Similaridades e diferenças entre cenários climáticos do CMIP3 e CMIP5 para o nordeste do Brasil.

Abstract

This study compares the reliability of precipitation projections from CMIP3 (scenarios A1B and A2) and CMIP5 (scenarios RCP4.5 and RCP8.5) climate models for the Northeast region of Brazil (NEB). The models were evaluated based on their ability to represent the historical period (1961–1990) and to project future periods (2021–2050 and 2051–2080). Using data from 97 INMET weather stations, model performance was assessed through statistical metrics (R, RMSE, and BIAS) and climate extreme indices, following the ETCCDI methodology. Results showed that CMIP3 models performed better in simulating historical climatology, with higher correlations, lower errors, and more stable bias. CMIP5 models showed greater uncertainty, particularly in simulating convective processes and moisture transport. During the DJF season, CMIP3 models better captured the influence of the Intertropical Convergence Zone (ITCZ). Extreme climate indices indicated a significant reduction in both extreme and total annual precipitation, a pattern more accurately represented by CMIP3 models, especially CNCM3. The increase in consecutive dry days and the decrease in rainy days were also better captured by CMIP3. Future projections indicate a progressive reduction in precipitation, with the most severe impacts in scenarios RCP8.5 (CMIP5) and A2 (CMIP3), suggesting increasing aridification and a shorter rainy season. Scenario RCP4.5 indicates a more moderate transition. Seasonal analysis reinforces the tendency of reduced rainfall during wet months and a prolonged dry season. The findings underscore the need to improve climate models and highlight the importance of regionalized projections. This research provides relevant insights to support public policies and water resource management in NEB in the face of climate change impacts.