technical features climate change refereed paper

CLIMATE CHANGE IMPACT ON ROUS WATER SUPPLY D Kirono, G Podger, W Franklin, R Siebert Abstract Under this scheme the secure yield is defined as the annual Climate change poses significant demand that can be supplied risks to the security of water from the headworks over the resources in many parts of 103 year historic record and . This study investigated the implications for which satisfies the 5/10/20 secure yields of a regional water rule, i.e.: supplier in NSW. The aim of • restrictions of any kind the study was to estimate future should not be applied for more secure yields, using the than 5% of the time (> 1862 Integrated Quantity and Quality days); Model (IQQM) and a range of • restrictions of any kind climate change scenarios, to should not be imposed more identify the most appropriate than one year in ten on average time horizon for making new (> 10 years); and investments in infrastructure. • the system should be able to The assessment concluded that, supply 80% of normal demand when taking into consideration (i.e. 20% reduction in impacts of global warming, the consumption) through a repeat need for a new source will be of the worst drought on record. most likely after 2018, Water presently comes from consequently planning needs to two main supply storages: commence in 2008. Rocky Creek and Introduction Emigrant Creek Dam. The There is an increasing body of former has a storage capacity of research that supports a picture 13,956 ML and a safe yield of of a warming world with about 9,600 ML/annum (DIPNR, 2004) while the significant changes in regional Figure 1. Rous Water scheme (Rous Water, 2006). climate systems (IPCC, 2001; latter has a capacity of 820 ML IPCC, 2007). This has with a safe yield of about significant implications for the 1,100 ML/annum. There are reliability of water supplies across Australia (NSW). The third section describes the also some other small sources (Jones and Preston, 2006). It is important framework built to assess climate change under Council control with a combined to consider the risk that climate change impacts on Rous Water’s supply. Section safe yield of about 900 ML/annum. poses to catchment yield within the context four presents and discusses the assessment Based on population projections and other of changing demand. procedure and subsequent results, and considerations, GeoLINK (2005) estimated This project investigated the implications section five summarises the main that the best estimate of the likely future that climate change may have on Rous conclusions of the study. demand from the Rous Water scheme in Water’s regional water supplies. The aim of the year 2030 is around 18,000 Overview of Rous Water Regional the study was to help identify the most ML/annum. This equates to an increase of appropriate time horizon for making new Water Supply 43% over the current demand of 12,600 investments in infrastructure. Rous County is located in northeastern ML/annum. To manage this growth in The first section provides an overview of NSW and is part of the Wilsons River demand, Rous Water adopted a water Rous Water’s regional water supply scheme, catchment, from Lofts Pinnacle in the west, management strategy in 1995 that was the second discusses the projected climate along Nightcap and Koonyum ranges near amended in 2004. The strategy provides a change profile for the coast (Figure 1). Rous Water is the range of options to meet water regional water supply authority providing requirements. One key objective of Rous The need for a new water in bulk to a number of Councils, Water’s strategy is to implement effective from Lismore to Ballina, with an demand management, the target being a source is most likely in approximate population of 93,000. The minimum 10% reduction in per capita supply system was designed based on the demand by the year 2011, relative to 2005. 2018: planning should former NSW Department of Public Works Recognising future demand may outstrip start in 2008. and Services definition of ‘secure yield’. existing sustainable yield, the strategy

68 MARCH 2007 Journal of the Australian Water Association technical features climate change refereed paper

identified two additional supplies, Lismore source and Dunoon Dam. Lismore source is a medium-term solution which is able to assist in meeting the high demand projection up to 2024. It consists of a pumping station capable of abstracting up to 30 ML/day from the upper reaches of the tidal pool in the Wilsons River, which is only utilised when Rocky Creek Dam is below 95% capacity and is subject to abstraction licence constraints. The proposed Dunoon Dam is located downstream of Rocky Creek Dam and captures local inflows as well as Rocky Creek dam spills (CMPS&F, 1995). An Overview of Climate Change in NSW During the 20th century, the globally averaged surface temperature increased by 0.6 ± 0.2ºC with the warmest year being 1998, followed by 2005 (WMO, 2005). In Figure 2. Framework for climate change impact assessment for Rous Water Scheme. NSW, the temperature has also been steadily increasing over the last fifty years (BOM, 2006). In northeastern NSW, change, scientists have developed climate Framework for Impact Assessment temperatures have increased at the rate of scenarios from global climate models º Impact and risk assessment is one stage in a approximately 0.4 C per decade. The BOM (GCMs). Best estimates for globally average larger risk management framework. Ideally, (2006) has also shown that rainfall has been surface air warming are expected to range risk management involves all related declining in the non summer months at a between 1.8ºC to 4.0ºC at 2090-2099 stakeholders. The decision-making process rate of roughly -20 to -50 mm per decade. relative to 1980-1999 (IPCC, 2007). In is commonly circular to allow the In summer, the rainfall has been increasing Australia, CSIRO uses both global and performance of chosen decisions to be at a rate of approximately +10 to +20 mm regional climate models in the development reviewed and revisited as new information per decade. Lismore rainfall records show of regional climate change projections. that the annual rainfall has been decreasing According to Hennessy et al (2004), the on climate change and its impacts are slightly at a rate of -6 mm per decade from models tend to simulate decreasing annual- available. the late 1880s to present. Winter rainfall average rainfall over NSW, particularly in Most research on the hydrologic impact of has been decreasing at a rate of -5 mm per winter and spring. In autumn the direction climate change uses a predictive approach. decade, whereas the summer rainfall has of the change is uncertain, while in summer It begins with generating climate change been increasing at a rate of +1 mm per there is a tendency for increases in the scenarios. Climate information is then fed decade. However it should be noted that north-east. Annual-average potential into hydrologic models and/or water- the trends in Lismore’s rainfall are not evaporation is projected to increase across management systems to evaluate the statistically significant at a 95% confidence NSW. The largest changes are projected in differences in system performance under level. winter with the smallest changes in different climate scenarios. Adaptations can The climate system is highly complex, and summer. Compared to changes in the other then be designed to manage those changes. therefore it is inappropriate to simply areas of NSW, the projected changes in the Through consultations with stakeholders, extrapolate past trends to predict future north-east, where Rous Water is located, the assessment framework for climate conditions. To estimate future climate are relatively small. change impacts and adaptation for Rous

Journal of the Australian Water Association MARCH 2007 69 technical features climate change refereed paper

Water has been established and is presented scenario may be considered as a more study. In this variation the storages are in Figure 2. Details about the main steps in optimistic future. It describes a convergent configured to 55% capacity, which this framework are described as follows. world with the global population that peaks represents the capacity at which restrictions in mid-century and declines thereafter, are imposed on the system. This First step: preparing climate change rapid change in economic structures toward configuration is used to verify the 20% scenarios a service and information economy, with rule. As described previously, future climate reductions in material intensity and the To be able to assess the impacts on flow scenarios are commonly developed through introduction of clean and resource-efficient due to changes in climate the time series GCMs. Currently, there is a range of technologies (IPCC, 2001). This step inputs used in IQQM were modified. In available GCMs, each developed by a produced a total of 18 simulations for each the FX94D scenario IQQM uses a mixture different scientific group across the world. year and for every station used in the of observed and predicted flows that vary These models differ in their approaches to Wilsons River IQQM. between sites. The Sacramento rainfall simulating the climate, hence different runoff model was used to extend flows for models may project different climate Second step: estimating supply under most of the tributaries. Emigrant Creek futures, even when driven by the same climate change Dam was based on the Australian Water scenario of future emissions. The standard The climate scenarios were applied to the Balance Model (AWBM) rainfall-runoff measure to compare climate models is their Integrated Quantity Quality (IQQM) model while Rocky Creek dam was based ‘sensitivity’ defined by how much eventual hydrologic, river system simulation package on a monthly correlation with a nearby warming they project when the-pre- (Simons et al. 1996) to generate climate gauge and then disaggregated by a mixture industrial atmospheric concentration of change flow sequences. IQQM consists of of techniques. For this study all tributaries CO is doubled from around 270 to 550 the Sacramento rainfall-runoff model and 2 needed to be modelled using rainfall-runoff parts per million by volume (p.p.m) river routing, water demand and allocation models and to be consistent the (today’s atmospheric CO is about 380 routines to simulate river flow and river 2 Sacramento model was used for all p.p.m). GCMs use a particular emissions regulation. This software has been tributaries. DIPNR had already developed scenario as the input to generate a implemented in most regulated and a large Sacramento models for Rocky Creek Dam projection of climate change over the next number of unregulated river systems in and Emigrant Creek Dam that had a good century. The IPCC commissioned a range NSW and Queensland. This study used the match with the short periods of observed of scenarios of greenhouse gas and sulfate Wilsons River IQQM implementation data and these were adopted. aerosols emissions up to the year 2100. The which is described in detail in DIPNR scenarios were reported in the Special (2004). Simulations with the Sacramento model Report on Emissions Scenarios (SRES, The number of observed rainfall, generally matched well with observed flows 2000). For example, the SRES-A1 scenario evaporation, and stream flow stations used at all sites. However, when the observed depicts a very rapid economic growth, in the Wilsons River model are 8, 1, and data were replaced with the simulated data, global population that peaks in mid- 14, respectively. Rainfall data was used to the result for secure yield was different to century and declines thereafter, and the account for soil moisture (which governs that from the original FX94D model. As rapid introduction of new and more the crop water demands of irrigators) and the model is most sensitive to inflows from efficient technologies (IPCC, 2001). to apply rainfall to the water surfaces of Rocky Creek Dam and Emigrant Creek Dam, the Sacramento models for these For this study, scenarios of regional change and river reaches. Rainfall data tributaries were adjusted with an aim to as a function of global warming (percent was also required for calculating catchment change per ºC of global warming) for inflows. The evaporation data were used to reproduce the original secure yield of potential evaporation (Ep) and estimate evapotranspiration from crops, 14,900 ML/annum. Unfortunately this was precipitation (P) were prepared based on evaporation from reservoirs and river not quite possible due to very small six GCM simulations through the use of reaches, and to synthesise streamflow. The differences (<2%) in critical events for the the CSIRO Climate Scenario generator, model incorporates existing irrigation 5% and 10% rules. The best that could be OzClim. OzClim is a PC-based climate development (3,000 ha); system demands achieved was a secure yield of 15,000 scenario generator that simplifies the (11 urban/rural demand centres); sources of ML/annum while keeping the difference in process of calculating scenarios from supply (Rocky Creek Dam, Emigrant overall volume at Emigrant Ck and Rocky climate change model outputs, applies Creek Dam, and Borefields); scheme Ck within 2%. Consequently, for this scenarios to impact models and manages operating rules and access constraints to study, all secure yields are compared against uncertainty (Page and Jones, 2001). A streamflows at Lismore source. a baseline secure yield of 15,000 ML/annum rather than the 14,900 range of GCM, emission scenarios and The published results for the Wilsons River ML/annum obtained by DIPNR in the climate sensitivities can be harnessed using model (DIPNR, 2004) are based on a FX94D scenario. this system. For this assessment, multiple model scenario known as FX04D. climate change simulations were conducted However, subsequent to this report, the Estimation of the secure yield for each for the years 2010, 2020, and 2030. These Lismore source access rules have been climate scenario was estimated based upon were based on six climate models assuming investigated further, resulting in a revised the minimum yield required to meet the a range of climate sensitivities (low, scenario known as FX94D, which is as yet 5/10/20 rule. The estimation of secure medium and high) and the SRES A1B, unpublished. Based on a discussion with yield was determined by an iterative A1F and B1 scenarios. The A1 scenarios Rous Water, this study has adopted the solution that successively modifies and runs can be considered as a pessimistic as they FX94D model scenario as the baseline from IQQM until the demand was just met in give high CO2 emissions. A1B depicts a which to assess the impacts of climate accordance with the 5/10/20 rule. This balance across all sources, while A1F changes. To assess the 20% rule a variation process was carried out for each of the 18 depicts a fossil intensive situation. The B1 of the FX94D model was created for this climate scenarios, for a total of 54 scenarios

70 MARCH 2007 Journal of the Australian Water Association technical features climate change refereed paper

(i.e. 18 scenarios for each of the years 2010, 2020 and 2030). To solve each scenario, IQQM was run approximately 10 times, resulting in approximately 540 runs to cover all of the climate scenarios and time periods. Third step: risk assessment The issue of climate change is beset by uncertainties. These uncertainties include the magnitude of global warming, regional changes in rainfall and evaporation, and regional supply sensitivity and coping capacity. Quantifying the uncertainties in climate change and its downstream consequences in units of probability or likelihood helps to identify robust adaptation strategies (Jones and Hennessy, Figure 3. Supply projections due to climate change (CC) versus demand projection. 2000). In this study an event-based Note: the high demand, best estimate demand, and low demand curves are taken probability, where the likelihood of and modified from GeoLINK (2005). recurring events is estimated, was used to describe the future state of climate change under the enhanced greenhouse effect. To do this, Monte Carlo methods (repeated For assessment purposes, the 5th, 50th, and demand and its trend, the historical record random sampling) were employed to 95th percentiles of the probability from 1996 to 2005 and its trend projection stochastically generate probabilistic distribution are then considered as the are also presented. The demand projections estimates of future climate change and its “wet”, “medium” and “dry” scenarios as (either based on the GeoLINK, 2005, or depicted in Figure 3. impacts on Rous Water’s secure yields. based on the trend analysis of the historic • In the wet climate change scenarios, demand) show an obvious increase. The An assessment was then performed by over the near term, the secure yield will fall slope of the actual demand trend is similar comparing the projected future supply, in 2010 before increasing to above the to the best-estimate demand provided by after accounting for climate change, with present supply in 2020 and will continue to GeoLINK (2005), suggesting that the two projected future demand. In this study the rise to more than 16,000 ML in 2030. The are relatively comparable. Demand is likely demand projections were taken from the slight fall in 2010 is hypothesised due to a to increase to more than 14,150 ML results of the GeoLINK (2005) study that relative similar rate of change in both (according to the trend analysis) or more were slightly modified so that they have the rainfall and evaporation (i.e. 2% above than 14,900 ML (according to GeoLINK, same starting point with the actual demand current value by 2010). As the models are 2005) by 2030. trend in 2006 (i.e. approximately 11,973 more sensitive to the change in evaporation The points of concern are where a given ML). The trends in actual demand were this causes a slight decrease in yield in supply scenario encounters a particular also estimated based on the available 2010. In 2020 and 2030, the increase in demand scenario as summarised in Table 1. historical demand from 1996 to 2005. rainfall is much higher than the increase in evaporation, hence the yield increases. In the combination of “Wet supply and Assessment Results and • In the medium climate change Low demand”, “Wet supply and Medium Recommendations scenarios, the secure yield will decline to demand”, and “Medium supply and Low Assessment around 14,000 ML in 2030. demand” scenarios, the supply will keep Simulations with future climate scenarios • In the dry climate change scenario, the pace with the demand, therefore no further within IQQM resulted in a range of secure yield will decline to approximately action would be required. estimated future secure yields for the Rous 11,600 ML in 2030. • In the combination of “Wet supply and Water scheme in 2010, 2020 and 2030. The projections of the future demand High demand” scenario, however, the need These results were subsequently used to (estimated by GeoLINK, 2005) are plotted for a new source will be in 2018. generate cumulative probability in Figure 3. To represent the actual • In the combination of “Medium distributions for projected yields. The supply and High demand”, and increases in supply are found to be “Medium Supply and Medium very unlikely (<10% probability), Table 1. Options on the appropriate time to have demand” scenarios, there is a need whereas the declines in supply are a new source, according to the different supply and demand scenarios. for a new source in 2014 and 2023, likely (>66% probability). This respectively. suggests that the probability of Supply scenarios changes in water supply is skewed Wet Medium Dry • In the combination of “Dry towards the “decrease in future supply and High demand”, “Dry Demand scenarios supply” scenarios. The best estimate supply and Medium Demand”, and Low - - 2025 (50% probability) changes in secure “Dry supply and Low demand”, the yields in 2010, 2020, and 2030 are - Medium - 2023 2016 need for a new source will be in 1.7%, -5.8%, and -8.1% respectively. High 2018 2014 2012 2012, 2016, and 2025, respectively.

Journal of the Australian Water Association MARCH 2007 71 technical features climate change refereed paper

Recommendations (e.g. periods of anomalous drought or CMPS&F. 1995. Rous Regional Water Supply The assessment provides several options on rainfall) not represented by the historical Strategy Planning Study, Scheme Options, Final Report. May 1995. the most appropriate time to build a new record or model simulations that could also DIPNR. 2004. Rous Water – Augmentation of water source, according to the different affect water supply within the time horizon in question. Changes in decadal mean Regional Water Supply Scheme, Hydrologic supply and demand scenarios. In terms of modelling study using Wilsons River IQQM. the future supply, this study has shown that rainfall may occur due to other factors Issue 3. NSW Department of Infrastructure, the probability of changes in water supply other than the greenhouse effect (e.g. long Planning and Natural Resource is skewed towards the “decrease in future term variations in natural climate), so there GeoLINK. 2005. Dunoon Dam, Population supply” (hence the “medium” and “dry”) may be additional changes and risks above and demand projections. Draft report, 19- scenarios. In terms of the future demand, and beyond those accounted for here. 08-2005. GeoLINK (2005) has estimated that the Hennessy, K., Page, C., McInnes, K., Jones, R., Conclusions probability of the future demand is skewed Bathols, J., Collins, D. and Jones, D. 2004. towards the high and medium demand This paper has described the framework Climate change in New South Wales, Part 1: scenarios. This suggests the followings. and the results of a climate change risk Past climate variability and projected changes assessment of Rous Water’s supplies based in average climate. Consultancy report for • The earliest time for a new source for the the New South Wales Greenhouse Office. upon The Wilsons river IQQM. Results of Rous Water system is after 2018. Thus, IPCC. 2001. Climate Change 2001: The scien- taking into consideration a 10 year the risk analysis suggest that decreases in tific basis. Summary for policymakers. In planning and construction time from the secure yields are likely in the future. The Houghton, J.T., Ding, Y., Griggs., D.J., time of this report there is only two years best estimate (50% probability) changes in Noguer, M., Van Der Linden, P.J. and before commissioning of a new source has secure yields in 2010, 2020, and 2030 are Xioaosu, D (eds) Contribution of working to be initiated. -1.7%, -5.8%, and -8.1% respectively. The group I to the third assessment report of the assessment suggests the earliest time to have Intergovernmental Panel on Climate • The medium time for a new source for a new source is 2018. Given that a ten-year Change, Cambridge University Press, the Rous Water system is 2023. Thus, from Cambridge. the time of this report there is about seven time lead is considered the minimum required to commission a new water source, IPCC, 2007. Climate change 2007: The physical years time before commissioning of a new science basis. Summary for policymakers. source has to be initiated. the plan has to be started in at least 2008. Contribution of working group I to the Within that time, it is recommended that Acknowledgments fourth assessment report of the Intergovernmental Panel on Climate actual demand be closely monitored. This This work was produced by CSIRO under Change. is because estimates of when a new source is contract to the Rous Water Regional Water Jones, R.N. and Hennessy, K.J. 2000. Climate required are highly sensitive to demand Supply, NSW Department of Commerce. change impacts in the Hunter Valley, A risk assumptions. In the case of a dry climate NSW DIPNR is acknowledged for the assessment of heat stress affecting dairy change scenario, for instance, GeoLINK’s agreement regarding the use of Wilsons cattle. A research report, CSIRO demand projection leads to a relatively Atmospheric Research, Victoria. River IQQM. Chris Ribbons and Richard similar suggestion with that of the actual Jones, R.N. and Preston, B.L. 2006. Climate Cooke are thanked for providing the trend of demand (i.e. only 2 years change impacts, risk and the benefits of required files and assistance in running difference). In the case of medium climate mitigation. A report for the Energy Futures them. Cher Page and Jim Ricketts helped change scenario, however, this differential Forum, CSIRO Marine and Atmospheric in setting up the OzClim program for the becomes larger (i.e. 5 years). In addition, Research, Victoria. analysis. The study has benefited from given a high demand and a dry-supply Page, C.M., and Jones, R.N. 2001. OzClim: constructive comments by Dr Roger Jones the development of a climate scenario scenario, a crisis may occur in around and Dr Benjamin Preston. generator for Australia. In: MODSIM 2001: 2012, by which point supply will be unable International Congress on Modelling and to meet the demand yet it will already be The Authors Simulation: proceedings, Australian National too late to commission a new source to University, F. Ghassemi, and others (editors). Dewi Kirono is a research scientist in address the supply/demand gap. Thus, Canberra, ACT: Modelling and Simulation information regarding the demand is Climate Change Impact and Risk Group, Society of Australia and New Zealand. p. valuable in updating the plan as to when a CSIRO Marine and Atmospheric Research. 667-671. new source has to be built. Email: [email protected]. Geoff Rous Water. 2004. Demand management plan, Podger is the Principal Research Scientist, 2004-2009. Rous Water Regional Water The results of the current study indicates River Basin Modeller within CSIRO Land Supply, Adopted Council Meeting March that, ideally, the demand will not be more and Water, and Water Management 2004. than the driest supply projection (i.e. about Research Program Leader within eWater Simons, M., Podger, G. and Cooke, R. 1996. 13,000 ML in around 2018). This means CRC. Email: [email protected]. IQQM – A hydrologic modelling tool for that the maximum tolerable increase of Wayne Franklin is the Operational water resource and salinity management. demand by 2018 is only approximately Environmental Software, Vol 11, Nos. 1-3, Services Manager within Rous Water. 10% of the current demand. If the wet pp 185-192, 1996 Email: [email protected]. supply scenario is taken into account, the SRES. 2000. Special report on emissions gov.au. Rob Siebert is the Project Manager maximum tolerable increase of demand by scenarios (SRES). Special report on the of the Lismore Source Project, Rous Water. 2018 is about 25% of the current demand. Intergovernmental Panel on Climate Email: [email protected]. Therefore, ongoing monitoring of rainfall Change. N. Nakicenovic and R. Swart (eds), gov.au Cambridge 2000. http://www.ipcc.ch patterns is also recommended. WMO (2005) WMO statement on the status of To interpret the results appropriately, there References the global climate in 2005. World is another factor that needs to be taken into BOM. 2006. Australian Bureau of Meteorological Organisation. account, specifically, climate variability Meteorology. www.bom.gov.au. http://www.wmo.ch/web/Press/index.html

72 MARCH 2007 Journal of the Australian Water Association