Abstract
This work addresses the performance and projections of climate simulations from CMIP5 global models and CORDEX regional climate models over Northeast Brazil (NEB). For this study, NEB climate was characterized by the following variables: near surface air temperature (TAS), precipitation (PR), potential evapotranspiration (ETo), and aridity index (IA). In verification methodology of the models we utilized statistical metrics such as mean square error, statistical bias, Pearson correlation and concordance index. In general, the ensemble mean of the simulations (M) produces a significant representation of NEB recent (1985-2005) climate. The projections of M for scenarios of future emissions of greenhouse gases, RCP4.5 and RCP8.5, show generalized increase in the temperature, with 2079-2099 (long-term) temperature anomalies over NEB ranging between 2.1 °C (RCP4.5) and 4 °C (RCP8.5). The ETo rates are more likely to increase during the 21st century, with the RCP8.5 long-term mean 15% higher than reference values for recent climate. Projection of precipitation is inconclusive due to significant spread among the individual simulations, with M indicating a anomaly of −1.6% in RCP8.5 long-term mean. Projected IA is reduced in most simulations of future climate, suggesting a greater chance of increased aridity over NEB during this century.
Keywords:
Northeast Brazil; CMIP5; CORDEX