Fire represents a preva‎lent natural hazard worldwide, leading to the devastation of trees, plants, and significant economic and environmental harm. Effectively managing fire risk and its ensuing environmental repercussions involves predicting and mapping fire occurrences in rangelands and natural areas. This study employs MaxEnt models and Bayesian belief networks (BBNs) to model and map fire risk, focusing on the Benroud Zanganeh area in Fars province. Utilizing 122 fire occurrence points from the past decade, 70% were designated as a training set for model construction, and 30% served as a test set for validation. Physiographic factors (slope, slope direction, elevation, and surface curvature), climatic variables (rainfall and temperature), human factors (distance from roads and residential areas), land use, and normalized differential vegetation index were incorporated to model fire occurrences. The study area was classified into four fire risk zones: low-risk, moderate risk, high-risk, and very high-risk. The MaxEnt model identified 12.13% and 51.21% of the area as having very high and low fire risk, respectively. In contrast, the BBNs model indicated 24.95% with very high fire risk and 10.49% with very low fire risk. The MaxEnt model highlighted distance from residential areas, distance from roads, elevation, and land use as the most influential factors in fire occurrence, while slope direction, surface curvature, and normalized differential vegetation cover index (NDVI) had minimal impact. Response curve analysis revealed a higher probability of fire occurrence in the west of the region. BBNs analysis emphasized the significant influence of slope, distance from residential areas, distance from roads, and land use on fire occurrence. The BBNs model, with an area under the receiver operating characteristics curve (AUC) of 0.96, outperformed the MaxEnt model (AUC 0.86). Although both models demonstrated acceptable performance in predicting fire occurrences, these findings offer valuable insights for forest and rangeland managers and disaster management departments. The results facilitate informed decision-making for implementing effective mitigation measures to safeguard valuable natural resources from future fires

حسين بشري , مصطفي تركش اصفهاني

مسعود سميعي

رضا جعفري , محسن احمدي