Recently, the impacts of climate change have become increasingly evident, affecting both population and economic sectors, highlighting the need for developing future-oriented adaptation strategies in response to the predicted intensification of its impacts. Climate models’ projections are powerful tools for evaluating future climate change impacts and developing adaptation strategies. However, potential biases within these projections may affect climate risk assessment in hydrology and agriculture. In this way, this study aimed to find a suitable bias-correction method from the Coupled Model Intercomparison Project Phase 6 climate and extreme climate variables for Brazil, providing the best models for each climate variable. We evaluated the performance and error of two types of bias-correction methods commonly applied in the literature: linear scaling and quantile mapping. The results showed that the non-parametric quantile mapping methods performed better for most climate variables. The linear scaling method performed slightly better for the maximum consecutive dry days index in certain models and regions. Nevertheless, the improvement was minimal compared to the raw data, indicating that bias correction has limited capacity to improve indexes that climate models represent poorly. The best models varied according to the climate variable, but ACCESS-ESM1-5, NorESM2-MM, CanESM5, EC-Earth3, and CMCC-ESM2 predominated in the variables’ ranking after bias correction. Our study supplies insight into the suitable bias-correction method and model selection across different variables and the Brazilian region. This level of detailed information promotes informed decision-making in climate risk assessment studies regarding agriculture, energy production, society, and disasters.
All Science Journal Classification (ASJC) codes
- Global and Planetary Change
- Environmental Science (miscellaneous)
- Economic Geology
- Computers in Earth Sciences