Danger of Vegetation Fires in the Cerrado-Amazon Transition Region Based on In Situ and Reanalysis Meteorological Data

Fire hazard indices are fundamental for mitigating socioeconomic and environmental damage. This study evaluated the performance of the Ängstrom, FMA, FMA+, EVAP/P, and P-EVAP indices in the Cerrado-Amazon transition region (2010–2022), Brazil, using data from National Institute of Meteorology (INMET) and reanalysis (Copernicus). The efficiency of the models was validated by the Skill Score and Percentage of Success methods, correlating them with the hotspots from the DBQueimadas (INPE). The results reveal climatic seasonality typical of tropical regions, with rainy summers and severely dry winters, with minimum relative humidity below 30%. Although the average annual rainfall is 1662.20 mm, spatial heterogeneity and seasonal water reduction drove a 42% increase in the number of fire occurrences, totaling 3.9 million hotspots in the period. The P-EVAP and FMA+ indices showed greater predictive accuracy, with P-EVAP reaching a Skill Score of up to 0.74, especially with reanalysis data. FMA showed intermediate performance, while Ängstrom and EVAP/P were less reliable. Regionally, the highest sensitivity and accuracy of the indices were observed in Maranhão and Tocantins. It is concluded that regional meteorological variability directly influences the risk of wildfires, with P-EVAP and FMA+ being the most effective tools for monitoring and preventing fires in the region.