As part of an Australia-China climate change bilateral project, in this study we report on a regional model, RegCM3, high-resolution climate simulation over the whole Australian continent. A 30-year model integration with a 20 km horizontal resolution is performed. As the model uses completely different configurations to the model being used in Australia for climate change studies, this study provides a complementary approach for exploring and assessing the skill of dynamical downscaling for this country. In the analysis, we explore model skill in three areas: how well the model simulates the observed surface climatology and if it captures some locally driven features; how skilful the model is in simulating climate variability in the region and the spatial coherence seen from observations; and whether the model has skill in simulating daily climate statistics. Despite some model deficiencies, such as an overestimation of monsoon rainfall, the overall results are encouraging. Detailed features in observed rainfall, temperature and surface evaporation are well simulated, including the extent of summer monsoon rainfall, the winter rainfall maximum in the southwest of Western Australia (SWWA) and the topographically driven high rainfall belt along the Great Dividing Range. The model gives reasonable skill in capturing the observed climate variability in the region, with about 10 per cent error of observed magnitudes. Over SWWA and the Murray-Darling Basin (MDB), it is skilful in reproducing the observed probability distribution functions of rainfall, maximum and minimum temperatures, except that the model tends to show lower probability of producing light rain events, and biases in rainfall and temperature.
Song et al. (Mon,) studied this question.