Evaluation of temperature and precipitation simulations of a subset of CMIP6 models for Brazil

Abstract

This article examines the ability of 40 CMIP6 climate models to simulate the mean temperature and mean precipitation observed in each state in Brazil from January 1985 to December 2014. Despite the simulation biases, the CMIP6 was able to simulate the annual and seasonal behavior and capture the trend of the analyzed variables. For surface temperature, the subset of the CMIP6 showed the best performance for the South and Southeast Brazil regions, highlighted in Paraná, Santa Catarina, and São Paulo states. Midwest Region has the best performance for precipitation. In relation to temperature trends over the 30-year study period, 1985–2014, all states in Brazil showed an increasing trend, both for the subset of the CMIP6 and observed data. However, precipitation in the North and Southeast regions did not show a clear trend for the analyzed period, since some states showed an increase whereas some showed a reduction. However, the states in the Central-West and Northeast regions show a trend toward a reduced precipitation. On the other hand, in the South Region, all states have increasing precipitation, as supported by observed and simulated data. Even with the inherent uncertainties in climate modeling and detected biases, this study is relevant to show the behavior of the present models nowadays climate simulations. The evaluation presented in this work established confidence levels to use the subset CMIP6 model in future climate studies, as detailed in this paper.