Drought projections for Australia: Updated results and analysis of model simulations

Abstract To meet increasing demand for information on future drought hazard to help Australia build resilience and preparedness under a changing climate, we developed new information on drought projections for Australia and four sub-regions based on the natural resources management (NRM) zones. The information reported here includes: two drought indices (the Standardised Precipitation Index, SPI, and the Standardised Soil Moisture Index, SSMI); four drought metrics (percent time spent in droughts, mean drought duration, mean drought frequency, and mean drought intensity); and two drought categories (drought and extreme drought). The projections are developed from CMIP5 global climate model simulations of rainfall and soil moisture for the historical (1900–2005) and future (2006–2100) climates. The multi-model results project significant increases in all the drought hazard metrics, except frequency, with larger changes in the SSMI compared to SPI. The more severe drought hazard under climate change is apparent over a larger area than previously indicated, particularly in southern and eastern Australia. Although the majority of modelling results indicate more severe drought conditions, the range in the results is large, mainly because of the uncertainty in the global climate model rainfall projections. A projected decrease in rainfall results in a projected increase in drought severity (which is further enhanced by the increase in potential evapotranspiration), and a projected increase in rainfall results in a projected decrease in drought severity (moderated by the increase in potential evapotranspiration). The assessment of the ability of models to reproduce historical observations does not show clusters of models that best simulate all the different drought metrics. Unlike previously assumed, the results show that the models that best reproduce the observed rainfall are not necessarily best in simulating the drought metrics. For this reason, all the models are used here to estimate the multi-model median and range of results. The large uncertainty in the projections can be confusing to end users and present challenges in adapting to climate change. The presentation and communication of projections here will also go some way towards overcoming this challenge.