Evaluation of nested climate change models in the study of extreme events

Climate models are an essential tool in climate change studies. However, before using those models in future climate analysis, their ability to describe the present climate must be evaluated. This study assessed the performance of the nested models Eta-HadGEM2-ES and Eta-MIROC5 in the probabilistic description of extreme weather events for precipitation and air temperature (minimum and maximum) in the state of Sao Paulo-Brazil. The generalized extreme value distribution (GEV) was used to estimate the probability of extreme events simulated by the climate models. These probabilities were compared with those observed from weather stations (1961-2005). Using two goodness-of-fit tests (Anderson-Darling and Lilliefors), it was possible to verify that the GEV can be fitted to the model's data for at least 85% of the state of Sao Paulo. Altogether, both Eta-HadGEM2-ES and Eta-MIROC5 need a correction factor to address the numeric difference between the studied area's GEV parameters. The Eta-MIROC5 had the lowest deviations compared to the Eta-HadGEM2-ES, indicating a better performance describing extreme events in Sao Paulo. The Eta-MIRO5 and Eta-HadGEM2-ES models are promising tools for analyzing climate change's influence on the occurrence and frequency of extreme events in Sao Paulo.