A comparison of GCM-simulated and observed mean January and July precipitation

Abstract

A high-resolution global precipitation climatology (developed by Legates and Willmott) is used to evaluate the simulated January and July precipitation fields of the GFDL, OSU, GISS and UKMO general circulation models (GCMs). Legates and Willmott's climatological estimates were derived from raingage observations and the gage biases were minimized. These estimates are spatially averaged to the resoltuion of each GCM and differences between the GCM-simulated field and the climatological averages are computed and mapped. Zonal averages for ten-degree bands also are examined. Precipitation rates along the Intertropical Convergence Zone (ITCZ) simulated by all four GCMs are considerably lower than the climatological estimates in both months. Moreover, inadequate representation of the seasonal migration and latitudinal extent of the ITCZ result in an underestimation of wet-season precipitation and an overestimation of dry-season rainfall. In the Southern Hemisphere, the January mid-latitude maximum is inadequately simulated and it is overestimated in July. Northern Hemispheric patterns generally are better simulated than those in the Southern Hemisphere. The spectrally-based GFDL model substantially overestimates polar precipitation while it is more accurately represented by the grid-point GCMs. Regional errors are commonly quite large (in many areas they exceed 2 mm day −1) which suggests cautious use of current-generation GCM prognostications for local- or regional-scale climate change studies.