Comparative analysis of the ability of a set of <scp>CMIP3</scp> and <scp>CMIP5</scp> global climate models to represent precipitation in South America

ABSTRACT The purpose of this study is to evaluate the ability of two sets of global climate models (GCMs) derived from the Coupled Model Intercomparison Projects Phase 3 ( CMIP3 ) and Phase 5 ( CMIP5 ) to represent the summer, winter, and annual precipitation mean patterns in South America south of the equator and in three particular sub‐regions, between years 1960 and 1999. Different metrics (relative bias, spatial correlation, RMSE , and relative errors) were calculated and compared between both projects to determine if there has been improvement from CMIP3 to CMIP5 models in the representation of regional rainfall. Results from this analysis indicate that for the analysed seasons, precipitation simulated by both CMIP3 and CMIP5 models' ensembles exhibited some differences. In DJF , the relative bias over Amazonia, central South America, eastern Argentina, and Uruguay is reduced in CMIP5 compared with CMIP3 . In JJA , the same occurs in some areas of Amazonia. Annual precipitation is also better represented by the CMIP5 than CMIP3 GCMs as they underestimate precipitation to a lesser extent, although in NE Brazil the overestimation values are much larger in CMIP5 than in CMIP3 analysis. In line with previous studies, the multi‐model ensembles show the best representation of the observed patterns in most seasons and regions. Only in some cases, single GCMs [ MIROC3 .2(hires) – CMIP3 – and MIROC4h – CMIP5 ] presented better results than the ensemble. The high horizontal resolution of these models suggests that this could be a relevant issue for a more adequate estimation of rainfall at least in the analysed regions.