Water security analysis of the Middle River supply system: a foreSIGHT application

Abstract

Scenario-neutral approaches have become the preferred method of interpreting climate driven impacts on natural and engineered systems when analysing the reliability of water resource systems facing climate change. Scenario-neutral approaches help deal with uncertainty and potential decision paralysis, by stress-testing systems across a range of plausible future conditions, instead of analysing systems under a narrower selection of climate projections. This results in better identification of failure modes and a richer elucidation of system behaviour. A case-study application of a methodology that supports the bottom-up approach—driven by a software tool ‘foreSIGHT’—is presented here. Applied to the Middle River Reservoir system, the scenario-neutral approach adopted here follows three steps to: i) provide insights into system dynamics, ii) reveal modes of failure due to changes in climate, and iii) understand when intervention may be required given multiple lines of evidence. The Middle River system (situated in Kangaroo Island, South Australia) comprises of a single reservoir with relatively small storage volume relative to the annual catchment inflows, resulting in large seasonal fluctuations in reservoir levels with regular filling in winter and drawdown over the dry season. The reservoir can hold approximately 200 ML more than the mean annual demand (450 ML), which is highly variable year to year. The occurrence of draw down below 6 and 2 m levels were of interest to the stakeholders, as they broadly indicate, the need to augment the water supply due to a decrease in water quality (below 6 m), and the reservoir emptying (below 2 m). The reliability of the water supply system in 2030 was the focus of the analysis, as this aligns with opportunities to upgrade infrastructure. Observed hydrological data for the Middle River catchment was analysed, and a coupled hydrological, water balance and demand model was used in order to determine a historical baseline of system performance over the period of 1961-2005. When modelling the system in response to the historical baseline as a benchmark of expected performance, it was found that in dry years the demand can be high enough to draw down the reservoir below 6 m. The modelled probability of that failure under a historical climate was ~2%. It was also found that following this period of stationary climate there has been an observed increase in PET of 8.3%, indicating that the system is already experiencing a ‘changed’ climate compared to the historical baseline. A climate stress-test of the Middle River water supply system following the foreSIGHT framework demonstrated that the system is sensitive to annual changes in rainfall and PET. These metrics both influence supply and demand, with decreases in rainfall primarily affecting streamflow, and increases in PET increasing the demand placed on the system. In many of these failures, the reservoir had not filled the previous year, therefore providing little warning to enable water augmentation strategies to be put into place. Under more extreme changes in climate, the reservoir cannot fill multiple years in a row, leading to multi-year failures. To establish an understanding of how the climate might have changed in 2030 relative to the historical baseline, multiple lines of evidence were considered including multiple sources of climate model projections for the year 2030, and the recent trend in observed PET. It was found using the scenario-neutral approach that in the climate conditions projected for the year 2030 the probability of failure of the reservoir dropping below 6 m could increase from 2% to 7-8%. Further, the observed trend in PET suggests the probability of failure may have already increased to 5%. This highlights a significant system sensitivity to relatively small changes in climate.