Coupled model simulations of current Australian surface climate and its changes under greenhouse warming: an analysis of 18 CMIP2 models

Coupled climate models have been extensively used to further our understanding of the dynamics and physics of the Earth's climate system and the potential changes of regional and global climates in the future, especially due to human activities such as fossil fuel burning and land-use activities. Nevertheless, there are still large uncertainties in our knowledge of the global climate system and in our representations of such a complex system. The confidence of our projected future climate change, therefore, inevitably depends on how well the current climate is simulated by coupled climate models and how large the scatter is among the model simulations of current and future climates. As one of the diagnostic subprojects within the Coupled Model Intercomparison Project phase II (CMIP2), we present an evaluation of 18 CMIP2 coupled model simulations over the Australian region. Monthly rainfall and surface air temperature climatologies over the Australian region have been derived from the 18 CMIP2 control simulations and compared with observations from the Australian Bureau of Meteorology. The gross spatial patterns of austral summer rainfall (DJF) are reasonably simulated by the majority of the models. However, there are significant model errors in simulating the intensity and location of the heavy Australian monsoon rainfall in the north and eastern parts of the continent, with about half of the models showing more than 100 mm/month biases and a number of models simulating wrong locations of the monsoon rainfall. The seasonal cycle of the surface temperature is reasonably reproduced in the models although there are biases of around 2-4 degrees C present in the model simulated surface air temperature climatology. Based on the 80-year model simulations of perturbed climate, with 1% per year increase of atmospheric CO2 concentration, the changes of surface air temperature and precipitation have also been analysed. The average annual surface temperature change in the last 20-year period of the model simulations against the model control simulations over the Australian region varies from 1.00 degrees C to 2.18 degrees C, with an ensemble average of 1.59 degrees C and 0.33 degrees C scatter measured by one standard deviation. The models give a mixed signal in predicting averaged Australian rainfall changes, with some models simulating more than 3 mm/month increase while others show more than 4 mm/month decrease with on average no change. The spatial distributions of the model-simulated surface temperature and precipitation changes have also been analysed. Surface temperature is increased over the whole continent in all models, while the changes in precipitation show large spatial variations. The ensemble mean model shows decreases in winter rainfall across southern Australia and over northwestern Australia during summer. Increased rainfall is simulated over parts of eastern Australia during winter, extending further north during summer. Besides the analysis of changes in mean climate, the potential impacts of global warming on Australian climate variability is explored in a preliminary way by analysing the changes in tropical Australian precipitation correlations with surface temperature variations over four key oceanic regions. Results suggest that the influence of tropical and subtropical sea-surface temperature (SST) forcing on the Australian climate may change under greenhouse warmin