Assessing forest fire likelihood and identification of fire risk zones using maximum entropy-based model in Khyber Pakhtunkhwa, Pakistan

Abstract

Pakistan has a limited forest coverage, with a significant portion, approximately 40%, concentrated in the Khyber Pakhtunkhwa (KP) region. This highlights the regional significance of KP in terms of forest wealth within the country. The substantial utilization and excessive exploitation of forests have negatively affected the ecosystems. This study aimed to focus on the environmental and social variables and their contribution to the onset of forest fires in KP using Maximum Entropy Model (Maxent). MODIS active fire data history from 2000 to 2022 was studied to establish the relation between forest fire likelihood and environmental conditions. The variables under study included raster data of temperature, wind, precipitation, elevation, slope, aspect, and population density with 2.5-min resolution accessed from Worldclim. The area under curve (AUC) fire probability value was determined to be 0.833, suggesting strong performance of the model. The jackknife analysis indicated the highest contribution of wind (34.2%) followed by precipitation (33.7%) and temperature (18.9%). Maxent was also used to study the potential fire risk zones. It was observed that 53% of the study area is under high-risk, 12% under moderate-risk, and 35% under low-risk. High-risk areas include Abbottabad, Mansehra, Battagram, Shangla, and some parts of Buner and Haripur. These results can prove to be helpful insight in developing preventive strategies for more focused fire management plans that can help reduce fire risk by considering environmental and socioeconomic conditions.