Austral summer precipitation interannual variability and trends over Southeastern South America in CMIP5 models

Abstract

ABSTRACTThe purpose of this study is to assess the ability of Coupled Model Intercomparison Project 5 (CMIP5) models in reproducing the variability and change of the austral summer precipitation observed in Southeastern South America (SESA) along the 20th century and beginning of the 21st. Models show a reduction in mean precipitation biases and inter‐model dispersion, and a significant improvement in the representation of the leading pattern of precipitation interannual variability (EOF1), in comparison with Coupled Model Intercomparison Project 3 (CMIP3) models. Changes of the EOF1 activity in the present climate, as represented by both, climate model simulations and rainfall gridded datasets, evidence an increase of the frequency of EOF1 positive events (associated with positive precipitation anomalies in SESA and negative ones in the South Atlantic Convergence Zone) and a decrease of the frequency of EOF1 negative events. Nevertheless there are still large uncertainties due to model differences and the internal variability of the climate system. In order to reduce the impact of model uncertainties, an ensemble of the climate simulations that represent better the features associated with EOF1 activity was built, regardless to which model they correspond. The results obtained with this ensemble confirm that largest precipitation trends in SESA are those represented by climate simulations associated with an increase (decrease) of EOF1 positives (negative) events. It was also found that positive precipitation trends in SESA resulted from climate simulations forced by anthropogenic sources are the largest and significantly different from those from simulations forced by natural sources only, which are not significantly different from zero.