Assessment of hydrologic impact of climate change on Ord River catchment of Western Australia for water resources planning: A multi-model ensemble approach

Climate change, a harsh reality of modern civilization, has significant impact on water resources. General Circulation Models (GCMs) can predict long term impact of climate change at large or continental scale with some degree of accuracy. But the locus of interest of hydrologist and decision makers is to evaluate probable impact of climate change at local scale, individual catchment, and or at basin scale. In Australia, limited study has been done to find the impact of climate change on water resources at local scale. In this study, hydrological impact of climate change on the Ord River catchment is assessed using modelled runoff from rainfall projections for two Intergovernmental Panel on Climate Change (IPCC) emission scenarios A2 and B1 for the period of 2046-2065 and 2081-2100. To address uncertainties due to differences in GCMs, a multi-model ensemble approach (with 11 GCMs data) has been adopted. For downscaling GCM data to a 5 km resolution (compatible with catchment modelling), Bureau of Meteorology Statistical Downscaling Model (BoM-SDM) is used. The Land Use Change Incorporated Catchment (LUCICAT) hydrologic model is applied to simulate future rainfall and runoff using downscaled rainfall. The model is calibrated with recently developed 5 km grid rainfall produced by the Bureau of Meteorology, Australia. The model calibrated well at all gauging stations of the Ord River catchment. Observed mean annual runoff and modelled mean annual runoff at all six gauging stations are within +/- 3%. Observed rainfall and runoff data suggest that rainfall and runoff has an increasing trend in the northern part of Western Australia in recent times. Findings suggest that the recent rainfall pattern may continue for mid this century (2046-2065) for both scenarios A2 and B1 with a slight increase of mean annual runoff by around 1.77% for scenario A2 and 1.06% decrease for scenarios B1, compared to hind cast mean annual runoff for the period 1981-2000. But during last part of this century (2081-2100), there might be a decrease of runoff for the Ord River catchment of around 12% for scenario A2, and around 7% for B1 compared to hind cast runoff for period 1981-2000. Out of 11 GCMs, eight have shown reasonably consistent results compared to mean annual runoff modelled form observed rainfall and hind cast mean annual runoff for period 1981-2000. Significant variation in terms of runoff projection has also been observed among the GCMs, indicating considerable uncertainty in applying downscaled GCM data for rainfall and runoff projection.