Whetton, P., M.H. England, S.P. O'Farrell, I.G. Watterson and A.B. Pittock
Climatic Change, 33, 497-519, 1996

KeyWords
General-circulation model. Increasing carbon-dioxide. Atmosphere model. Transient-response. Variability. Precipitation. Simulation. Gases. Co2.

Abstract
The extent of agreement amongst current global climate models (GCMs) on the global pattern of rainfall change simulated under enhanced greenhouse conditions is assessed. We consider the results of five experiments which use a simple mixed layer ocean formulation and five which use a fully dynamic ocean model ('coupled experiments'). For many regions of the northern hemisphere there is strong agreement amongst both mixed layer and coupled experiments on the sign of simulated rainfall change. However, in the southern hemisphere there are large, and apparently systematic, differences between the coupled and mixed layer experiments. In particular, whereas the mixed layer experiments agree on simulated rainfall increase in summer in the tropics and subtropics of the Australian sector, the coupled experiments agree (although more weakly) on rainfall decreases. These differences appear to relate to the much reduced warming simulated by the coupled experiments in the high latitudes of the southern hemisphere. However, recent oceanographic evidence suggests that this suppressed warming may be considerably overestimated. We conclude therefore that despite the in-principle advantages of coupled models, it may be too soon to base some regionally specific climate change scenarios solely on the results of coupled experiments.