 -1% -2% -12% -37% -22% -30% -50%

Currency Creek 8.3 7.4 -4% 11% -9% -51% 2% -16% -54%

Net evaporation 0.1 4.8 4% 4% 2% 3% 13% 11% 10%

River loss 0.4 0.4 4% 2% 6% 9% 2% 6% 9%

Sub-total 121.9 115.7 -23% -3% -18% -53% -7% -22% -55%

Unattributed fluxes

Total 0.2 0.1 -59% -7% -25% -61% -7% -25% -61%

4.3.2 Inflows and water availability

Inflows

To assess the impact of the various scenarios on the typical flows, the mean and median daily flows at the end-of-system are tabulated for Scenario A together with the percentage change from Scenario A (Table 4-4).

Table 4-4. Mean and median flows (ML/day) under Scenario A and for other scenarios relative to Scenario A

A B Cwet Cmid Cdry Dwet Dmid Ddry

ML/d Percent change from Scenario A

Marne River

Mean 21.95 -13% -6% -24% -63% -7% -26% -64%

Median 0.007 -57% -43% -71% -100% -43% -71% -100%

Bremer River

Mean 54.28 -23% -6% -27% -69% -6% -28% -70%

Median 5.05 -66% -13% -49% -100% -17% -53% -100%

Angas River

Mean 40.73 -37% -4% -20% -57% -8% -23% -60% g Median 6.63 -58% -10% -34% -81% -26% -45% -85%

Finniss River

Mean 116.83 -36% -4% -20% -55% -5% -20% -56%

Median 21.36 -51% -7% -29% -79% -10% -31% -80%

stem modellin Tookayerta Creek y s

r Mean 45.75 -1% -2% -12% -37% -22% -30% -50%

Median 19.92 -1% -4% -15% -49% -25% -35% -60%

Currency Creek 4 Rive Mean 20.28 -4% 11% -9% -51% 2% -16% -54%

Median 3.61 1% 26% -7% -64% 19% -11% -65%

Water availability

Water availability in the EMLR is defined as the total water available in the system – total outflows at the end of system plus any upstream diversions. The water availability indicator (Table 4-5) shows the change in water availability relative to the baseline scenario of historic climate and current development.

Table 4-5. Change in annual water availability under scenarios B, C and D relative to Scenario A

A B Cwet Cmid Cdry Dwet Dmid Ddry GL/y Percent change from Scenario A 120.4 -23% -3% -18% -52% -6% -21% -54%

A time series of annual water availability under Scenario A is shown in Figure 4-2. This is the summed availability for all six catchments. The lowest annual water availability was 24 GL in 1982 while the highest annual water availability was 283 GL in 1923. Figure 4-3 shows the differences from Scenario A in annual water availability for Scenario B and the C and D scenario variants.

44 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

) 350 300 250 200 150 100 50 0 Annual water availability (GL availability water Annual 1895 1915 1935 1955 1975 1995

Figure 4-2. Annual water availability under Scenario A

4 River system modelling

(a)

50 )

-50

-100

-150

-200 B

-250 Annual water availablity (GL availablity water Annual 1895 1915 1935 1955 1975 1995

(b) (c)

50 50 ) ) 0 0

-50 -50

-100 -100

-150 -150 C range D range -200 -200 Cmid Dmid -250 -250 Annual water availablity (GL availablity water Annual Annual water availablity (GL availablity water Annual 1895 1915 1935 1955 1975 1995 1895 1915 1935 1955 1975 1995

Figure 4-3. Annual water availability relate to Scenario A under (a) Scenarios B, (b) Scenario C and (c) Scenario D

4.3.3 Storage behaviour

Analysis of the storage behaviour of 6800 individual farm dams in the models was not practical. Storage behaviour was analysed by aggregating the total volumes across all catchment on each day giving a combined storage capacity of 18,129 ML for current levels of development. To assess behaviour of the aggregated storage a ‘spilling threshold’ was set at 75 percent of the combined maximum capacity. This was used as the basis for determining the average and maximum years between spills; the minimum storage volume and date were also assessed (Table 4-6). The behaviour of the aggregated farm dam storage across the models gives an indication of the level of regulation of this system as well as how reliable total storage is during extended periods of low or no inflows.

Table 4-6. Details of combined farm dam behaviour

Total for all farm dams A B Cwet Cmid Cdry Dwet Dmid Ddry Minimum storage volume (ML) 3003 2083 2858 2482 947 3077 2626 1002 Minimum storage date 08/04/1915 08/04/1915 08/04/1915 08/04/1915 08/04/1915 08/04/1915 08/04/1915 08/04/1915 Average years between spills 0.6 0.8 0.6 0.7 1.6 0.7 0.9 1.9 Maximum years between spills 2.5 3.6 2.5 2.6 7.6 2.6 3.6 8.6

(a) (b)

20000 20000 g

15000 15000

10000 10000 Volume (ML) 5000 Volume (ML) 5000 stem modellin A A Cw et y B Cmid Cdry s r 0 0 1942 1944 1946 1948 1950 1942 1944 1946 1948 1950

4 Rive (c)

20000

15000

10000

Volume (ML) 5000 A Dw et Dmid Ddry 0 1942 1944 1946 1948 1950

Figure 4-4. Behaviour of the water storages over the maximum days between spills under (a) scenarios A and B, (b) scenarios A and C, (c) scenarios A and D

4.3.4 Consumptive water use

Net diversions

As the water balance models used for this study apply mostly to farm dam storages, the term diversions is used here to represent the total water diverted (gross farm dam water use). It includes water lost through evaporation off the farm dam. Farm dam storages within the model are supplied by a fixed annual demand that is between 30 and 50 percent of each dam’s volume. Table 4-7 and Figure 4-5 show the total average annual diversions for each subcatchment under Scenario A and the percentage change of all other scenarios compared to Scenario A.

46 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

Table 4-7. Change in net diversions in each sub-catchment relative to Scenario A

A B Cwet Cmid Cdry Dwet Dmid Ddry Reach GL/y Percent change from Scenario A Marne River 1.4 -2% 0% -8% -31% 6% -3% -29% Bremer 1.8 -5% 1% -7% -30% 6% -3% -28% Angas 2.1 -6% 3% -2% -14% 23% 17% 0% Finniss 3.5 -6% 2% -2% -14% 10% 5% -9% Tookayerta 0.9 1% 3% 0% -5% 15% 10% 3% Currency 1.0 -11% -7% -12% -23% -11% -16% -29% Total 10.7 -6% 1% -5% -19% 10% 3% -14%

4 River system modelling

(a) (b)

4.0 4.5 Crange 4.0 Drange )

) 3.5 Cmid Dmid 3.5 3.0 A A 3.0 2.5 2.5 2.0 2.0 1.5 1.5 1.0 1.0 Annual diversionsAnnual (GL Annual diversionsAnnual (GL 0.5 0.5 0.0 0.0 M arne B remer A ngas Finniss Tookayert a Currency M arne B remer A ngas Finniss Tookayert a Currency

Figure 4-5. Total average annual diversions for sub-catchments under (a) scenarios A and C and (b) scenarios A and D

Figure 4-6 shows an annual time series of total net diversion for Scenario A and the difference from Scenario A for scenarios B, C and D. The maximum annual diversion for Scenario A is 12.3 GL in 1970 and the minimum is 5.7 GL in 1914.

(a) (b)

14 2

) 12 1 10 0 8 -1 6 -2 4 -3 2 -4 Annual differenceAnnual (GL) Annual diversionsAnnual (GL 0 -5 1895 1915 1935 1955 1975 1995 1895 1915 1935 1955 1975 1995

(c) (d) g 2 2 1 1 0 0 -1 -1 -2 -2 stem modellin y -3 -3 s r -4 -4 Annual differenceAnnual (GL) differenceAnnual (GL) -5 -5 1895 1915 1935 1955 1975 1995 1895 1915 1935 1955 1975 1995

4 Rive

(e) (f)

2 2 1 1 0 0 -1 -1 -2 -2 -3 -3 -4 -4 Annual differenceAnnual (GL) Annual differenceAnnual (GL) -5 -5 1895 1915 1935 1955 1975 1995 1895 1915 1935 1955 1975 1995

(g) (h)

2 2 1 1 0 0 -1 -1 -2 -2 -3 -3

-4 differenceAnnual (GL) -4 Annual differenceAnnual (GL) -5 -5 1895 1915 1935 1955 1975 1995 1895 1915 1935 1955 1975 1995

Figure 4-6. Total diversions for (a) Scenario A and difference between total water use for Scenario A and (b) Scenario B (c) Scenario Cwet; (d) Scenario Cmid; (e) Scenario Cdry; (f) Scenario Dwet; (g) Scenario Dmid; and (h) Scenario Ddry

48 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

Level of use

The level of use for the region is indicated by the ratio of total net diversions to total inflows. Total net diversions are defined as the water extracted from farm dams. This indicates the share of the available water that is diverted for consumptive use. Net diversions for a system with only farm dam storages refers to the total amount of water extracted from the farm dams for consumptive use plus the total net evaporative loss from farm dams. That is, the total water removed from the system.

Net diversions are used to reflect the change in mass balance of the system. They do not take into consideration the difference in water quality that may exist between diversions and returns. In the case of the EMLR models no returns are modelled. Table 4-8 shows the level of use indicators for each of the scenarios. The level of use is relatively low with 9 percent of the water resource being diverted under Scenario A and the maximum level being 16 percent under Scenario Ddry.

4 River system modelling

Table 4-8. Relative level of use under scenarios A, B, C and D

A B Cwet Cmid Cdry Dwet Dmid Ddry 9% 11% 9% 10% 15% 10% 12% 16%

Use during dry periods

Table 4-9 shows the average use, as well as the average annual use for the lowest one, three and five-year periods for Scenario A and the percentage change from Scenario A for each other scenario. These figures indicate the impact on water use during dry periods.

Table 4-9. Indicators of net use including net evaporation during dry periods under scenarios A, B, C and D

Annual Diversion A B Cwet Cmid Cdry Dwet Dmid Ddry GL/y Percent change from Scenario A Lowest 1-year period 5.7 -21% 0% -12% -56% 5% -8% -55% Lowest 3-year period 8.7 -10% 0% -7% -33% 8% 0% -30% Lowest 5-year period 9.7 -8% 1% -7% -28% 9% 0% -25% Average 10.7 -6% 1% -5% -19% 10% 3% -14%

Reliability

The average reliability of water products in a system where farm dam storage is the main supply can be indicated by the ratio of total diversions (water use from farm dams) to the total maximum possible diversion allowed by the fixed demand model. That is, if the maximum possible demand from all farm dams in the model is set between 30 and 50 percent of the farm dam capacity, then reliability is set against that benchmark. For Scenario C the maximum possible use is 6.2 GL/year, for Scenario D it is increased proportionally with the increase in development to 7.4 GL/year.

(a) (b)

8 8

6 6

4 4

C range D range 2 Cmid 2 Dmid

Annual diversionAnnual (GL) A diversionAnnual (GL) A

0 0 1895 1945 1995 1895 1945 1995

g Figure 4-7. Comparison of allocated and used water under (a) Scenario A and Scenario C, and (b) Scenario A and Scenario D

Table 4-8 show the difference between the maximum possible diversion and diversion for each of the scenarios in volume reliability plots. stem modellin y s r (a) (b)

8 8

4 Rive 7 7 6 6 5 5 4 4 3 3 Allocated 2 Allocated 2 Diverted (Cwet) 1 Diverted (A) Diverted (Cmid) Diverted (B) volume (GL)Annual 1 Diverted (Cdry) Annual volume (GL) volume Annual 0 0 0 20 40 60 80 100 0 20406080100 % of years exceeded % of years exceeded

(c)

8 7 6 5 4 3 Allocated 2 Diverted (Dwet) Diverted (Dmid) 1 Diverted (Ddry) Annual volume (GL) volume Annual 0 0 20406080100 % of years exceeded

Figure 4-8. Diversion reliability under (a) scenarios A and B; (b) Scenario C and (c) Scenario D

Table 4-10 shows for each of the farm dam net diversions the average annual difference between available water and diverted water. This table gives an indication of the level of utilisation of the farm dams in the EMLR region.

50 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

Table 4-10. Summary of average reliability under scenarios A, B, C and D

A B Cwet Cmid Cdry Dwet Dmid Ddry GL/y Allocated water 6.6 6.6 6.6 6.6 6.6 7.5 7.5 7.5 Diversion 6.2 6.0 6.1 6.0 5.4 6.9 6.8 6.0 Difference 0.4 0.6 0.5 0.6 1.2 0.6 0.8 1.5

4.3.5 River flow behaviour 4 River system modelling

End-of-system flow characteristics

Figure 4-9, Figure 4-10 and Figure 4-11 show the end-of-system flow duration curves for the six modelled catchments. Each of the scenarios is plotted on the same plot. Cease-to-flow is considered to occur when model flows are less than 1 ML/d.

(a) Marne River (b) Bremer River

1000 10000 P P A A B 1000 B 100

100 10 10

Daily flow(ML) 1 Daily flow(ML) 1

0.1 0.1 0 20406080100 0 20406080100

g (c) Angas River (d) Finniss River

10000 10000 P P A A 1000 B 1000 B

100 100 stem modellin y s r 10 10 Daily flow(ML) Daily flow(ML) 1 1 4 Rive 0.1 0.1 0 20406080100 0 20406080100

(e) Tookayerta Creek (f) Currency Creek

10000 10000 P P A A 1000 B 1000 B

100 100

10 10 Daily flow (ML) Daily flow (ML) 1 1

0.1 0.1 0 20406080100 0 20406080100 % time flow is exceeded % time flow is exceeded

Figure 4-9. Daily flow duration curves for modelled subcatchments under scenarios P, A and B

52 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

(a) Marne River (b) Bremer River

1000 10000 C range C range Cmid Cmid A 1000 A 100 P P 100 10 10

Daily flow(ML) 1 Daily flow(ML) 1

0.1 0.1 0 20406080100 0 20406080100

4 River system modelling (c) Angas River (d) Finniss River

10000 10000 C range C range Cmid Cmid 1000 A 1000 A P P 100 100

10 10

Daily flow(ML) 1 Daily flow(ML) 1

0.1 0.1 0 20406080100 0 20406080100

(e) Tookayerta Creek (f) Currency Creek

10000 10000 C range C range Cmid Cmid 1000 A 1000 A P P 100 100

10 10

Daily flow (ML) flow Daily 1 (ML) flow Daily 1

0.1 0.1 0 20406080100 0 20406080100 % time flow is exceeded % time flow is exceeded

Figure 4-10. Daily flow duration curves for modelled subcatchments under scenarios P, A and C

(a) Marne River (b) Bremer River

1000 10000 D range D range Dmid Dmid A 1000 A 100 P P

100 10 10

Daily flow (ML) flow Daily 1 Daily flow(ML) 1

0.1 0.1 0 20406080100 0 20406080100

g (c) Angas River (d) Finniss River

10000 10000 D range D range Dmid Dmid 1000 A 1000 A P P

100 100 stem modellin y s r 10 10 Daily flow (ML) flow Daily Daily flow(ML) 1 1 4 Rive 0.1 0.1 0 20406080100 0 20406080100

(e) Tookayerta Creek (f) Currency Creek

10000 10000 D range D range Dmid Dmid 1000 A 1000 A P P

100 100

10 10 Daily flow (ML) Daily flow (ML) 1 1

0.1 0.1 0 20406080100 0 20406080100 % time flow is exceeded % time flow is exceeded

Figure 4-11. Daily flow duration curves for modelled sub-catchments under scenarios P, A and D

Figure 4-12, Figure 4-13 and Figure 4-14 show the mean monthly flow for the pre-development scenario, Scenario A, Scenario B, Scenario C and Scenario D for each of the end-of-system flow gauges. This shows how the seasonality at the end-of-system has changed between all of the scenarios. It also shows the average change in volume for each month.

54 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

(a) Marne River (b) Bremer River

3000 6000 P P 2500 A 5000 A B B 2000 4000

1500 3000

1000 2000

Monthly (ML) flow 500 Monthly (ML) flow 1000

0 0 JFMAMJ JASOND JFMAMJ JASOND

4 River system modelling (c) Angas River (d) Finniss River

4000 10000 P P 3500 A A 8000 3000 B B

2500 6000 2000 1500 4000 1000 Monthly (ML) flow Monthly (ML) flow 2000 500 0 0 JFMAMJ JASOND JFMAMJJASOND

(e) Tookayerta Creek (f) Currency Creek

3500 2000 P P 3000 A A B B 2500 1500 2000 1000 1500

1000 500 Monthly (ML) flow Monthly (ML) flow 500 0 0 JFMAMJ JASOND JFMAMJ JASOND

Figure 4-12. Seasonal flow duration curves under scenarios P, A and B at (a) Marne River; (b) Bremer River, (c) Angus River, (d) Finniss River, (e) Tookayerta Creek, and (f) Currency Creek

(a) Marne River (b) Bremer River

3000 6000 Crange Crange 2500 Cmid 5000 Cmid A A 2000 P 4000 P

1500 3000

1000 2000

Monthly (ML) flow 500 Monthly (ML) flow 1000

0 0 JFMAMJJASOND JFMAMJJASOND

g (c) Angas River (d) Finniss River

4000 10000 Crange Crange 3500 Cmid Cmid 8000 3000 A A P P 2500 6000 stem modellin

y 2000

s 1500 4000 r 1000 Monthly (ML) flow Monthly (ML) flow 2000 500

4 Rive 0 0 JFMAMJJASOND JFMAMJJASOND

(e) Tookayerta Creek (f) Currency Creek

3500 2000 Crange Crange 3000 Cmid Cmid A 1500 A 2500 P P 2000 1000 1500

1000 500 Monthly (ML) flow Monthly (ML) flow 500 0 0 JFMAMJJASOND JFMAMJJASOND

Figure 4-13. Seasonal flow duration curves under scenarios P, A and C at (a) Marne River; (b) Bremer River, (c) Angus River, (d) Finniss River, (e) Tookayerta Creek, and (f) Currency Creek

56 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

(a) Marne River (b) Bremer River

3000 6000 Drange Drange 2500 Dmid 5000 Dmid A A P P 2000 4000

1500 3000

1000 2000

Monthly (ML) flow 500 Monthly (ML) flow 1000

0 0 JFMAMJ JASOND JFMAMJ JASOND

4 River system modelling (c) Angas River (d) Finniss River

4000 10000 Drange Drange 3500 Dmid Dmid 8000 3000 A A P P 2500 6000 2000 1500 4000 1000 Monthly (ML) flow Monthly (ML) flow 2000 500 0 0 JFMAMJ JASOND JFMAMJJASOND

(e) Tookayerta Creek (f) Currency Creek

3500 2000 Drange Drange 3000 Dmid Dmid A 1500 A 2500 P P 2000 1000 1500

1000 500 Monthly (ML) flow Monthly (ML) flow 500 0 0 JFMAMJ JASOND JFMAMJ JASOND

Figure 4-14. Seasonal flow duration curves under scenarios P, A and D at (a) Marne River; (b) Bremer River, (c) Angus River, (d) Finniss River, (e) Tookayerta Creek, and (f) Currency Creek

Table 4-11 shows the size of daily events with 2, 5 and 10-year recurrence intervals for P, A, B, C and D scenarios for each of the catchments. Note this analysis estimates the average peak daily flow and not the peak flow for a day, which is considerably higher in most river systems.

Table 4-11. Daily flow event frequency and magnitude for scenarios P, A, B, C and D

Return interval P A B Cwet Cmid Cdry Dwet Dmid Ddry Years ML/d Percent change from Scenario A Marne 2 689 589 -14% -5% -23% -65% -9% -24% -66% 5 1220 1175 -13% -9% -29% -68% -10% -30% -69% 10 1742 1726 -7% -2% -17% -58% -2% -18% -59% Bremer 2 2428 2238 -21% -1% -26% -77% -2% -27% -78% 5 4060 3999 -19% -5% -23% -72% -5% -23% -73% 10 5131 5107 -21% -6% -23% -61% -6% -24% -62% g Angas 2 1962 1987 -28% -3% -12% -47% -3% -17% -52% 5 2726 2710 -24% 1% -10% -41% -3% -15% -44% 10 3276 3274 -29% -3% -12% -40% -4% -13% -43% Finniss stem modellin y 2 1237 1218 -38% -10% -23% -59% -10% -24% -59% s r 5 1744 1761 -31% -3% -14% -47% -3% -15% -48% 10 2015 2018 -31% -5% -16% -39% -5% -16% -39%

4 Rive Tookayerta 2 503 511 -1% -3% -13% -33% -23% -31% -48% 5 677 692 0% 0% -8% -30% -22% -27% -45% 10 810 827 -1% 1% -9% -31% -21% -28% -46% Currency 2 705 685 -4% 11% -5% -41% -1% -16% -49% 5 1033 1025 -2% 5% -8% -39% -6% -17% -46% 10 1269 1247 0% 11% 0% -38% 0% -11% -46%

The percent of time flow occurs for these scenarios are presented Table 4-12. Cease-to-flow is considered to occur when model flows are less than 1 ML/day.

Table 4-12. Percent of time stream continues to flow under scenarios A, B, C and D

Outflow Name A B Cwet Cmid Cdry Dwet Dmid Ddry

Marne River 53.4% 52.1% 52.7% 51.6% 45.6% 52.7% 51.5% 45.6% Bremer River 67.1% 55.9% 64.7% 58.7% 41.8% 64.2% 58.2% 41.3% Angas River 95.5% 84.6% 94.2% 90.6% 73.9% 93.1% 89.1% 71.9% Finniss River 99.8% 99.7% 99.8% 99.8% 98.7% 99.8% 99.8% 98.7% Tookayerta Creek 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Currency Creek 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%

58 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

4.3.6 Sharing of water resources

Non-diverted water shares

There are several ways of considering the relative level of impact on non-diverted water and diversions. Table 4-13 presents two indicators for relative impact on non-diverted water:

• the average annual non-diverted water as a proportion of the total available water • the average annual non-diverted share relative to the non-diverted share for Scenario A.

Table 4-13. Relative level of available water not diverted for use under scenarios A, B, C and D

A B Cwet Cmid Cdry Dwet Dmid Ddry 4 River system modelling Non-diverted water as a percentage of total 91% 89% 91% 90% 85% 90% 88% 84% available water Non-diverted share relative to Scenario A non- 100% 69% 148% 93% 69% 148% 93% 69% diverted share

Combined water shares

Figure 4-15 combines the results from water availability, use and non-diverted water into a bar chart. The size of the bars indicates total water availability and the sub-division of the bars indicates the diverted and non-diverted fractions.

180 160 Div erted 140 Non diverted 120 100 80 60 40 Annual water (GL) water Annual 20 0 A B Cw et Cmid Cdry Dw et Dmid Ddry

Figure 4-15. Comparison of diverted and non-diverted shares of water under scenarios A, B, C and D

4.4 Discussion of key findings

4.4.1 Water availability

Current water availability in EMLR averages 120.4 GL/year. Scenario B would lead to a 23 percent reduction in water availability – nearly 28 GL/year less. Scenario C results indicate that under a 2030 climate EMLR water availability would be between 3 and 52 percent lower than current levels, with the best estimate 2030 climate result being an 18 percent reduction (nearly 22 GL/year). This is a very significant reduction given the already low-yielding ephemeral systems of the EMLR.

In Scenario D, expansions in farm dams (2948 ML) and commercial plantation forestry (2000 ha) were simulated. These developments were capped to reflect the current development limit in relation to farm dams – 30 percent of the mean winter runoff from subcatchments as defined in the River Murray Catchment Water Management Plan (2003). The reduction in streamflow under Scenario D ranged from 6 to 54 percent. For the best estimate 2030 climate with future development the predicted reduction in mean annual runoff is 21 percent. This is a 3 percent additional reduction on top of the climate change impacts as a result of additional farm dams and future commercial plantation forestry development. This is a modest increase; however, it reflects the already high level of farm dam development in the EMLR.

The changes in mean and median daily flows varied between catchments. For Currency Creek both the mean and median daily flow increased under Scenario Cwet. All other catchments showed decreases in those indicators.

The worst affected catchments were those drier catchments that already had low median flows such as the Marne, in which the median flow falls to zero under scenarios Cdry and Ddry.

There is an increase in the size of some events from pre-development to current development (Table 4-11). This is due to farm dams spilling rainfall that falls directly on the farm dam. This creates higher peak flood events than would have occurred if rain fell on the ground.

4.4.2 Storage behaviour

The average period between spills increases from 0.6 years under Scenarios A to 0.8 years under Scenario B (the recent

climate scenario). This is a 33 percent increase in the average period. Under the best estimate 2030 climate the average g period would be 0.7 years, and with future development this would increase to 0.9 years. The maximum period between spills is predicted to increase from 2.5 years under Scenario A, to 3.6 years under scenarios B and Dmid and to 7.6 and 8.6 years under scenarios Cdry and Ddry respectively. Scenarios Cwet, Cmid and Dwet are not significantly impacted.

4.4.3 Consumptive water use stem modellin y s r Net Diversions

Current net diversions for EMLR are 10.7 GL/year. Under Scenario B – the recent climate scenario – these decrease by

4 Rive 6 percent.

Under 2030 climate the impacts on total net diversions for the EMLR range from a 1 percent increase (wet extreme) to a 19 percent decrease; the best estimate result is a 5 percent decrease. The increase in net diversions under Cwet indicates that while there is a net reduction in inflow, the climate changes are at the threshold between less water being available in the catchment and more water being diverted. This conditions shifts for Scenario Cmid to a net reduction in diversions.

As the demand on farm dams was not adjusted for climate the only impact on diversions is the reduction in inflows. For small reductions in inflows there is no significant reduction in diversions as spill volumes are reduced. This of course affects end-of-system flows. As inflows further reduce a threshold is reached where diversions are impacted by reduction in inflows. As the ratio between the demand and inflow increases the amount of evaporation decreases which affects the net diversion – since net diversions include net evaporation.

The results discussed above for the entire EMLR system mask the variations across catchments. In catchments with higher rainfall and a base flow component, net diversions may actually increase under a drier climate. Tookayerta Creek catchment, for example, has a high level of base flow and is generally the least impacted according to all indicators.

Scenario D results are more complicated. The drier catchments (Marne River, Bremer River) generally behave as under Scenario C, however, with a lower level of diversion. This is due to the in-line nature of farm dams and the requirement for a more upstream dam to fill and spill, before a more downstream dam can commence filling. This dynamic reduces the aggregated diversion in the system.

Reliability

The reliability of the system to provide water was assessed against the maximum possible use for a fixed demand. For scenarios B and C the maximum possible usage from farm dams was 6.6 GL/year. The ability for each farm dam to use the maximum is controlled by the volume in the dam available during the time of demand. Assessment of Scenario A suggests that although 6.6 GL/year is the average maximum possible water available for usage, on average, farm dams were only able to access 6.2 GL/year – a 7 percent deficit. Comparison with Scenario C suggests a deficit ranging from 7 to 18 percent with Scenario Cmid (and Scenario B) showing a deficit of 9 percent – a 2 percent reduction from Scenario A. Scenario D leads to a deficit ranging from 8 to 20 percent with Dmid showing a deficit of 10 percent – a 3 percent reduction from Scenario A.

60 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

4.4.4 End-of-system flows

End-of-system flows were analysed by inspecting the flow duration curves for each catchment and each scenario. Results can be typically grouped into the drier, ephemeral catchments (Marne River, Bremer River and Angas River) and the wetter perennial catchments (Finniss River, Tookayerta Creek, and Currency Creek). The end-of-system impact on high flows is generally covered in the water balance section. More scrutiny is given here to end-of-system low flows.

It should be noted that both streamflow measurement and modelling becomes more difficult for lower flows. Therefore the certainty of the results in relation to low flows and zero flows is lower and should be approached with more caution.

Nonetheless, analysis of zero flows suggests that again, the drier catchments exhibit far greater increases in the number of zero-flow days than the wetter catchments. Of all the catchments, the Bremer River shows the greatest increase in zero-flow days (13 percent increase under scenarios Cmid and Dmid compared to Scenario A). This result however, may be due to the inclusion of a loss node in the Bremer River model which is used to simulate the river losses in the lower 4 River system modelling part of the catchment. The wetter catchments exhibited little, if any, increase in the number of zero-flow days.

4.4.5 Comparison with SIMHYD modelling

Unlike other MDB regions, the river modelling for EMLR in this project does not use the results from the SIMHYD rainfall runoff-modelling (Chapter 3) for assessing changes in inflows. The distributed nature of the WaterCRESS model and in particular the spatial distribution of inflows, means that inflows were explicitly modelled using WaterCRESS. A comparison can therefore be made between the WaterCRESS and SIMHYD estimates of flows for the six EMLR catchments. Table 4-14 and Figure 4-16 (a–f) below show the comparison between the mean annual end-of-catchment flows for each catchment for each of the eight scenarios modelled. Table 4-14 compares the changes in end-of- catchment flows between the two models for each catchment and scenario.

Table 4-14. End-of-catchment outflows for WaterCRESS and SIMHYD under each scenario

A B Cwet Cmid Cdry Dwet Dmid Ddry

GL/y

Marne SIMHYD 8 5 8 7 4 7 6 4

WaterCRESS 8 7 8 6 3 7 6 3

Bremer SIMHYD 21 15 21 17 11 20 17 11

WaterCRESS 20 15 19 14 6 19 14 6

Angas SIMHYD 14 11 14 12 8 13 11 7

WaterCRESS 15 9 14 12 6 14 11 6

Finniss SIMHYD 42 33 42 36 24 41 35 23

WaterCRESS 43 27 41 34 19 41 34 19

Tookayerta SIMHYD 17 14 17 15 10 15 13 9

WaterCRESS 17 17 16 15 10 13 12 8

Currency SIMHYD 7 6 7 7 5 7 7 5

WaterCRESS 7 7 8 7 4 8 6 3

Table 4-15. Percentage reduction in end-of-catchment outflow for WaterCRESS and SIMHYD under each scenario

A B Cwet Cmid Cdry Dwet Dmid Ddry

GL/y Percent change from Scenario A

Marne SIMHYD 8 -38% 1% -17% -47% -7% -24% -52%

WaterCRESS 8 -13% -6% -24% -63% -7% -26% -64%

Bremer SIMHYD 21 -28% 0% -16% -46% -3% -18% -48%

WaterCRESS 20 -23% -6% -27% -69% -6% -28% -70%

Angas SIMHYD 14 -22% 0% -16% -45% -7% -22% -51%

WaterCRESS 15 -37% -4% -20% -57% -8% -23% -60%

Finniss SIMHYD 42 -22% 0% -15% -43% -4% -18% -46%

WaterCRESS 43 -36% -4% -20% -55% -5% -20% -56% g Tookayerta SimHyd 17 -17% 0% -12% -38% -11% -22% -46%

WaterCRESS 17 -1% -2% -12% -37% -22% -30% -50%

Currency SimHyd 7 -17% 0% -13% -39% 0% -13% -39%

WaterCRESS 7 -4% 11% -9% -51% 2% -16% -54% stem modellin y s r WaterCRESS predicts a greater reduction in flows under scenarios Cdry and Ddry than SIMHYD; although not large in volume, the differences between the WaterCRESS and SIMHYD results are more pronounced in the drier catchments. The differences are likely to be because, firstly, the SIMHYD modelling is based on climate inputs on a 5 km by 5 km grid 4 Rive across the entire EMLR while the WaterCRESS models interpolate between 17 climate stations, resulting in slightly different climate estimates across the EMLR. Secondly, the impacts of farm dams are implicitly modelled in SIMHYD by calibrating for the period 1995 to 2005. Farm dam development that occurred during this period has been averaged as part of the calibration process. WaterCRESS is configured to explicitly model current development (albeit in a simplified manner) and consequently is expected to predict larger impacts particularly under a drier climate where development is more pronounced. Finally, the impact of commercial plantation forestry in SIMHYD is estimated using FCFC (Brown et. al., 2007) that takes into account Zhang curves (Zhang et. al., 2001) and changes in flow duration curve shape, while WaterCress uses an areal reduction factor of 0.85. The reduction factor used in WaterCRESS causes a larger reduction than what was applied to SIMHYD flows (Table 4-16).

Table 4-16. Percentage change in runoff due to commercial plantation forestry in WaterCRESS and SIMHYD under best estimate 2030 climate scenario

Catchment SIMHYD WaterCress Percent Tookayerta Creek 16% 17% Currency Creek 6% 7%

Mass balance results (Table 4-3) indicate that the reduction in end-of-system flows in most cases is due to a reduction in subcatchment inflows rather than an increase in water diverted for farm dams. In most cases WaterCRESS predicted a greater reduction in inflow than SIMHYD. The relative changes due to future climate and development are more comparable for the ‘mid’ and ‘wet’ scenarios.

62 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

(a) Marne River (b) Bremer River

10 25 SimHyd SimHyd WaterCRESS WaterCRESS 8 20

6 15

4 10

2 5 Annual DischargeAnnual (GL) Annual DischargeAnnual (GL)

0 0 A B Cw et Cmid Cdry Dw et Dmid Ddry A B Cw et Cmid Cdry Dw et Dmid Ddry

4 River system modelling

20 50 SimHyd SimHyd WaterCRESS WaterCRESS 40 15

30 10 20

5 10 Annual DischargeAnnual (GL) DischargeAnnual (GL)

0 0 A B Cw et Cmid Cdry Dw et Dmid Ddry A B Cw et Cmid Cdry Dw et Dmid Ddry

(e) Tookayerta Creek (f) Currency Creek

20 10 SimHyd SimHyd WaterCRESS WaterCRESS 8 15

6 10 4

5 2 Annual DischargeAnnual (GL) DischargeAnnual (GL)

0 0 A B Cw et Cmid Cdry Dw et Dmid Ddry A B Cw et Cmid Cdry Dw et Dmid Ddry

Figure 4-16. Comparison between the mean annual flows for each catchment under scenarios A, B, C and D

4.5 References

Alcorn M (2006) Surface Water Assessment of the Currency Creek Catchment. Report DWLBC 2006/07, Department of Water, Land and Biodiversity Conservation, Government of South Australia. Alcorn M (2007) Surface Water Assessment of Bremer River Catchment Draft Report, Department of Water, Land and Biodiversity Conservation, Government of South Australia. Brown AE, Podger, GM, Davidson AJ, Dowling TI and Zhang L (2007) Predicting the impact of plantation forestry on water users at local and regional scales. An example for the Murrumbidgee River Basin, Australia. Forest Ecology Management doi:10.1016/j.foreco. 2007.06.011. Bureau of Meteorology (2007) Meteorological and agricultural information for rural interests. Available at: http://www.nrm.gov.au/silo. Cresswell DJ (2002) WaterCRESS, Water Community Resource Evaluation and Simulation System. Unpublished reference manual. Department of Water, Land and Biodiversity Conservation, Government of South Australia.

Greenwood A and Cresswell D (2007) The Impact of Plantation Forestry on Runoff in the Mount Lofty Ranges, Case Study: Burnt Out Creek. Technical Note DWLBC 2007/XX, Department of Water, Land and Biodiversity Conservation, Government of South Australia. 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, 309–330. Savadamuthu K (2002) Impact of Farm Dams on Streamflow in the Upper Marne Catchment. Report DWR 02/01/0003, Department for Water Resources, Government of South Australia. Savadamuthu K (2003) Streamflow in the Upper Finniss Catchment Report DWLBC 2003/18, Department of Water, Land and Biodiversity Conservation, Government of South Australia. Savadamuthu K (2004) Surface Water Assessment of the Tookayerta Catchment Report DWLBC 2004/23, Department of Water, Land and Biodiversity Conservation, Government of South Australia. Savadamuthu K (2006) Surface Water Assessment of the Upper Angas Sub-Catchment Report DWLBC 2006/09, Department of Water, Land and Biodiversity Conservation, Government of South Australia. Zhang L, Dawes WR and Walker GR (2001) The response of mean annual evapotranspiration to vegetation changes at a catchment scale. Water Resources Research 37, 701–708.

g stem modellin y s r 4 Rive

64 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

5 Uncertainty in surface water modelling results

This chapter describes the assessment of uncertainty in the surface water modelling results which was conducted to provide an independent comparison for the river modelling results. It has four sections:

• a summary of major issues and observations, and key messages and the uncertainty around the results • an overview of the approach • presentation of results, and • discussion of key findings.

5.1 Summary 5 Uncertaint

Using multiple lines of evidence an assessment was made of the uncertainty that is internal to the river model (as opposed to that associated with the climate scenarios), and the implications that this has for the confidence in the modelling results and their appropriate use. y nsraewtrmodellin water surface in 5.1.1 Issues and observations

• The density of the gauging network in the Eastern Mount Lofty Ranges is high compared to the average across the Murray-Darling Basin, reflecting the large number of unconnected streams in this region. However, there is a lack of long records and ratings are poor. End-of-system flows are not currently gauged. • The models of the Eastern Mount Lofty Ranges surface water system are relatively simple. They represent rainfall-runoff, relatively small in-stream losses to floodplains and wetlands, the storage behaviour of aggregated farms dams and diversions from these farm dams. End-of-system for several of the models is well upstream of the receiving waters. These models do not consider the (probably large) losses in the lower reaches.

• Because of the requirement for an upstream and a downstream gauge, independent water accounts could only g be constructed for one reach on the Bremer River. However, model performance was also assessed at one results location on the Finniss River.

5.1.2 Key messages

Based on the minimal assessments possible for the accounting period (1990 to 2006) the following conclusions can be made with respect to the river models for EMLR: • The internal uncertainty in the model appears to be small compared to the external uncertainty associated with future climate change. Model improvements therefore may not substantially reduce the uncertainty in decision making under changed climate. The best prospect is to reduce the uncertainty in climate projections. • The model is conceptually sound. It mainly includes well-measured processes and reproduces monthly and inter-annual flow patterns well. The model is considered robust for assessing relative changes in average flows. • The model overestimated total flow volumes by 13 percent in the Finniss River and by 22 percent in the Bremer River. Estimates of absolute current and future flow volumes are thus uncertain. • Uncertainty in model predictions associated with the possibility that groundwater influences future surface water processes in unexpected ways is generally considered low, although this may be an issue in Tookayerta Creek.

5.2 Approach

5.2.1 General

In Chapter 4, scenarios of climate and/or development were analysed using a river model to assess expected changes in water balance, flow patterns and consequent water security. Uncertainty in that analysis can be formulated in two parts:

• External uncertainty: uncertainty external to the model including uncertainty associated with the forcing data used in the model determined by processes outside the model such as climate, land use and water resource development. • Internal uncertainty: uncertainty in predictions that arises because the model is an imperfect representation of reality. May include uncertainty associated with the conceptual model, the implementing algorithms and software code, and application of the model to a specific region (Refsgaard and Henriksen, 2004).

This Chapter only addresses the internal uncertainty in the model. Fully quantifying uncertainty is impossible, and when

results scenarios take the model beyond circumstances that have been observed in the past, the quantifiable uncertainty may g only be a small part of the total uncertainty (Weiss, 2003; Bredehoeft, 2005). A quantitative analysis was therefore combined with a qualitative interpretation of the adequacy of the model for its purpose (similar to ‘model pedigree’, cf. Funtowicz and Ravetz, 1990; Van der Sluijs et al., 2005). Multiple lines of evidence were considered and synthesised to

modellin derive an overall assessment of confidence in model performance. These lines of evidence are: r • the quality of the hydrological observation network • the components of total estimated stream flow 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

in surface wate baseline river model simulations and from water accounting) y • measures of the baseline models performance in simulating observed stream flow patterns • the projected changes in flow pattern under the scenarios compared to the performance of the model in reproducing historic flow patterns.

None of these lines of evidence are conclusive in their own right. In particular: 5 Uncertaint • the model may be ‘right for the wrong reasons’, e.g. 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 pertinent if model boundary conditions are outside historically observed conditions, such as in climate change studies like this one.

Table 5-1. Possible framework for considering implications of assessed uncertainties

Low threat High threat Current water sharing arrangements Current water sharing arrangements are likely to appear sufficient for ongoing be inadequate for ongoing management of water management of water resources. resources, as they do not adequately consider

Low future threats. uncertainty

Current water sharing arrangements Current water sharing arrangements may be appear sufficient for ongoing inadequate for ongoing management of water management of water resources, but resources. Further work to reduce the major

High careful monitoring and adaptive sources of uncertainty can help guide changes management is recommended. to water sharing arrangements. uncertainty

66 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

With these limitations an attempt has been made to assess the likelihood that the river model provides realistic estimates of the changes that would occur under the scenarios evaluated. Qualitative model assessment is preferably done by expert elicitation (Refsgaard et al., 2006), but the time frame of the project did not allow this. Instead a tentative assessment was made that was adjusted following review by experts within and beyond the project as well as stakeholder representatives. Model uncertainty needs to be interpreted against the sum of internal and external uncertainty. The range of projections under different scenarios that are evaluated in this study provides an indication of the external uncertainty. Only where internal uncertainty clearly exceeds external uncertainty will model improvements help to reduce overall uncertainty. The implication of overall uncertainty for the use of the results presented in this study 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 considered to be a largely subjective assessment. A possible framework for users of the project results to consider the implications of the assessed uncertainties is shown in Table 5-1.

5.2.2 Information sources 5 Uncertaint

Information on the gauging network was obtained from the Water Resources Station Catalogue (www.bom.gov.au/hydro/wrsc) and from the South Australian Department of Water, Land and Biodiversity Conservation (www.dwlbc.sa.gov.au/subs/surface_water_archive/ ). Time series of water balance components as modelled under the

baseline scenario (Scenario A) and all other scenarios were derived as described in Chapter 4. The data used in water y accounting are described in the following section. modellin water surface in

5.2.3 Water balance accounting

Generic aspects of the water accounting methods are described in Chapter 1. This section includes a description of the basic purpose of the accounts which is to inform the uncertainty analysis carried out as part of this study 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 for wetland and irrigation water use, as well as the calculations for attribution of apparent ungauged gains and losses. Aspects of the methods that pertain specifically to the EMLR are

presented below. g results

Framework

The EMLR region is characterised by several short river systems that drain into the River Murray (Chapter 4). Only the Bremer River has more than one gauging station as required for water accounting. For the reach on the Bremer River between the two gauges water accounts were established for the period July 1990 to June 2006. The associated catchment area is shown in Figure 5-1. Model results for this reach were provided by DWLBC.

results g modellin r in surface wate y

5 Uncertaint Figure 5-1. Map showing the subcatchments used in modelling, with the reaches for which river water accounts were developed (‘accounting reach’) and contributing head water catchments with gauged inflows (‘contributing catchment’). Black dots and red lines are nodes and links in the river model respectively.

Diversion data

No diversion data were available. Irrigation development in the upland part of the region is small and mainly associated with hillside farm dams. More extensive irrigation occurs along the River Murray and this is considered in the river model for the Murray.

Wetland and irrigation water use

The result of the remote sensing analysis (see Chapter 1 for description of methods) is shown in Figure 5-1. Irrigation areas were identified in the lower modelled reaches but not in the reach for which water accounts were developed. No sizeable wetland areas were identified. Therefore no net irrigation or floodplain and wetland water use estimates were used in accounting.

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 1. In using SIMHYD estimates, it was noted that these are available for both the area of local runoff as well as the three contributing catchments (Figure 5-2). Therefore downward adjustment of SIMHYD estimates was expected.

68 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

5.2.4 Model uncertainty analysis

The river model results and water accounts were used to derive measures of model uncertainty. The different analyses are described below. In the interest of brevity details on the formulas used to calculate the indicators are not provided here but can be found in Van Dijk et al. (2007). Calculations were made for each reach separately but summary indicators were compared between reaches.

Completeness of hydrological observation network

Statistics on how well all estimated river gains and losses were gauged, or where not gauged how well they 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 to the grand totals of all

estimated inflows or gains, and/or all outflows or losses, respectively 5 Uncertaint • 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 graphically in Appendix C. y nsraewtrmodellin water surface in Comparison of modelled and accounted reach water balance

The river reach water balance terms, as modelled by the baseline river model (Scenario A), and as accounted, were compared for the period of water accounting. Neither necessarily represent reality, but large differences may indicate larger uncertainty in the model.

Climate range calibrated

If the period of model calibration is characterised by climate conditions that are a small subset, or atypical of the range of climate conditions that is historically observed, this probably increases the chance that the model will behave in g unexpected ways for climate conditions outside the calibration range. As an indicator of this, the percentage of the results overall climate variability range (i.e. annual rainfall) for the 111–year baseline simulation period that was covered by the extremes in the calibration period was calculated.

Performance of the river model in explaining historical flow patterns

All the indicators used in this analysis were based on the Nash-Sutcliffe model efficiency (NSME; Nash and Sutcliffe, 1970) indicator. 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, NSME values have been calculated for log- transformed and ranked flows, as well as for high (highest 10 percent) and low (lowest 10 percent) monthly flows. Where observed monthly flows include zero values, NSME cannot be calculated for the log-transformed flows. Likewise, if more than 10 percent of months have zero flow, NSME cannot be calculated for low flows.

Using NMSE, the efficiency of the water accounts in explaining observed outflows is calculated. This indicates the scope for improving the model to explain more of the observed variability. If NSME is much higher for the water accounts than for the model, this suggests that the model can be improved to reduce uncertainty. If the two are of similar magnitude, a better model is less likely to be possible without additional hydrological measurements.

Streamflow patterns at the end-of-reach gauge were compared visually to the flows predicted by the baseline river model and to 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). This 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. 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 one or less suggests that the projected scenario change is not significant when compared to river model uncertainty. A ratio that is considerably greater than one indicates that the future scenario model is much poorer at producing historic observations than the baseline model, suggesting that the scenario leads to significant changes in flow. The change-uncertainty ratio 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 of monthly and annual change-uncertainty ratios is made for each scenario.

5.3 Results results g 5.3.1 Density of the gauging network

Figure 5-2 shows the location of streamflow, rainfall, and evaporation gauges in the region. Table 5-2 provides modellin r information on the measurement network. The EMLR region has a relatively dense gauging network compared to the other regions in the Basin (ranked 3 out of 18). The rainfall, streamflow and evaporation gauging networks are all about three to four times denser than the Basin average. Figure 4.1 shows four additional gauges in the lower stream reaches . These are discontinued gauges with unrated records shorter than 2 years. There are however, deficiencies in the stream gauging network. More than half of all active streamflow gauges are in the Bremer River catchment and there are no gauges in Reedy Creek – a major subcatchment (Figure 5-2). There is no long-term monitoring of end-of-system flows. in surface wate y Whilst Table 5-2 records three evaporation stations, one of these replaces an earlier station at the same location so there have effectively only ever been two evaporation stations in the region at any one time.

Table 5-2. Some characteristics of the gauging network of the Eastern Mount Lofty Ranges region (4693 km2) compared with the entire Murray-Darling Basin (1,062,443 km2) 5 Uncertaint

Gauging network characteristics Eastern Mt Lofty Ranges Murray-Darling Basin No. No. per 1000 km2 No. No. per 1000 km2 Rainfall Total stations 99 21.09 6232 5.87 Stations active since 1990 57 12.15 3222 3.03 Average years of record 45 45 Streamflow Total stations 17 3.62 1090 1.03 Stations active since 1990 16 3.41 881 0.83 Average years of record 18 20 Evaporation Total stations 3 0.64 152 0.14 Stations active since 1990 1 0.03 104 0.10 Average years of record 18 27

70 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

5 Uncertaint y nsraewtrmodellin water surface in g results

Figure 5-2. Map showing the rainfall, streamflow and evaporation observation network, along with the subcatchments used in modelling

5.3.2 Review of model calibration and evaluation information

The EMLR region is modelled using six separate WaterCRESS models (see Chapter 4). WaterCRESS is a distributed rainfall-runoff model that includes explicit representation of farm dams. River regulation, groundwater interactions and diversions (other than from farm dams) are not represented. Reports on five model applications in the region were made available by DWLBC (Savadamuthu, 2002, 2003, 2004, 2006; Alcorn, 2006). A summary of the model reports is provided in Table 5-3.

Table 5-3. Details of calibration and validation periods, number of years between 1895–2006 (112 years) with annual rainfall less than the driest and more than the wettest year in the calibration period, respectively, and prior assessment of the models performance

Reach Downstream Area Number of Calibration/ Years Assessment gauging station (km2) validation* out of code and name period 112

Rainfall Stream drier wetter gauges flow gauges

42605290, Upper Marne 240 6 1 1975–1988 1 21 Very good reproduction of flows. 42610122 4266292, Upper Angas 190 4 1 1996–1999 27 43 Unreliable stream flow data. Reasonable to 4266293, good reproduction of flows. 42605030 2605040 Upper Finniss 193 5 1 1970–2000 1 0 Possible problems with high stream flow data. Good to excellent reproduction of flows, somewhat less for low flows. 42610200 Tookayerta 100 1 1 1997–2002 5 30 Little stream flow data available. Very good

results reproduction of flows. g 42605300 Currency Creek 89 5 1 1973–1992 1 0 High flow observations may be unreliable. Very good reproduction of flows, somewhat less for low flows and dry years. * Only partial validation considering performance for calibration period. Source: DWLBC modellin r

The rainfall-runoff models were parameterised using aerial surveys of land use and farm dams. Between 550 and 1250 farm dams were identified in each subcatchment, storing between 1.3 GL and 5.9 GL of water per subcatchment. Rainfall data used in the models included four privately owned gauges in the Upper Marne subcatchment. It was concluded in the model report that this provided a reasonable representation of spatial distribution, but that use of data from private sources introduced some uncertainty. in surface wate y The calibration period of the models was limited by the lack of long-term gauging records. At worst, there were only three years of unreliable stream gauging data available (Upper Angus subcatchment). At best, there were 30 years of streamflow data available (Upper Finniss Creek) although high flow observations there were considered unreliable. The full stream gauging period was used for calibration of the models to maximise the range of conditions included. A consequence of this was that no separate period was used to validate the calibration. The short monitoring period also 5 Uncertaint meant that conditions outside calibration conditions occur frequently for the three subcatchments. One third of the years in the 112-year modelling period have been wetter than the wettest year in the 4-year calibration period for the Angus River model.

Model performance for the calibration period ranges from good in the Upper Angus subcatchment, where the calibration period was shortest (NSME of 0.77 for annual flow and 0.75 for monthly flow), to excellent in Upper Finniss Creek where the calibration period was the longest and most representative (NSME of 0.90 for annual flow and 0.92 for monthly flow). It was concluded that model performance could be improved with additional gauging stations, recommissioning of stations to achieve longer records, and better monitoring design to gauge low flows and improve rating curves. Use of daily rather than monthly evaporation data in the models was also identified as a potential area for improvement. In Tookayerta Creek it was noted that the high flow contribution from groundwater suggests that improved measurement and modelling of groundwater hydrology is warranted in this catchment.

5.3.3 Model uncertainty analysis

Construction of water accounts requires an upstream and a downstream gauge along a river reach. Consequently for the EMLR, accounts could be established for only one reach of the Upper Bremer River. Streamflow data for the full accounting period (1990 to 2006) is available for one station on the Finniss River (426504). This data was used to evaluate performance of the Finniss River model. All indicators and results are listed by reach in Appendix C; a summary of the assessments is provided below.

72 ▪ Water availability in the Eastern Mount Lofty Ranges October 2007 © CSIRO 2007

Completeness of hydrological observation network

The Upper Bremer reach is a strongly gaining reach with main stem outflows estimated at 2.6 times the main stem inflows. An estimated 45 percent of all gains and 89 percent of all losses during the accounting period were gauged. Most of the total estimated gains (83 percent) could be attributed to processes assessed in the water accounts, leaving 17 percent of unattributed gains or error.

It was concluded that the hydrology of this reach is relatively straightforward and reasonably well understood. Because regulation and losses are small, the main challenge in this, and probably other river reaches in the region, is a correct estimation of catchment rainfall-runoff response.

Comparison of modelled and accounted reach water balance

A summary of the regional water balance simulated by the river model and derived by water accounting is shown in Table 5-4. In both cases, numbers are averages for the period 1990 to 2006. It is noted that there are no irrigation areas 5 Uncertaint or ephemeral or permanent wetlands, other than farm dams, in the accounting area. Neither the river model nor the water accounts include estimates of groundwater exchanges, as no data were available to constrain such estimates, and because groundwater discharges are implicit in local inflow estimates.

The following conclusions can be drawn from a comparison of the modelled and accounted water balance: y nsraewtrmodellin water surface in • 69 percent of the ungauged gains were explained by SIMHYD local inflow estimates (Appendix C). The remaining ungauged losses could not be attributed. They make up 17 percent of the total inflows, within the errors possible from gauging and runoff modelling. • The river model estimates of ungauged inflows are close to the sum of local inflows and unattributed gains in the water accounts. • The end-of-reach gauge suggests there are 2.0 GL/year of losses in the reach. These cannot be attributed to particular processes and could be noise in gauging. They represent 12 percent of the gauged outflow. • The modelled end-of-reach flows are 22 percent or 3.5 GL/year greater than observed stream flows. This is an overestimate of the water resource in the reach for the accounting period, especially considering that the model

includes current farm dam development but the gauging covers a longer historical period with less development g

on average. results

Table 5-4. Regional water balance modelled and estimated on the basis of water accounting

Water balance (Jul 1990 – Jun 2006) Model (A) Accounts Difference GL/y Total gains 19 18 1 Gauged inflows 6 8 -2 Ungauged inflows 13 7 6 Unattributed inflows 0 3 -3 Total losses 20 18 2 End-of-system outflows 19 16 4 Other gauged outflows 0 0 0 Diversions 0 0 0 River flux to groundwater 0 0 0 Other ungauged losses 0 0 0 Unattributed losses 0 2 -2

Performance of the river model in explaining historic flow patterns

The better the baseline model simulates streamflow patterns, the greater the likelihood that it represents the response of flows to changed climate, land use and regulation, notwithstanding the possibility that the model is right for the wrong reasons through compensating errors.

Appendix C gives NSME values of model performance. The model reproduced observed flow patterns very well in the Bremer (0.82 for monthly flows and 0.80 for annual flows) and very well to excellent at the Finniss River gauge (0.94 for monthly flows and 0.84 for annual flows). Low flows were significantly overestimated by the model in both cases. These findings corroborate the prior assessment that model performance was very good for the Upper Finniss River with the exception of low flows. Note also, that the model systematically overestimates total flow volumes over the accounting period in both rivers: by 13 percent in the Finniss River and 21 percent in the Bremer River.

Scenario change-uncertainty ratio

A high change-uncertainty ratio (CUR) corresponds with a scenario change in flows that is likely to be significant given the uncertainty or noise in the model, and a value close to 1.0 means that the modelled change is of similar magnitude to results g the uncertainty in the model. For both catchments considered, the changes under the Cdry scenario were strong (4.7–7.2, Appendix C) compared to the uncertainty in the model. Average stream flow was reduced by 56 to 70 percent under the Cdry scenarios. The difference in significance between C and equivalent D scenarios was very small in all cases. The changes projected under the B, Cdry and Cmid scenarios were weak (CUR of 0.2 to 1.5), even though changes in modellin r average flow ranged from 4 percent to 28 percent. This was because, as noted above, the baseline model overestimated flows over the accounting period, meaning that the reduced flow under future scenarios actually agreed better with the historical data.

5.4 Discussion of key findings in surface wate y 5.4.1 Completeness of the gauging network The hydrology of the EMLR surface water system is well gauged when compared to the rest of the Basin, although greater temporal and spatial resolution of rainfall data would reduce the uncertainty in the modelling.

While the stream flow gauging network is relatively dense, the relatively short monitoring periods and quality of the

5 Uncertaint streamflow gauging limit predictive capacity. This is exacerbated by the relatively large number of unconnected systems which require a large number of gauges to measure the majority of fluxes.

Water accounts could only be established for one strongly gaining reach; model performance could be assessed for one additional reach. No major diversions occur in the reaches considered in the uncertainty assessment.

5.4.2 Conceptual understanding of regional surface hydrology Despite the sparseness of the gauging network, the conceptual understanding of the hydrology of the EMLR surface water system is robust. The surface water system is characterised by multiple, relatively simple catchments dominated by rainfall-runoff response and runoff interception by farm dams. Floodplain and wetland losses are probably substantial in the lower reaches of some of the streams, but these reaches are not included in the river models.

Groundwater interactions were identified as a source of uncertainty for the Tookayerta Creek catchment in the western uplands, where it contributes considerably to streamflow. Groundwater discharge is implicitly described in the model through the base flow component, but it is unclear to what extent groundwater development may introduce uncertainty in river model projections, perhaps particularly in the lower reaches of the system (Chapter 6).

Surprises associated with river regulation and development are possible given the relatively high level of rural development. This includes the theoretical possibility of further farm dam development and other intercepting activities associated with land use and management and water resource capture or diversion (Chapter 3).

5.4.3 Performance and uncertainty in aspects of the river model Overall model performance appears to be adequate for assessments of rainfall-runoff changes due to farm dam development and climate change. However, no independent validation could be performed, and individual models have uncertainty due to short-term streamflow data of uncertain quality. However, the same conceptual models have been used in multiple catchments with apparently similar hydrological functioning. This suggests a degree of cross-validation between applications that creates greater confidence in the results.

The climate calibration range was excellent for some of the catchment models and poor to very poor for others (Table 5-3). Independent comparison with observed flow at two stations for 1990 to 2006 confirms prior assessments that the models reproduce monthly and annual flow sequences very well, but reproduce low flows less well, noting however, that this may also be related to unreliable gauging. Water accounting suggested that the model in each of the two catchments evaluated overestimated total flows by 13 percent and 22 percent.

Water accounting was performed for one reach of the Bremer River. A non-linear adjustment of SIMHYD estimates could 5 Uncertaint explain most of the ungauged gains as local inflows (for a more comprehensive evaluation of SIMHYD performance see Chapter 3). The water balance derived from accounting agreed well with the modelled water balance.

5.4.4 Implications for use of the results of this study

When compared to the uncertainty in the model, the changes in flow pattern predicted under the scenarios for two y nsraewtrmodellin water surface in stations were strong for the driest C and D scenario, but weak for the other scenarios. The latter was due to streamflow overestimation by the baseline model. This suggests that at least for these two stations, the modelling may not give a good estimate of future streamflow volumes or of current flow volumes. The small number of stations prevents extrapolation of these findings across the region, but the generally good historic model performance elsewhere suggests that probably relative changes and perhaps absolute changes are still likely to be predicted well.

Overall the hydrological system is considered relatively simple, the river models conceptually sound and the various indicators of model performance sufficient for the changes in average flows to be assessed with a reasonable degree of confidence. However, projections of changes in low flows are more uncertain. Arguably the greatest conceptual uncertainty is associated with the limited data available to test the adequacy of the model simulated impacts of current and future farm dam development. This uncertainty needs to be assessed considering the magnitude of estimated future g results farm dam development impacts relative to those associated with climate change.

The internal model uncertainty appears small when compared to the external uncertainty associated with uncertainty in climate change predictions. Future farm dam and commercial forestry development are projected to have small impacts relative to that of future climate (Chapter 3). Model improvements or better certainty over land use development may therefore not substantially reduce the uncertainty in decision making, but could increase certainty of climate predictions.

5.5 References

Alcorn M (2006) Surface Water Assessment of the Currency Creek Catchment, South Australia. Report DWLBC 2006/07, Knowledge and Information Division, Department of Water, Land and Biodiversity Conservation, Adelaide Bormann H (2005) Evaluation of hydrological models for scenario analyses: Signal-to-noise-ratio between scenario effects and model uncertainty. Advances in Geosciences 5, 43–48. Bredehoeft J (2005) The conceptual model problem—surprise, Hydrogeology Journal 13,37–46. Funtowicz SO and Ravetz J (1990) Uncertainty and Quality in Science for Policy. Kluwer Academic Publishers, Dordrecht. Kirby J, Mainuddin M, Podger G and Zhang L (2006a) Basin water use accounting method with application to the Mekong Basin. In: S. Sethaputra and K. Promma (Eds.), Proceedings on the International Symposium on Managing Water Supply for Growing Demand, Bangkok, Thailand. Jakarta. UNESCO, pp 67–77. Kirby J, Mainuddin M, Mobin-ud-Din A, Marchand P and Zhang L (2006b) Water use account spreadsheets with examples of some major river basins. In: Proceeding of the 9th International River Symposium. 4-7 September 2006, Brisbane. Nash JE and Sutcliffe JV (1970) River flow forecasting through conceptual models, 1: a discussion of principles. Journal of Hydrology 10, 282–290. McVicar TR and Jupp DLB (2002) Using covariates to spatially interpolate moisture availability in the Murray-Darling Basin. Remote Sensing of Environment 79, 199–212. Pappenberger F and Beven KJ (2006) Ignorance is bliss: Or seven reasons not to use uncertainty analysis. Water Resources Research 42, W05302, doi 10.1029/2005WR004820. Refsgaard JC and Henriksen HJ (2004) Modelling guidelines–terminology and guiding principles. Advances in Water Resources 27, 71– 82.

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, 1586–1597. Savadamuthu K (2002) Impact of farm dams on streamflow in the Upper Marne. Report DWR 02/01/0003, South Australia Department for Water Resources, Adelaide. Savadamuthu K (2003) Surface Water Assessment of the Upper Finniss Catchment, South Australia. Report DWLBC 2003/18, Department of Water, Land and Biodiversity Conservation, Adelaide. Savadamuthu K (2004) Surface Water Assessment of the Tookayerta Catchment, South Australia. Report DWLBC 2004/23, Department of Water, Land and Biodiversity Conservation, Adelaide. Savadamuthu K (2006) Surface Water Assessment of the Upper Angas Sub-catchment, South Australia. Report DWLBC 2006/09, Department of Water, Land and Biodiversity Conservation, Adelaide. 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, 481–492. Van Dijk AIJM (2006) 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. Van Dijk AIJM and Mattersdorf G (2007) Comparison of MODIS-based scaling of potential evapotranspiration with on-ground observations. Geophysical Research Abstracts 9, 11692. SRef-ID: 1607-7962/gra/EGU2007-A-11692. Van Dijk AIJM et al. (2007) River model uncertainty assessment. A report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project. CSIRO, Australia. In prep. Walker WE, Harremoës O, Rotmans J, van der Sluijs JP, van Asselt MBA, Janssen P and Krayer von Krauss MP (2003) Defining uncertainty a conceptual basis for uncertainty management in model-based decision support. Integrated Assessment 4, 5-17. results g Weiss C (2003) Expressing scientific uncertainty. Law, Probability and Risk 2, 25-46. modellin r in surface wate y 5 Uncertaint

6 Groundwater assessment

This chapter describes the groundwater assessments for the Eastern Mount Lofty Ranges region. It has seven sections:

• a summary of major issues and observations, key messages and uncertainty

• a description of the groundwater management units in the region

• a summary of the hydrogeological context

• a description of the trends in groundwater levels and salinity

• a description of surface water-groundwater connectivity Gonwtrassessment 6 Groundwater • assessment of water balances for lower priority groundwater management units, and

• a discussion of key findings.

6.1 Summary

6.1.1 Issues and observations • Groundwater use in the Eastern Mount Lofty (EMLR) region in 2004/5 was 18.7 GL/year, which was 1.3 percent of the total groundwater use in the MDB (not including the confined GAB aquifers, unincorporated areas and stock and domestic). Most of this (15.6 GL/year) occurred in the Eastern Mount Lofty Ranges (EMLR) groundwater management unit (GMU).

• The extraction to rainfall recharge (E/R) ratio was used as the main indicator for this region. This ratio provides an indicator of the level of development: 0–0.3 low, 0.3–0.7 medium, 0.7–1.0 high, >1.0 very high. There are other forms of recharge besides rainfall (diffuse dryland) recharge e.g. stream recharge may be a significant component of recharge for this reporting region but difficult to quantify. The E/R is not relevant to the Angas Bremer GMU where most extraction occurs from the confined aquifer. High to very high development would require very good groundwater information to ensure extraction is not adversely impacting on the groundwater resource or ecosystems that are dependent on groundwater.

• Recharge scaling factors as estimated by a soil-vegetation-climate model (WAVES) is used to modify rainfall recharge estimates for the different scenarios. These new values are then used to estimate E/R.

6.1.2 Key Messages • The current extraction is one quarter of rainfall recharge in the EMLR GMU (E/R=0.25) and two-thirds of rainfall recharge in the Marne Saunders GMU (E/R=0.67); for the Marne Saunders GMU rainfall recharge may be as low as 40 percent of the total recharge. Groundwater development is therefore considered low in the EMLR GMU and low-moderate in the Marne Saunders GMU.

• The main concern for groundwater at current extraction levels is increasing groundwater salinity in the Angas Bremer GMU and the Currency Creek region.

• The reduction in stream flow due to groundwater extraction is estimated to be nearly 7.0 GL/year.

• Under a continuation of the recent (last ten years) climate E/R would increase to 0.33 for the EMLR GMU and 0.79 for the Marne Saunders GMU. Under the best estimate 2030 climate E/R would be 0.26 for the EMLR GMU; ranging from 0.22 (wet extreme 2030 climate) to 0.39 (dry extreme 2030 climate). For the Marne Saunders GMU the 2030 climate values are 0.69 (best estimate), 0.59 (wet extreme) and 1.01 (dry extreme). A continuation of the recent climate or the dry extreme of the 2030 climate range would effectively raise the level of groundwater developed to moderate for the EMLR GMU and to high to very high for the Marne Saunders GMU, again noting that flood recharge will reduce E/R values.

• Without intervention, groundwater extraction from all groundwater management units is likely to increase in the future. For the best estimates of future extraction (a 60 percent increase) and 2030 climate E/R increases to 0.42 for the EMLR GMU (range: 0.36–0.63) and to 1.03 for the Marne Saunders GMU (range: 0.88–1.52). This degree of growth in groundwater extraction would therefore raise the level of groundwater development to moderate for the EMLR GMU and to high to very high for the Marne Saunders GMU. Even allowing for possible flood recharge, the level of development under Scenario D for the Marne Saunders would require further groundwater analysis to support management.

• This level of increase in extraction would reduce streamflow by an about additional 4 GL/year – a total average impact of groundwater development of about 11 GL/year. Current groundwater allocation planning may however, prevent these levels of extraction being realised.

6.1.3 Uncertainty A simple water balance approach was used for this region. This is appropriate given the low priority rating of these GMUs in the context of the overall project. However, it is noted that this would not be appropriate for addressing local groundwater management issues. The importance of flood recharge in the Angas Bremer and Marne Saunders GMU implies that a more complex analysis may be warranted to evaluate the implications of variations in flood recharge under each of the climate change scenarios. The estimated impacts of groundwater extraction on stream flow are considered to assessment r have a low level of confidence.

6.2 Groundwater management units in EMLR The EMLR region contains three groundwater management units (GMUs): Eastern Mount Lofty Ranges, Angas Bremer and Marne Saunders. These GMUs have been assigned a low priority ranking in the context of the overall MDB Water

6 Groundwate Assessment project (Table 6-1). The priority ranking takes into account level of development, size of the available resource and degree of connectivity between rivers and aquifers relative to other GMUs across the MDB. The priority ranking provides a basis for focussing effort to those aquifers affecting most the total water resource across the MDB.

A simple water balance approach is used in this region. This is consistent with the low priority ranking. It is noted that while this limited assessments is appropriate with the constraints and for the terms of reference of this project, additional work may be required for local management of groundwater resources. The approach includes a background description of the hydrogeology (Chapters 2 and 6) and an evaluation of the impact of changing rainfall recharge and extraction under each of the scenarios (Chapter 6).

More specifically, the following form the groundwater assessment for the EMLR region:

• Survey of trends in groundwater levels and salinity,

• Description of surface-groundwater connectivity, and

• Estimation of water balances under different scenarios.

The key outputs developed are:

• Extraction to rainfall recharge ratio (E/R) for different scenarios, and

• Impact of groundwater extraction for different scenarios.

The scenarios being analysed are current (2004/5) groundwater extraction under current climate (A), climate indicative for the last ten years (B), climate projections for 2030 (Cdry, Cmid and Cwet) and estimated future groundwater extractions under future climate and land use scenarios (Ddry, Dmid and Dwet). For each of these scenarios, rainfall recharge is estimated by multiplying the rainfall recharge under current climate by a recharge scaling factor (RSF) appropriate for the climate scenario. This RSF is calculated using a one-dimensional Soil-Vegetation Atmosphere water transfer model (WAVES).

Table 6-1. Description of GMUs, including annual extraction, entitlement and recharge details

Code Priority GMU Current Total Long-term average Recharge from Estimated 2030 ranking groundwater entitlement extraction limit rainfall groundwater extraction extraction GL/y GL GL/y S14 Low Eastern Mount Lofty Ranges 15.6 46* 46* 61.4 25.0 S18 Low Angas Bremer 1.2 6.5 6.5 *** 3.0 S23 Low Marne Saunders 2.0 4** 4** 3 3.0 Unincorporated area (east of Not available Not Not licensed Marne Saunders) licensed *A water allocation plan for the EMLR GMU is currently being prepared. DWLBC recommended interim allocation; to be confirmed once environmental investigations are complete. **A water allocation plan for the Marne Saunders is currently being prepared. Annual allocation volume is an interim value.

*** The confined aquifer receives minimal recharge from rainfall or stream flow. assessment 6 Groundwater Angas Bremer and Marne Saunders extraction values are for the 2004/05 water use year. The land use survey for the EMLR GMU was completed in 2002/03.

Figure 6-1. Map of groundwater management units within the Eastern Mount Lofty Ranges region

6.3 Hydrogeological context The hydrogeological context is described in Chapter 2. Some of the salient points are briefly revisited.

The EMLR region is topographically separated into a highland (Hills) zone and a lowland (Plains) zone. The Hills zone is coincident with fractured basement rock aquifers of the Mount Lofty Ranges and the Plains Zone with sedimentary aquifers of the Murray Basin. Groundwater is also contained within the alluvial fill within highland valleys.

The fractured rock aquifers in the highland region consist of a number of different geological units which include the Barossa Complex, the Kanmantoo Group, the Adelaidean, and the Normanville Group. Of these, the latter two are more important as a resource.

Unconsolidated sedimentary aquifers of the Plains include the Renmark Group which is not generally considered as a productive aquifer due to the presence of better quality water in the overlying Murray Group Limestone. This unit is highly fossiliferous and of sandy limestone containing groundwater of varying salinity. Significant recharge occurred to this assessment r aquifer in the Marne Saunders and Angas Bremer GMUs several thousand years ago during a much wetter climate and now supplies groundwater for irrigation, stock and domestic uses. However, there is very little current recharge in the Angas Bremer GMU with current extractions gradually depleting the lower salinity groundwater resource and resulting in salinity increases (Zulfic and Barnett, 2007). The Marne Saunders GMU is recharged through rainfall infiltration and flood recharge.

In the Hills zone, groundwater moves from the higher points in the landscape to the lowest where discharge occurs along 6 Groundwate streams, providing baseflow in this zone. Groundwater also feeds permanent pools along main drainage lines. Where the streams flow out of the hills onto the plains, the streams change from gaining to losing and begin to recharge the underlying sedimentary aquifers. This form of recharge may be locally and seasonally important. For example recharge during floods in the Marne Saunders GMU is the main source of recharge to the productive aquifer (Barnett et al., 2001).

6.4 Trends in groundwater levels and salinity Groundwater levels within the fractured rock aquifers of the EMLR are largely dependent on rainfall with observed groundwater levels at their lowest point in 2004 due to below average rainfall, Figure 6-2: Adelaidean and Barossa Complex). In some areas (e.g. Mt Barker, Figure 6-3) groundwater levels have recovered due to the redevelopment of irrigated land to urban uses.

418

) 416 414

412

RSWL (mAHD RSWL 410 408 1984 1990 1995 2001 2006

Figure 6-2. Hydrograph for bore MOR220 showing a declining trend in groundwater level with the lowest point in 2004 associated with below average rainfall

313

) 312 311 310 309

RSWL (m AHD RSWL 308 307 1979 1984 1990 1995 2001 2006

Figure 6-3. Hydrograph for bore MCF009 showing a declining trend in groundwater level throughout the 1990s with a rising trend after

2000 assessment 6 Groundwater

Groundwater levels within the Permian Sands aquifer shows seasonal drawdowns, but there are no adverse trends apparent at current levels of extraction (Figure 6-4).

248

) 244 240

236 232 RSWL (m AHD RSWL 228 1979 1984 1990 1995 2001 2006

Figure 6-4. Hydrograph for bore NGK006 displaying neither a rising nor falling trend

Analysis of all monitoring wells in the Angas Bremer GMU is currently being undertaken by DWLBC toward the development of the new Water Allocation Plan (Zulfic and Barnett, 2007). This work indicates that there are three distinct groups of groundwater level trends in the Angas Bremer GMU:

• Groundwater levels in wells located south of Langhorne Creek and west of Bremer River generally declined until 1984, were stable between 1984 and 1992, increased sharply in response to high Spring rainfall in 1992, declined between 1993 and 1995, increased between 1995 and 2005, and have remained stable since 2005.

• Groundwater levels east of Bremer River were generally stable (with seasonal variation) until 1989, and have been increasing ever since.

• Groundwater levels north of Langhorne Creek on the western side of Bremer River decreased until 1992 and have increased since 1993 with groundwater levels becoming more stable in the recent years.

The salinity of groundwater in the confined Murray Group Limestone aquifer is increasing in the Angas Bremer and the Currency Creek regions. A plot of levels and salinity for the Langhorne Creek town water supply is provided in Figure 6-5.

7 1500 BRM007 RSWL TDS Tow n Water Supply 1450 6 1400 5 1350 4 1300 1250 3 1200 1150

RSWLAHD) (m 2 1100 1 1050 Total Dissolved Solidsmg/L 0 1000 1968 1976 1984 1993 2001 2009

Figure 6-5. Groundwater levels and salinity variations for the Langhorne Creek town water supply over the last 15 years

The rising trend in salinity persists despite the significant recharge following the floods in 1992 and the reduction in groundwater extraction of the past 15 years. It is assumed that the salinisation of the confined aquifer is influenced by assessment r downward leakage of saline groundwater from the overlying unconfined aquifer and lateral migration of saline groundwater from up-gradient sources. Monitoring indicates the area of low salinity groundwater in the Angas Bremer is becoming smaller as extraction continues.

There is a similar issue in the Currency Creek area where the salinity of groundwater within the Murray Group Limestone aquifer has increased at a rate of up to 50 mg/L/year (Barnett, 2007). 6 Groundwate 6.5 Surface water-groundwater connectivity Surface water-groundwater connectivity within the EMLR region has not been mapped within this assessment; however, sufficient information is available to make some general statements concerning the interaction between surface water and groundwater largely based on observed discharge/recharge relationships.

In the Hills Zone groundwater moves from the higher points in the landscape to the lowest where discharge occurs along streams. This discharge acts as the baseflow to streams in this zone and can occur throughout the year, even during summer and dry periods. Groundwater also feeds permanent pools along main drainage lines. Zulfic and Barnett (2003) provide the baseflow values listed in Table 6-2. Barnett (DWLBC, pers. comm.) has estimated the base flow in the Marne to be 1.86 GL/y (base flow index of 0.23).

Table 6-2. Subcatchment partitioning of streamflow into run-off and baseflow for selected catchments

Subcatchments Run-off Baseflow GL/y Bremer River (AW 426533)* 10.04 7.16 Angas River (AW 426503)* 2.79 1.52 Finniss River (AW 426504)* 14.92 9.56 * Run-off and baseflow values from Zulfic and Barnett (2003).

Where the streams flow out of the hills onto the plains, they change from gaining to losing and begin to recharge the underlying sedimentary aquifers. This form of recharge may be locally and seasonally important. For example recharge during floods in the Marne Saunders GMU is thought to be a significant source of recharge to the productive aquifer.

6.6 Water balances for lower priority GMUs

6.6.1 Groundwater extraction and recharge Current groundwater extraction data reported for the EMLR reporting area are based on metering in the Angas Bremer and Marne Saunders GMUs and on land use survey in the EMLR GMU (Table 6-1). This land use survey/crop water requirement method used in the EMLR GMU is useful as a means of estimating the rate of groundwater use but does have a tendency to overestimate extraction.

Rate of future extraction has been estimated in consultation with DWLBC based on the current understanding of demand and capacity of the aquifer to meet demand. The values of future extraction can be considered ‘best-guesses’ only, and current groundwater allocation planning may prevent these levels of extraction being realised. For the purposes of this assessment future groundwater extraction in the EMLR GMU is assumed to 25 GL/year (Table 6-1) which is less than

the recommended (interim) allocation on the basis that elevated salinity and limited bore yield may limit the rate of assessment 6 Groundwater extraction in some areas.

The rate of groundwater extraction in the Angas Bremer GMU (estimated at 3 GL/year) is not expected to reach current allocation because of salinity impacts associated with extraction. DWLBC are working to re-estimate the extraction limit for the Angas Bremer GMU. The estimate of future extraction rate for Marne Saunders is set at 3 GL/year which is less than the recommended allocation.

The estimates of rainfall recharge presented in this section were provided by DWLBC (2007). Rates of rainfall infiltration measured by analysis of chloro-fluoro-carbons (CFCs), carbon-14, chloride data and stable isotopes of water were extrapolated across all management zones by DWLBC to generate an estimate of the annual volume of rainfall recharge across the EMLR reporting region.

The recharge rates have been converted to annual volumes for each management zone across the EMLR by DWLBC (summarised in Table 6-3). These calculations include areas where the salinity of groundwater may exceed the current beneficial use, mostly within areas of the Kanmantoo Formation.

Table 6-3. Estimated rainfall recharge for subcatchments within the EMLR region

Subcatchments Geology mm/y GL/y Eastern Mount Lofty Ranges Northern Murray Group Limestone - unconfined 1 0.3 Kanmantoo - Fractured rock 16 9.7 Bremer Adelaidean 50 6.2 Kanmantoo Group 21 9.8 Currency Creek Permian Sands 82 0.5 Kanmantoo Group 27 2.6 Tookayerta Creek Permian Sands 65 8.6 Kanmantoo Group 25 1.2 Angas Kanmantoo Group 21 5.4 Adelaidean 59 2.4 Finniss Permian Sands 57 3.5 assessment r Kanmantoo Group 25 4.2 Barossa Complex 97 6.9 Angas Bremer Hills zone Minor Plains Zone Minor

6 Groundwate Marne Saunders Hills Zone Kanmantoo Group 15 3.0 Plains Zone Plains Minor Total 64.4 Source: DWLBC, 2007; Cresswell and Gibson, 2004; and Barnett et al., 2001

Cresswell and Gibson (2004) estimated recharge to the Murray Group Limestone aquifer in the Angas Bremer GMU to be 2 GL/year based on an analysis of flood events over 25 years and the definition of the extent of recharge zones along the creek by interpretation of airborne radiometric surveys. Additional work undertaken by DWLBC since Cresswell and Gibson (2004) indicated that while there is a component of recharge during flood events it is likely to be a small volume and insufficient to sustain current extraction (Zulfic and Barnett, 2007). This conclusion has been drawn in part by work by Cresswell and Herczeg (2004) who used carbon-14 dating to infer that vertical recharge across the aquitard into the confined aquifer occurred slowly at time scales of a few hundred to a few thousand years. However, there appears to be some uncertainty regarding the importance of flood recharge with suggestion that recharge may occur where the confining layer has been eroded near streams.

Recharge during flood events in the Marne Saunders is more significant than in the Angas Bremer (estimated by Barnett et al. (2001) to be 4 GL/year) and it is likely that the magnitude of flood recharge would change across the scenarios considered in this assessment. However, variations in flood recharge have not been modelled. The dependence of flood recharge on climate change has been modelled for other reporting regions where both good groundwater models and good flow records exist. The dependence is sensitive to flood recharge versus in-channel recharge, climate, rules for diversion as water availability changes and conductance between extraction points and the river (including low conductance unsaturated zones).

6.6.2 Recharge modelling The simple water balance approach requires the application of recharge scaling factors (RSFs). Values of diffuse dryland recharge as estimated in the previous section underpin management of the GMUs within the Eastern Mt Lofty Region. The RSFs are used to multiply these values to provide estimates of dryland recharge under different climate scenarios to

be used in further analyses. For Scenario A, the RSF is necessarily 1.0, and for other climate scenarios would be expected to be approximately 1.0. The precedent of impacts of climate being reported as percentage change as followed. The RSFs can be obtained by dividing the percentage change by 100 and adding to 1.0.

For the purposes of further groundwater analyses, only the three chosen scenarios (Cdry, Cmid and Cwet) are required.

These scenarios, which represent a combination of Global Climate Model (GCM) and CO2 emission scenarios, were chosen on the basis of ranking in surface water run-off in order to reflect the range of predictions in surface run-off. However, it should be recognised that recharge is not necessarily perfectly correlated with surface run-off or mean rainfall. 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. For the semi-arid to sub-humid parts of catchments, the extreme events become more important. A number of GCMs show increase in extreme events. Some caution is required in outputs dependent on this aspect of the GCMs, as the main scenarios are chosen on the basis of mean runoff, which is more dependent on average and seasonal rainfall. Gonwtrassessment 6 Groundwater Recharge will also depend on the land use and soils. These can be locally variable, local spatial variation in RSFs. An estimate provided for small areas e.g. small GMUs will be sensitive to these local variations while larger areas will have a broader range of soils and land uses, which will lead to more robust estimates.

To better understand the assumption of choosing the scenarios on the basis of run-off, RSFs were estimated for all 15 global climate models under the three emission scenarios. In all cases, a one dimensional soil-vegetation-atmosphere water transfer model (WAVES: Zhang and Dawes, 1998) has been used for selected points around the MDB for combinations of soils and vegetation. Spatial data on climate, vegetation and soils were then used to interpolate values to regions.

Table 6-4 shows the percentage change in the modelled mean annual recharge averaged over EMLR region for scenario C (2030 climate) relative to scenario A for the 45 scenarios (15 global climate models for each of the high, medium and low global warming scenarios). The percentage change in the mean annual recharge and the percentage change in mean annual rainfall from the corresponding global climate models are also tabulated in Table 6-4. The plots show that there is a wide variability between the global climate models and scenarios regarding climate change in the EMLR with over 60 percent of the scenarios predicting less recharge and the rest predicting more recharge. It is the high global warming scenario that predicts both the highest and lowest change in recharge for the EMLR.

60% High global warming 40% Medium global warming 20% Low global warming

-20%

-40%

-60% iap mri ipsl gfdl mpi csiro cnrm miub miroc inmcm ncar_pcm giss_aom cccma_t47 cccma_t63 ncar_ccsm

Figure 6-6. Percentage change in mean annual recharge from the 45 Scenario C simulations relative to Scenario A recharge

Table 6-4. Summary results from the 45 Scenario C simulations. Numbers show percentage change in mean annual rainfall and recharge under Scenario C relative to Scenario A. Those in bold type have been selected for further modelling.

High global warming Medium global warming Low global warming

GCM Rainfall Recharge GCM Rainfall Recharge GCM Rainfall Recharge

Ipsl -19% -36% giss_aom -13% -25% giss_aom -6% -11%

giss_aom -20% -34% ipsl -12% -25% Ipsl -5% -11%

Gfdl -18% -19% gfdl -12% -14% Gfdl -5% -6%

Inmcm -10% -17% inmcm -7% -12% Csiro -4% -5%

Csiro -14% -14% csiro -9% -11% inmcm -3% -5%

Cnrm -14% -12% cnrm -9% -11% Cnrm -4% -4%

Miub -8% -7% iap -3% -6% Iap -1% -2%

Iap -4% -7% miub -5% -5% Mri -3% -2%

ncar_ccsm -4% 0% ncar_ccsm -2% -1% Miub -2% -1%

Mri -10% 1% mri -6% -1% ncar_ccsm -1% 0%

ncar_pcm 2% 7% ncar_pcm 1% 5% ncar_pcm 1% 2% assessment r cccma_t47 -2% 11% cccma_t47 -1% 6% cccma_t63 0% 3%

cccma_t63 1% 13% mpi -2% 6% cccma_t47 -1% 3%

Miroc 1% 13% cccma_t63 1% 7% miroc 0% 4%

Mpi -4% 14% miroc 1% 8% Mpi -1% 13%

6 Groundwate

The ‘dry’, ‘mid’ and ‘wet’ scenario C variants as chosen on the basis of surface water modelling are shown in Table 6-4 in bold type. The choice of global climate models for surface run-off 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 run-off being respectively the highest, third lowest and the median for RSF. The large variability in RSFs is related to the large variability in rainfall produced by the various global climate models. Rainfall and RSFs are correlated, although not perfectly. Also, some global climate models with decreases in rainfall lead to RSFs greater than 1.0. This results from the more extreme events as captured by the GCMs being more frequent while the mean rainfall decreases.

The scenarios for further analysis for the EMLR and Marne Saunders GMUs are shown in Table 6-5. The RSFs are calculated by dividing the values in Table 6-5 by 100 and adding 1.0. The RSF for the last ten years of Scenario A was also estimated relative to the long-term average to be -24%, not dissimilar to the Scenario B run.

Table 6-5. Summary results of the scenarios for modelling for each GMU in the EMLR region. Numbers show percentage change in mean annual recharge under scenario relative to Scenario A.

GMU B Cdry Cmid Cwet

Percent change relative to Scenario A

Eastern Mount Lofty Ranges -23% -35% -3% 14%

Marne Saunders -16% -34% -3% 14%

6.6.3 Current and future groundwater extraction versus rainfall recharge When applied the rainfall recharge scaling factors calculated using values in Table 6-5 produced the scaled recharge values for scenarios A, B and C listed in Table 6-6. It can be seen that under Scenario B, there is 14.6 GL/y less recharge. By comparison, over the last ten years, there has been 15.4 GL/y less recharge.

Table 6-6. Scaled recharge under scenarios A, B and C

GMU Rainfall Recharge Scaled Rainfall Recharge

A B Cdry Cmid Cwet GL/y Eastern Mount Lofty Ranges 61.38 47.26 39.90 59.54 69.97 Gonwtrassessment 6 Groundwater Marne Saunders 3.00 2.52 1.98 2.91 3.42

The ratio of current groundwater extraction to recharge is shown in Table 6-7. The ratio of extraction to rainfall recharge (E/R) is used as an indicator of the potential level of stress within the aquifer. Where the ratio is greater than 1.0 the groundwater resources of the GMU are being extracted at a rate greater than rainfall recharge is replenishing the groundwater. For the purposes of this report, levels of development are defined as: low, E/R 0–0.3; medium, E/R 0.3– 0.7; high, E/R 0.7–1.0 and very high, E/R>1.0. It should be noted that stream recharge forms an important additional form of recharge in this region.

Table 6-7. Comparison of current and projected future groundwater extractions with scaled rainfall recharge

GMU E/R Scaled E/R A B Cdry Cmid Cwet Ddry Dmid Dwet Eastern Mount Lofty Ranges 0.25 0.33 0.39 0.26 0.22 0.63 0.42 0.36 Marne Saunders 0.67 0.79 1.01 0.69 0.58 1.52 1.03 0.88

The comparison of extraction to rainfall recharge for each of the scenarios for the EMLR suggests a low level of development. Under the climate scenarios, this remains at a low level, except under the driest scenario, where the level of development becomes moderate. The impact of a dry climate may be estimated to be greater if the volume of recharge to the areas of low salinity groundwater is used rather than the volume of rainfall recharge estimated using the whole of the catchment area.

The ratio of extraction to rainfall recharge is greater for the Marne Saunders; however, this simple calculation becomes less useful where other recharge mechanisms such as flood recharge become important. For example, in the Marne Saunders there is estimated flood recharge of 4.0 GL/year which means the ratio of extraction to recharge (rainfall and flood recharge) will be lower for each scenario when taking the flood recharge component into account. The magnitude of flood recharge will be dependent on climate so this component of recharge is unlikely to remain constant with climate change and may not always sustain extraction when rainfall recharge diminishes. The effect of the climate change scenarios on the volume of flood recharge has not been assessed at this stage in the project.

The scenarios of predicted future development (at 2030) coupled with the impact of climate change demonstrate that the future extraction within the EMLR GMU could be as high as 63 percent of rainfall recharge under a dry climate scenario. Future extraction within the Marne Saunders is estimated to reach 152 percent of rainfall recharge under the driest climate change scenario. Even accounting for flood recharge, this would require further groundwater analysis to support.

6.6.4 Impact of extraction on stream flow The reduction in aquifer water levels due to groundwater extraction can impact stream flow. In the case of a gaining river, the rate of groundwater discharge to the river decreases. In the case of a losing river, a decline in groundwater level will increase leakage from the river into the underlying aquifer. Both processes will reduce flow in the river.

If groundwater discharge also occurs in other forms, such as evapotranspiration (ET) or groundwater through-flow, the reduction in stream flow will be less than the extracted volume.

There is little information regarding the degree of surface water-groundwater connectivity across the reporting region, but initial estimates are provided in Table 6-8 based on an understanding of depth to groundwater in the vicinity of rivers and streams, topography and geology. There is a low level of confidence attached to the estimated impacts of groundwater extraction on stream flow. All connectivity estimates are ‘best guesses’ and more detailed modelling (beyond the scope of this assessment) is required to better quantify impacts of groundwater extraction.

The following principles were used to guide the estimation of connectivity. Connectivity is expressed as the fraction of groundwater extracted that is derived from stream flow (either as induced leakage or captured discharge):

• The areas where rivers flow across Plains region have been assigned zero connectivity on the basis that generally the groundwater levels lie below the base of the river channel which means that groundwater

assessment extraction cannot impact stream flow. r • The Adelaidian regions are generally situated across the upper part of the catchments where there is lower topographic relief and where groundwater discharge occurs as through-flow and ET. These areas have been assigned a low level of connectivity (0.2). • The Kanmantoo regions are generally higher topographic relief areas where groundwater discharge to streams is a more significant component of the groundwater balance. These areas have been assigned a moderate level

6 Groundwate of connectivity (0.4). • The Permian regions generally are areas with higher topographic relief with high permeability sediments and where groundwater discharge to streams is a more significant component of the groundwater balance. These areas have been assigned a high level of connectivity (0.8).

The full impact of current extraction on stream flow is estimated to be 7.0 GL/year, once groundwater equilibrium is reached. This is estimated to increase to around 11.2 GL/year with the assumed rates of future groundwater extraction. The time lags will vary with the different geologies with Permian, equilibrium is expected to be attained quickly while for Adelaidean it may be as much as 50 years from beginning of extraction to full impact being realised.

Almost all estimated impacts are expected to occur within the EMLR GMU since large reaches of the rivers in the Marne Saunders and Angas Bremer GMUs have a low degree of connection to the productive aquifers. The future impact on Tookayerta is 19.5% of stream inflows, Angas 8.8%, Bremer 8.2%, Marne 7.4% and Finniss 6.7%. The impact of extraction on flows that reach the River Murray and Lake Alexandrina will be buffered by the losses of water that occur across the plains.

Table 6-8. Surface water-groundwater connectivity showing an estimate of the volumetric impact extraction has on stream flow

Region Connectivity Current extraction Current impact Future extraction Future impact GL/y Northern Kanmantoo 0.4 0.6 0.2 0.9 0.4 Northern Limestone 0 0.04 0 0.06 0 Bremer Adelaidian 0.2 3.3 0.7 5.4 1.1 Bremer Kanmantoo 0.4 1.1 0.4 1.7 0.7 Currency Permian 0.8 0.1 0.09 0.2 0.1 Currency Kanmantoo 0.4 0.3 0.1 0.46 0.2 Tookayerta Permian 0.8 2.6 2.1 4.2 3.4 Tookayerta Kanmantoo

0.4 0.03 0.01 0.05 0.02 assessment 6 Groundwater Angas Adelaidian 0.2 0.6 0.1 0.9 0.2 Angas Kanmantoo 0.4 2.0 0.8 3.3 1.3 Finniss Adelaidian 0.2 3.1 0.6 5.0 1.0 Finniss Kanmantoo 0.4 0.3 0.1 0.5 0.2 Finniss Permian 0.8 1.5 1.2 2.4 1.9 Marne Saunders (Hills) 0.5 1.0 0.5 1.60 0.8 Marne Saunders (Plains) 0 1.0 0 1.40 0 Angas Bremer 0 1.2 0 3.0 0 Total 18.77 7.0 31.07 11.2

6.7 Discussion Groundwater use in the EMLR region in 2004/5 was 18.7 GL/year – 1.3 percent of the total groundwater use in the MDB (not including the confined GAB aquifers). Most of this (15.6 GL/year) occurred in the EMLR GMU.

6.7.1 Application of the extraction to rainfall recharge (E/R) ratio E/R was used as the main indicator for this region. There are other forms of recharge besides rainfall (diffuse dryland) recharge e.g. stream recharge may be a significant component of recharge for this reporting region but difficult to quantify. E/R is not relevant to the Angas Bremer GMU where most extraction occurs from the confined aquifer. The majority of recharge in the Marne Saunders GMU is thought to be flood recharge.

E/R indicates the level of development: low, 0–0.3; medium, 0.3–0.7; high, 0.7–1.0; very high, >1.0 . High to very high development would require very good groundwater information to ensure extraction is not adversely impacting on the groundwater resource or ecosystems that are dependent on groundwater. If there were no other forms of recharge, values greater than one would be unsustainable. Management plans may identify groundwater requirements for ecosystems or baseflow to streams, which would lead to E/R ratios less than one.

6.7.2 Recharge scaling factors RSF as estimated by a soil-vegetation-climate model (WAVES) have been used to modify rainfall recharge estimates for the different scenarios. The use of these new values to estimate E/R inevitably leads to a large variability in estimates of E/R, due to the large variability in climate change scenarios. Nonetheless, the median recharge is expected to decrease under the climate scenarios. For some global climate models, a decrease in rainfall is associated with an increase in recharge. In these cases, the extreme events are greater even though the average rainfall is lower. Recharge is sensitive to the more extreme events. Over the last ten years, there has been 15.4 GL/y less recharge. The Scenario B estimates that a continuation of the climate indicative of the last ten years would lead to a reduction of 14.6 GL/y.

6.7.3 EMLR The EMLR has a low level of development. Under Scenario D this shifts to a moderate level of development. A continuation of the recent climate or the dry extreme of the 2030 climate range would also effectively raise the level of groundwater developed to moderate for the EMLR GMU.

6.7.4 Marne Saunders E/R is higher for the Marne Saunders (0.67). Under a continuation of the recent (last ten years) climate E/R would increase to 0.79. Under the 2030 climate E/R would be 0.69 (best estimate), 0.59 (wet extreme) and 1.01 (dry extreme). A continuation of the recent climate or the dry extreme of the 2030 climate range would effectively raise the level of groundwater developed to high to very high for the Marne Saunders GMU. Under future climate and land use projections, E/R under the best estimate 2030 climate increases to 1.03 for the Marne Saunders GMU (range: 0.88–1.52). Potential future growth in groundwater extraction could therefore raise the level of groundwater development to high to very high for the Marne Saunders GMU.

For the Marne Saunders GMU rainfall recharge may be as low as 40 percent of the total recharge. The inclusion of flood recharge would therefore decrease E/R values. If the 40 percent value for flood recharge was assumed for all scenarios, the respective ratios would modify to 0.29 (A), 0.34 (B), 0.25, 0.30, 0.43 (C) and 0.38, 0.44 and 0.65 (D). Even allowing

assessment for possible flood recharge, the level of development under Scenario D would require further groundwater analysis to r support management. With the reductions in flow predicted in Chapter 4 for various scenarios, the flood recharge component would also be likely to reduce.

6.7.5 Angas Bremer The main concern at current extraction levels is increasing groundwater salinity in the Angas Bremer GMU and the

6 Groundwate Currency Creek region. E/R values are not useful for the Angas Bremer region as extraction occurs from the semi- confined aquifer.

6.7.6 Stream impacts of groundwater extraction The reduction in streamflow due to current groundwater extraction is estimated to be 7.0 GL/year, once groundwater equilibrium is reached. While for some areas, this equilibrium may be reached quickly, other areas may take up to 50 years .Predicted future groundwater development would reduce stream flow by a further 4.2 GL/year; the total impact on streamflow in the future due to groundwater development therefore being 11.2 GL/year, once equilibrium is reached. This would only have a minor effect on the River Murray as most of the stream flow is lost across the alluvial plains before reaching the River Murray. It may affect downstream irrigators.

6.8 References

Barnett SR (2007) Currency Limestone Management Area - Status Report 2007. South Australia. Department of Water, Land and Biodiversity Conservation. DWLBC Technical Note 2007/10. Barnett SR, Zulfic D and Yan W (2001) Marne River Catchment Groundwater Assessment. Department for Water Resources, Groundwater Assessment Resource Assessment Division. Report DWR 2001/009. Cresswell RG and Gibson DL (2004) Application of Airborne Geophysical Techniques to Groundwater Resource Issues in the Angas– Bremer Plains, South Australia (SASMMSP Site Summary Report), South Australia. Department of Water, Land and Biodiversity Conservation. Report, DWLBC 2004/ 35. ISBN 0–9756945–2–9. Cresswell RG and Herczeg AL (2004) Groundwater Recharge, Mixing and Salinity across the Angas-Bremer Plains, South Australia: Geochemical and Isotopic Constraints. CSIRO Land & Water Report 29/04 / BRS Technical Report DWLBC (2004) Marne–Saunders Prescribed Water Resources Area Discussion Paper. Department of Water, Land and Biodiversity Conservation. DWLBC (2007) Eastern Mt Lofty Ranges. DRAFT REPORT. Department of Water, Land and Biodiversity Conservation. MDBC (2007) Updated summary of estimated impact of groundwater extraction on stream flow in the Murray–Darling Basin. Draft Report. Prepared by REM on behalf of MDBC Canberra. Zhang L and Dawes WE (1998) WAVES - An integrated energy and water balance model [Published as a website] .CSIRO Land and Water Technical Report No 31/98. Zulfic D and Barnett S (2003) Eastern Mount Lofty Ranges groundwater assessment. Department of Water Land and Biodiversity Conservation, South Australia. Report DWLBC 2003/25. Zulfic D and Barnett SR (2007). Angas Bremer Prescribed Wells Area - Groundwater Status Report 2007. South Australia. Department of Water, Land and Biodiversity Conservation. DWLBC Technical Note 2007/09.

7 Environment

7.1 Summary

7.1.1 Issues and observations

• Assessment of the environmental implications of changes in water availability is largely beyond the terms of reference of this project (see Chapter 1). 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. 7 Environment

• Water planning has only recently commenced in the Eastern Mount Lofty Ranges region and at this time there are no formal environmental water allocations or environmental flow rules for the region. While environmental studies have commenced, at this time there are no agreed ecologically-relevant hydrologic indicators for assessing environmental change in the region. Therefore no quantitative environmental assessments have been undertaken for this region.

• Contextual information on river-related environmental assets and values for the region is provided in Chapter 2. The region includes three wetlands of national importance (the Tookayerta & Finniss Catchments, the Marne River Mouth, and Ambersun – West Swamp) and the creeks and rivers of the region flow into the River Murray (an ’Icon Site’ under the Living Murray Initiative) and the internationally important Coorong and Lakes Alexandrina and Albert Ramsar wetland. These latter two environmental assets are considered in reporting for the Murray region. Additionally, the Eastern Mount Lofty Ranges region contains the regionally important Fleurieu Swamps in the Currency Creek, Tookayerta Creek and Finniss River catchments.

• Changes in end-of-system flows for the creeks and rivers draining the Eastern Mount Lofty Ranges region are reported on in Chapter 4. As the WaterCRESS models do not consider the substantial losses in the lower plains reaches of the catchments, only the end-of-system flow for the Finniss, Tookayerta and Currency catchments should be considered realistic indicators of flows to the downstream receiving waters. The lower reaches of these later catchments contain the regionally important Fleurieu Swamps. These wetlands may have been affected by the considerable reductions in low flows in the Finniss River and Currency Creek caused by water resource development. They would also be likely to be affected by future flow changes caused by climate change or by any commercial plantation forestry development in Currency Creek.

Appendix A Rainfall-runoff results for subcatchments

Apx Table A-1. Summary of modelling results for the subcatchments under scenarios A and C

Scenario A Scenario Cdry Scenario Cmid Scenario Cwet Modelling Area Rainfall APET Runoff Runoff Runoff Rainfall Runoff Rainfall Runoff Rainfall Runoff catchment coefficient contribution km2 mm Percent Percent change from Scenario A 4265213 4 327 1267 4 1% 0% -20% -43% -5% -10% 4% 15%

4265214 55 323 1274 5 1% 0% -20% -43% -5% -11% 3% 8% A ppendix 4265215 23 327 1267 4 1% 0% -20% -43% -5% -10% 4% 15% 4265216 110 359 1271 9 3% 1% -19% -46% -5% -15% 2% 3% 4265223 82 325 1264 4 1% 0% -20% -41% -5% -12% 4% 12% A

4265224 317 465 1248 21 4% 5% -19% -46% -5% -16% 0% 1% subcatchments for results Rainfall-runoff 4265225 5 327 1267 4 1% 0% -20% -43% -5% -10% 4% 15% 4265226 31 327 1267 4 1% 0% -20% -43% -5% -10% 4% 15% 4265243 88 420 1232 12 3% 1% -19% -43% -5% -14% 0% 1% 4265244 43 403 1238 10 2% 0% -19% -43% -5% -13% 0% 1% 4265245 74 552 1214 35 6% 2% -19% -44% -5% -15% 0% 0% 4265253 56 469 1213 22 5% 1% -19% -40% -5% -12% 0% 1% 4265323 788 375 1242 6 2% 3% -20% -42% -5% -13% 2% 6% 4265324 75 379 1253 7 2% 0% -20% -43% -5% -12% 0% 1% 4265325 200 369 1259 7 2% 1% -20% -44% -5% -12% 1% 3% 4265483 349 289 1292 3 1% 1% -20% -40% -5% -11% 4% 13% 4265492 100 376 1250 6 2% 0% -20% -43% -5% -12% 1% 3% 4266292 18 420 1232 12 3% 1% -19% -43% -5% -14% 0% 1% 4266293 122 674 1204 81 12% 7% -19% -45% -5% -16% 0% 0%

42605030 60 723 1192 89 12% 4% -19% -45% -5% -16% 0% 0% 42605040 171 879 1194 163 19% 20% -19% -44% -5% -15% 0% 0% 42605290 239 494 1274 23 5% 4% -19% -47% -5% -17% 0% 1% 42605300 57 764 1173 97 13% 4% -19% -39% -5% -12% 0% 0% 42605330 473 567 1231 49 9% 16% -19% -46% -5% -16% 0% 0% 42610122 238 296 1301 3 1% 1% -20% -40% -5% -11% 4% 13% 42610200 101 765 1196 100 13% 7% -19% -38% -5% -12% 0% 0% 42610282 93 447 1268 21 5% 1% -19% -48% -5% -17% 0% 0% 42610292 130 301 1291 4 1% 0% -20% -41% -5% -11% 4% 14% 42610333 33 584 1194 44 8% 1% -19% -39% -5% -13% 0% 0% 42610334 69 525 1210 34 7% 2% -19% -39% -5% -12% 0% 1% 42610335 285 432 1223 17 4% 3% -19% -40% -5% -12% 0% 1% 42610502 204 703 1201 95 14% 14% -19% -43% -5% -15% 0% 0% 4693 463 1244 30 7% 100% -19% -44% -5% -15% 1% 0%

Apx Table A-2. Summary of modelling results for the subcatchments under scenarios A and D

Modelling A runoff Plantations Farm dam increase Ddry runoff Dmid runoff Dwet runoff catchment increase mm ha ML ML/km2 Percent change from Scenario A 4265213 4 0 0 0.0 -43% -10% 15% 4265214 5 0 0 0.0 -43% -11% 8% 4265215 4 0 0 0.0 -43% -10% 15% 4265216 9 0 28 0.3 -49% -20% -2% 4265223 4 0 0 0.0 -41% -12% 12% 4265224 21 0 249 0.8 -48% -19% -3% 4265225 4 0 0 0.0 -43% -10% 15%

4265226 4 0 0 0.0 -43% -10% 15% 4265243 12 0 0 0.0 -43% -14% 1% 4265244 10 0 0 0.0 -43% -13% 1% 4265245 35 0 0 0.0 -44% -15% 0% 4265253 22 0 0 0.0 -40% -12% 1% 4265323 6 0 0 0.0 -42% -13% 5% 4265324 7 0 0 0.0 -43% -12% 1% 4265325 7 0 22 0.1 -44% -13% 2% 4265483 3 0 0 0.0 -40% -11% 13% 4265492 6 0 0 0.0 -43% -12% 3% 4266292 12 0 0 0.0 -43% -14% 1% 4266293 81 0 551 4.5 -51% -22% -7% 42605030 89 0 405 6.8 -53% -25% -10% 42605040 163 0 186 1.1 -45% -16% -1% 42605290 23 0 468 2.0 -52% -24% -7% 42605300 97 0 0 0.0 -39% -12% 0% 42605330 49 0 423 0.9 -48% -19% -3% 42610122 3 0 0 0.0 -41% -11% 13%

Appendix A Rainfall-runoff results for subcatchments 42610200 100 884 375 3.7 -46% -22% -11% 42610282 21 0 127 1.4 -52% -23% -6% 42610292 4 0 0 0.0 -41% -11% 14% 42610333 44 0 0 0.0 -39% -13% 0% 42610334 34 1110 0 0.0 -49% -27% -15% 42610335 17 0 55 0.2 -44% -18% -5% 42610502 95 0 491 2.4 -46% -18% -4% 30 1994 3379 0.7 -46% -18% -3%

Appendix B River water modelling reach mass balances

Marne River

A B Cwet Cmid Cdry Dwet Dmid Ddry

GL/y

Storage volume

Change over period 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 App Inflows ni ie ae modellin water River B endix Imported water 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Urban runoff 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Sub-catchment flows 10.8 9.6 10.3 8.5 4.8 10.3 8.5 4.8

Sub-total 10.8 9.6 10.3 8.5 4.8 10.3 8.5 4.8

Diversions

Total estimated use from farm dams 1.1 1.0 1.0 1.0 0.8 1.1 1.1 0.9

Sub-total 1.1 1.0 1.0 1.0 0.8 1.1 1.1 0.9

Outflows

End-of-system outflow 8.9 7.7 8.4 6.8 3.3 8.3 6.6 3.2

Net evaporation 0.8 0.8 0.8 0.8 0.6 0.8 0.8 0.7 g

River loss 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 balances mass reach

Sub-total 9.7 8.5 9.2 7.5 4.0 9.1 7.4 3.9

Unattributed fluxes

Total 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Mass balance error -0.07% -0.07% -0.07% -0.06% 0.03% -0.08% -0.08% 0.01%

Bremer River

A B Cwet Cmid Cdry Dwet Dmid Ddry

GL/y

Storage volume

Change over period 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Inflows

Imported water 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4

Urban runoff 1.0 0.9 1.0 0.9 0.8 1.0 0.9 0.8

Sub-catchment flows 21.2 16.8 20.2 15.9 7.3 20.2 15.9 7.3

Sub-total 22.6 18.1 21.5 17.2 8.4 21.5 17.2 8.4

Diversions

Total estimated use from farm dams 1.4 1.4 1.4 1.4 1.1 1.5 1.5 1.2

Sub-total 1.4 1.4 1.4 1.4 1.1 1.5 1.5 1.2

Outflows

End-of-system outflow 19.8 15.3 18.7 14.5 6.1 18.6 14.3 5.9

Net evaporation 0.9 0.9 0.9 0.9 0.8 1.0 0.9 0.8

River loss 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 reach mass balances

g Sub-total 21.1 16.7 20.1 15.8 7.3 20.0 15.7 7.2

Unattributed fluxes

Total 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

modellin r Mass balance error -0.06% -0.05% -0.05% -0.05% 0.00% -0.07% -0.07% -0.01% wate r endix B Rive pp A

Angas River

A B Cwet Cmid Cdry Dwet Dmid Ddry

GL/y

Storage volume

Change over period 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Inflows

Imported water 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Urban runoff 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2

Sub-catchment flows 16.9 11.4 16.3 14.0 8.5 16.3 14.0 8.5

Sub-total 17.1 11.7 16.5 14.2 8.7 16.5 14.2 8.7 App Diversions

Total estimated use from farm dams 1.1 1.1 1.1 1.1 1.0 1.4 1.4 1.3 modellin water River B endix

Sub-total 1.1 1.1 1.1 1.1 1.0 1.4 1.4 1.3

Outflows

End-of-system outflow 15.1 9.5 14.4 12.1 6.6 13.9 11.6 6.2

Net evaporation 0.9 1.0 1.0 1.0 1.0 1.2 1.2 1.2

River loss 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Sub-total 16.0 10.6 15.4 13.1 7.7 15.1 12.8 7.4

Unattributed fluxes

Total 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Mass balance error -0.14% -0.18% -0.14% -0.15% -0.22% -0.17% -0.19% -0.28% balances mass reach

Finniss River

A B Cwet Cmid Cdry Dwet Dmid Ddry

GL/y

Storage volume

Change over period 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Inflows

Imported water 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Urban runoff 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Sub-catchment flows 46.0 30.8 44.4 37.8 22.7 44.4 37.8 22.7

Sub-total 46.0 30.8 44.4 37.8 22.7 44.4 37.8 22.7

Diversions

Total estimated use from farm dams 1.9 1.9 1.9 1.9 1.8 2.1 2.1 2.0

Sub-total 1.9 1.9 1.9 1.9 1.8 2.1 2.1 2.0

Outflows

End-of-system outflow 42.7 27.4 40.9 34.3 19.2 40.6 34.0 19.0

Net evaporation 1.3 1.5 1.4 1.4 1.6 1.6 1.6 1.7

River loss 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 reach mass balances

g Sub-total 43.9 28.8 42.3 35.8 20.8 42.1 35.6 20.6

Unattributed fluxes

Total 0.2 0.1 0.1 0.1 0.1 0.2 0.1 0.1

modellin r Mass balance error -0.09% -0.12% -0.09% -0.10% -0.13% -0.10% -0.11% -0.14% wate r endix B Rive pp A

Tookayerta Creek

A B Cwet Cmid Cdry Dwet Dmid Ddry

GL/y

Storage volume

Change over period 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Inflows

Imported water 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Urban runoff 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Sub-catchment flows 17.4 17.2 17.0 15.4 11.2 13.8 12.6 9.2

Sub-total 17.4 17.2 17.0 15.4 11.2 13.8 12.6 9.2 Ap

Diversions p ni ie ae modellin water River B endix Total estimated use from farm dams 0.3 0.3 0.3 0.3 0.3 0.4 0.4 0.3

Sub-total 0.3 0.3 0.3 0.3 0.3 0.4 0.4 0.3

Outflows

End-of-system outflow 16.7 16.6 16.3 14.8 10.5 13.0 11.8 8.3

Net evaporation 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.5

River loss 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Sub-total 17.1 16.9 16.7 15.1 10.9 13.5 12.2 8.8

Unattributed fluxes

Total 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

g reach mass balances mass reach Mass balance error 0.00% 0.00% 0.00% 0.00% 0.01% -0.01% -0.01% -0.01%

Currency Creek

A B Cwet Cmid Cdry Dwet Dmid Ddry

GL/y

Storage volume

Change over period 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Inflows

Imported water 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Urban runoff 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Sub-catchment flows 8.3 7.9 9.1 7.6 4.5 8.4 7.0 4.2

Sub-total 8.3 7.9 9.1 7.6 4.5 8.4 7.0 4.2

Diversions

Total estimated use from farm dams 0.5 0.5 0.5 0.5 0.4 0.5 0.5 0.4

Sub-total 0.5 0.5 0.5 0.5 0.4 0.5 0.5 0.4

Outflows

End-of-system outflow 7.4 7.1 8.2 6.7 3.7 7.6 6.2 3.4

Net evaporation 0.5 0.4 0.4 0.4 0.4 0.4 0.4 0.4

River loss 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 reach mass balances

g Sub-total 7.9 7.5 8.6 7.1 4.1 7.9 6.6 3.8

Unattributed fluxes

Total 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

modellin r Mass balance error 0.02% 0.02% 0.01% 0.02% 0.08% 0.02% 0.04% 0.11% wate r

endix B Rive pp A

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 project 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.

Table C-1. Explanation of components of the uncertainty assessments A pni ie ytmmdlucranyassmn yreach by uncertainty assessment model system River C ppendix

Table Description Land use 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. Gauging data 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-to- month 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. Correlation with Information on the likely nature of ungauged components of the reach water balance. ungauged gains/losses 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) Water 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.

Table Description Change- Information on the significance of the projected changes under different scenarios, considering the performance of uncertainty ratios 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 one 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 one 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.

uncertainty assessment by reach Appendix C River system model

Reach 1

Downstream gauge 426533 BREMER RIVER @ near Hartley Upstream gauge 426557 MOUNT BARKER CREEK @ D/S Mt. Barker 426658 DAWESLEY CREEK @ U/S Brukunga Mine

Reach length (km) - Note: the sub-catchment has gauges for two small headwater areas, Area (km2) 473 and the reach balance is defined between them and the outflow Outflow/inflow ratio 2.62 gauge. The map, which is for the whole sub-catchment, does not Net gaining reach show these areas. They account for about 45 % of inflows.

This is a strongly gaining reach. Flows are dominated by local storm Land use ha % runoff. Dryland 47,310 100 Irrigable area - - Some of the inflows are gauged. Estimated local runoff explains most Open water* - - of the ungauged gains but adjustment was required. There are no River and wetlands - - recorded diversions and ungauged losses are unknown Open water* - - * averages for 1990–2006 Baseline model performance is good. Accounting explains observed flows moderately well. The projected changes with the scenarios are

Gauging data Inflows Outflows Overall 100 and gains and losses unattributed Fraction of total 80 gains Gauged 0.45 0.89 0.67 A 60 ungauged Attributed 0.83 0.89 0.86 gains reach by uncertainty assessment model system River C ppendix Fraction of variance 40 Gauged 0.17 0.91 0.54 gauged 20 gains Attributed 0.89 0.98 0.93 0 unattributed losses -20 Correlation with ungauged Gains Losses Linear adjustment ungauged normal ranked normal ranked -40 losses

-0.79 -0.76 (GL/y) losses and gains Reach Main gauge inflows -0.17 -0.27 -60 Tributary inflows -0.88 -0.77 -0.18 -0.36 gauged losses Main gauge outflows -0.97 -0.90 -0.19 -0.54 -80 Distributary outflows - - - - -100 Recorded diversions - - - - -0.76 -0.73 Estimated local runoff -0.21 -0.33 Adjusted -63% 90/91 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

100

gauged accounted model 10

0.1

0.01 Monthly streamflow (GL/mo)

0.001 Jan-90 Jan-91 Jan-92 Jan-93 Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Jan-06

Water balance Model (A) Accounts Difference Model efficiency Model (A) Accounts 100 Jul 1990 – Jun 2006 Monthly Gains GL/y GL/y GL/y Normal 0.82 0.84 Main stem inflows 6.5 6.0 0.4 Log-normalised - - Tributary inflows 3.0 2.1 0.9 Ranked 0.75 0.79 10 Local inflows 9.8 6.7 3.1 Low flows only - - Unattributed gains and noise - 3.0 -3.0 High flows only 0.47 0.62 Losses GL/y GL/y GL/y Annual 1 Main stem outflows 19.3 15.8 3.5 Normal 0.80 0.92 Distributary outflows 0.0 0.0 0.0 Log-normalised 0.82 0.90 Net diversions 0.3 0.0 0.3 Ranked 0.93 0.81 0.1 River flux to groundwater 0.0 - 0.0 River and floodplain losses 0.0 0.0 0.0 Definitions: Monthly streamflow (GL/mo) . Unspecified losses 0.0 - 0.0 - low flows (flows<10% percentile ) : 0.0 GL/mo 0.01 Unattributed losses and noise - 2.0 -2.0 - high flows (flows>90% percentile) : 3.5 GL/mo

Change-uncertainty ratios 0.001 0 20406080100 P B Cwet Cmid Cdry Dwet Dmid Ddry Annual streamflow 0.3 0.6 0.2 3.3 0.5 0.2 3.3 Pecentage of months flow is exceeded Monthly streamflow 0.6 0.7 0.6 2.3 0.7 0.6 2.4

gauged 1000 120 B C D A + wet 100 100 O mid B

– dry 80 Cwet 10 Cmid 60 Cdry 1 0.01 0.1 1 10 100 1000 40 Dwet Annual streamflow (GL/y) streamflow Annual 0.1 Dmid 20 Monthly Change-Uncertainty Ratio Ratio Change-Uncertainty Monthly Ddry 0.01 0 Annual Change-Uncertainty Ratio 90/91 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

Reach 2

Downstream gauge 426504 FINNISS RIVER @ 4Km East Of Yundi

100

0.1

0.01

Monthly streamflow(GL/mo) gauged model

0.001 Jan-90 Jan-91 Jan-92 Jan-93 Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Jan-06

Model efficiency Model (A) 100 This is a comparison of a gauged and modelled outflow. No inflow Monthly gauge was available, and so no water account has been done. The Normal 0.94 only meaningful results for this reach are those for the model Log-normalised 0.74 uncertainty analysis. 10 Ranked 0.85

Low flows only <0 The model fit is very good. High flows only 0.50 Annual 1 The Cdry and Ddry scenarios show change outside the range of Normal 0.84 uncertainty in modelling. The change predicted in the other scenarios Log-normalised 0.88 is within or close to the range of uncertainty. Ranked 0.85 0.1

Definitions: Monthly streamflow (GL/mo) . (GL/mo) streamflow Monthly - low flows (flows<10% percentile ) : 0.0 GL/mo 0.01 - high flows (flows>90% percentile) : 5.9 GL/mo

0.001 Change-uncertainty ratios 0 20406080100 P B Cwet Cmid Cdry Dwet Dmid Ddry Annual streamflow 1.2 0.7 0.6 7.2 0.6 0.6 7.3 Pecentage of months flow is exceeded Monthly streamflow 1.6 0.9 1.1 4.7 0.9 1.2 4.8

gauged 1000 60 A 50 100 B

40 Cwet 10 Cmid 30 Cdry 1 0.01 0.1 1 10 100 1000 20 Dwet Annual streamflow (GL/y) streamflow Annual uncertainty assessment by reach 0.1 Dmid 10 Monthly Change-Uncertainty Ratio Ratio Change-Uncertainty Monthly Ddry 0.01 0 Annual Change-Uncertainty Ratio 90/91 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 Appendix C River system model

Enquiries

More information about the project can be found at www.csiro.au/mdbsy. This information includes the full terms of reference for the project, an overview of the project methods and the project reports that have been released to-date.