Abstract

Recent experience with drought and a shifting climate has highlighted the vulnerability of urban water supplies to “running out of water” in Perth, south-east Queensland, Sydney, Melbourne and Adelaide and has triggered major investment in water source infrastructure which ultimately will run into tens of billions of dollars. With the prospect of continuing population growth in major cities, the provision of acceptable drought security will become more pressing particularly if the future climate becomes drier.

Decision makers need to deal with significant uncertainty about future climate and population. In particular the science of climate change is such that the accuracy of model predictions of future climate is limited by fundamental irreducible uncertainties. It would be unwise to unduly rely on projections made by climate models and prudent to favour solutions that are robust across a range of possible climate futures.

This study presents and demonstrates a methodology that addresses the problem of finding “good” solutions for urban bulk water systems in the presence of deep uncertainty about future climate. The methodology involves three key steps: 1) Build a simulation model of the bulk water system; 2) Construct replicates of future climate that reproduce natural variability seen in the instrumental record and that reflect a plausible range of future climates; and 3) Use multi-objective optimisation to efficiently search through potentially trillions of solutions to identify a set of “good” solutions that optimally trade-off expected performance against robustness or sensitivity of performance over the range of future climates.

A case study based on the Lower Hunter in New South Wales demonstrates the methodology. It is important to note that the case study does not consider the full suite of options and objectives; preliminary information on plausible options has been generalised for demonstration purposes and therefore its results should only be used in the context of evaluating the methodology. “Dry” and “wet” climate scenarios that represent the likely span of climate in 2070 based on the A1F1 emissions scenario were constructed. Using the WATHNET5 model, a simulation model of the Lower Hunter was constructed and validated. The search for “good” solutions was conducted by minimizing two criteria, 1) the expected present worth cost of capital and operational costs and social costs due to restrictions and emergency rationing, and 2) the difference in present worth cost between the “dry” and “wet” 2070 climate scenarios. The constraint was imposed that solutions must be able to supply (reduced) demand in the worst drought. Two demand scenarios were considered, “1.28 x current demand” representing expected consumption in 2060 and “2 x current demand” representing a highly stressed system. The optimisation considered a representative range of options including desalination, new surface water sources, demand substitution using rainwater tanks, drought contingency measures and operating rules.

It was found the sensitivity of solutions to uncertainty about future climate varied considerably. For the “1.28 x demand” scenario there was limited sensitivity to the climate scenarios resulting in a narrow range of trade-offs. In contrast, for the “2 x demand” scenario, the trade-off between expected present worth cost and robustness was considerable. The main policy implication is that (possibly large) uncertainty about future climate may not necessarily produce significantly different performance trajectories. The sensitivity is determined not only by differences between climate scenarios but also by other external stresses imposed on the system such as population growth and by constraints on the available options to secure the system against drought.

Please cite this report as:

Mortazavi-Naeini, M, Kuczera, G, Kiem, AS, Henley, B, Berghout, B &Turner, E 2013, Robust optimisation of urban drought security for an uncertain climate, National Climate Change Adaptation Research Facility, Gold Coast, 74 pp.

Visit the project page

Banner image Glennies Creek Dam copyright Tim J Keegan, Flickr Creative Commons