Evaluating drought risk dynamics: Comparison of a Climate-Informed Multi-time Scale Stochastic (CIMSS) framework to the AR(1) model

Abstract

Persistent drought across much of Australia, increasing water demand due to population growth, climate variability and the drying impacts of anthropogenically-induced climate change have placed increasing stress on our water supply systems. This paper represents an alternative approach to assessing water supply system security through the use of short-term drought risks and dynamic conditional simulation techniques. For the case study based on Hunter Valley hydrological data and reservoir conditions it was found that short-term drought risks were significantly higher compared to long-term drought risks, increasing non-linearly as the system becomes increasingly stressed. It is argued that this short-term drought risk approach is a useful strategic planning tool for water authorities and it represents an advance in thinking and provides a more realistic and informative estimation of drought risk than traditional long-term approaches. In comparing the climate-informed multi-time scale stochastic (CIMSS) framework to the widely-used AR(1) model, it was found that the CIMSS framework estimated short-term drought risks that were up to double that estimated by the AR(1) model, dependent on climate regime. However, the long-term risks showed insignificant differences between the models. This illustrates the value of using stochastic rainfall models that actively capture climate behaviour in addition to the hydrological record. These types of models can demonstrate the markedly climate-dependent nature of dynamic short-term drought risks. Hence, they can be considered more informative of water supply system risks than statistical techniques that rely merely on the hydrological record for their calibration.