Hunan Province, as a region frequently affected by forest fires, has significant importance in understanding the spatiotemporal distribution of forest fire occurrences for effective emergency response. This study examines the dynamic changes in forest fires in Hunan Province from 2001 to 2022 using MODIS fire hotspot data. By integrating meteorological data, we construct a Logistic Regression based forest fire risk probability model. During 2001–2022, the number of fire hotspots in Hunan Province initially increased and then fluctuated and decreased, with the hotspots concentrated in the southern and southwestern regions of the province. The forest fire risk probability model established in this study demonstrates a high accuracy with an AUC value of 0.916, which effectively reflects forest fire risks. Under different warming scenarios (1.5°C/2°C/3°C), the risk of forest fires significantly increased. Compared to the reference period (2016–2022), the area with extremely high risk increased by 2.4%, 9.8%, and 30.8%, respectively.