This study evaluates the performance of the ERA5 and ERA5-Land reanalysis datasets in representing daily maximum and minimum temperatures in the semi-arid region of northeastern Brazil—an area highly vulnerable due to its climatic and socioeconomic conditions. Validation was conducted by comparing reanalysis data with local station observations using statistical metrics such as Root Mean Square Error (RMSE), Pearson’s correlation coefficient (ρ), Mean Square Error (MSE), coefficient of determination (R²), and bias. The results indicate that ERA5 provides a more accurate representation of maximum temperatures, with RMSE values below 1.6 °C and Pearson correlation coefficients above 0.8, outperforming ERA5-Land. ERA5 consistently emerges as the superior model for maximum temperature, showing stronger agreement with observed data (ρ > 0.90; RMSE < 1.5 °C in most cases) and significantly lower bias. For minimum temperature, although ERA5 maintains robust performance, ERA5-Land demonstrates greater accuracy in specific stations and time periods, particularly in terms of bias (reaching − 0.05 °C), suggesting that model selection for minimum temperature should consider location and temporal context. Furthermore, the results confirm that ERA5 successfully captures the region’s known spatial climate pattern, more accurately representing milder temperatures in coastal cities and more pronounced thermal events in remote and elevated areas. By validating the reliability of ERA5 data in a region with limited observational coverage, this study provides a solid foundation for researchers and policymakers, enabling the confident use of reanalysis data in local-scale climate impact models. This contributes to the development of more robust and evidence-based adaptation strategies for key sectors such as agriculture and water resources.
Ferreira et al. (Mon,) studied this question.