Spatial Modeling of Fire Occurrence Probability in a Protected Area in Western Mexico

Abstract

Forest fires can generate economic losses, social vulnerability, and environmental damage. Therefore, more knowledge about the causes and effects on the territory is needed. The present study aimed to develop a spatial model to determine the influence of socio-economic and environmental variables on the probability of fire in a protected area. The study was developed in La Primavera Flora and Fauna Protection Area, located on the periphery of Guadalajara’s Metropolitan Area in western Mexico. We used remote sensing information and field data of hot spots of fire from 2012 to 2018 to model forest cover's susceptibility to wildfires based on a set of socio-economic and landscape variables in the Dinamica EGO software. We built and evaluated a fire occurrence probability map (AUC = 0.88) and estimated that 20 970.90 ha of forests in the area are prone to wildfires. Population density, communication routes, and agriculture were the variables with the highest weight of evidence. The protected area presents different levels of fire occurrence probability due to factors such as urban growth and land-use cover change. The model allows a spatially explicit visualization of the sites most susceptible to fires, enhancing the prevention and management of forest fires in the study area.