Srinivasan95a Srinivasan, G., M. Hulme, C. Jones, P. Jones and T. Osborn, 1995a: An evaluation of the spatial and interannual variability of tropical precipitation as simulated by GCMs (Diagnostic Subproject 21). Abstracts of the First International AMIP Scientific Conference, Monterey, California, 71. 
Precipitation is one of the most difficult variables to simulate in a General Circulation Model and arguably one of the most important. The Atmospheric Model Intercomparison Project (AMIP) provides an opportunity to examine the simulation of precipitation in a wide array of models. Monthly precipitation fields produced by a subset of 19 currently available AMIP model experiments are evaluated for the tropical region using a land-only observed dataset for the period 1980-88. The models show large variations in their ability to reproduce observed tropical precipitation, although spatial correlations indicate that some of the models simulate the pattern of observed precipitation fields fairly well. The correlations are strongest during boreal winter (DJF) and weakest during the boreal summer (JJA). Individual models also exhibit a consistent dry or wet bias as compared to the observed precipitation fields. Comparison between model and observed precipitation time series for two Central Pacific locations show that most models are unable to reliably reproduce interannual precipitation variability in this region. The exceptions are the MPI, ECM and JMA models which simulate with a reasonable degree of fidelity the observed precipitation characteristics of this region and generally over the whole tropics-as demonstrated by their high spatial, and relatively good anomaly, correlations.