The Australian summer monsoon—a model intercomparison study

A model intercomparison study was performed to assess the large-scale characteristics of the contemporary Australian summer monsoon in a total of 9 uncoupled and coupled general circulation model (GCM) simulations in comparison to the ERA-40 Re-analysis and the NCEP-NCAR Re-analysis. While all GCMs adequately reproduced low-level westerlies over the Australian monsoon region during austral summer, most models had some difficulty in reproducing either the spatial distribution or the magnitude of monsoon rainfall, with a dry bias over the monsoon region, highlighting in particular uncertainties associated with convection parameterizations. Most simulations also developed a split intertropical convergence zone. Of those that did not, a low resolution coupled model FOAM (Fast Ocean Atmosphere Model) also produced the most realistic annual cycle of precipitation.

Most GCMs had difficulty in generating intraseasonal variability within the monsoon region at Madden-Julian Oscillation timescales. On the contrary, the similar-to-observed periodicity of the Tropospheric Biennial Oscillation (TBO) and El Niño Southern Oscillation (ENSO) variability in the uncoupled simulations showed that the model atmospheres responded correctly to interannual variations in the surface forcing within the monsoon region. Fully coupled models, too, were quite successful in generating TBO-like variability. The coupled GCMs, however, generated ENSO variability on timescales ranging between 2 and 7 years.

Each of the uncoupled GCMs produced similar-to-observed correlations between Australian monsoon precipitation and tropical sea surface temperatures, as did coupled model simulations with a more realistic ENSO periodicity. Hence, the tendency of some coupled models to generate ENSO with a biennial periodicity is thought to be linked to coupled model deficiencies rather than a strong coupling with the Australian monsoon.