Abstract: Forest fires disrupt the stability of forest ecosystems, and scientifically evaluating the driving factors of forest fires is crucial for effective prevention. This study focuses on Anning City in Yunnan and uses kernel density analysis to investigate the spatial patterns of forest fires. We established, refined, and validated a Structural Equation Model (SEM) to identify the driving factors of these fires. Results reveal that fire incidents are concentrated in the northern part of Anning City, with a tendency to expand throughout the region. The spatial clustering of fires is dispersed, indicating that no single factor drives the occurrence of fires. The adjusted SEM identified and weighted several factors: nighttime light index (0.329), population density (0.278), GDP (0.201), distance to roads (0.192), maximum sustained wind speed (0.344), average wind speed (0.322), dew point (0.288), precipitation (0.106), elevation (0.412), aspect (0.369), and slope (0.219), the model has been validated and the predicted results are accurate. These results offer valuable data and theoretical insights for predicting fire occurrence probabilities and assessing fire risks.

Key words: forest fire, spatial distribution, driving factors, structural equation model, fire risk