Evaluation for the continent of Australia of the simulation of the surface climate using the Biosphere/Atmosphere Transfer Scheme (BATS) coupled into a global climate model

Abstract

The Biosphere Atmosphere Transfer Scheme (BATS) of Dickinson et al. (1986) has been coupled to the National Center for Atmospheric Research's Community Climate Model (CCM1). The land surface climatology generated by averaging the results of a 3 yr model integration on a monthly basis is evaluated for the continent of Australia by comparison with published descliptions of a wide range of parameters. The major inconsistencies in the BATS simulahon are associated with surface hydrology: modelled precipitation is too high in the central desert area and the large maxima of rainfall on the coast are grossly underestimated. This may be entirely a result of the coarse resolution of the GCM grid but, since rainfall is of crucial importance to agricultural and water supply activities, the inconsistencies clearly merit further consideration. Further problems associated with the surface hydrological regime include the difficulty of comparing total modelled evapotranspirat~on with evaporation calculated from evaporation pan measurements. In a semi-arid to arid environment such as Australia the latter are gross overestimates. Other differences between the BATS climate and measurements are confined to the central desert area and the crop and short grass regions in the south: simulated maximum temperatures are too high in the desert and minimum temperatures marginally too high in the southeast. Surface net radiation appears to be underestimated in the desert by up to 50 W m-' (probably due to the combined effect of overestimated temperatures and, more importantly, overestimated cloud amounts). Some of these secondary inconsistencies could be partially the result of the poor simulation of surface hydrological regimes which, in turn, seems likely to be primarily the fault of the coarse spatial resolution of the model. Proposals are outlined for improved methods of validation and testing the predictions of such complex biospheric submodels, at least at continental scale.