PROBABILITY OF OCCURRENCE OF FOREST FIRES IN APURÍMAC

Abstract

Forest fires are a global concern due to their negative impact on ecosystems and the life they harbor. In Peru, between 2013 and 2017, approximately 200,000 hectares burned, with the Apurímac region being one of the most affected. The aim of this study was to develop a wildfire probability map, based on 1312 records collected between 2003 and 2022, using Geographic Information Systems (GIS) and MaxEnt machine learning algorithm. 17 predictor environmental variables were used (11 climatic, two orographic and four related to human activity), statistically selected from a total of 31 variables. The predictive capacity of the model was ‘good’, with an Area Under the Curve of 0.819. 5.4 % (1134.2 km2) and 15 % (3171.54 km2) of the surface of Apurímac, mainly in the north, present a ‘very high’ and ‘high’ probability of occurrence of forest fires, respectively. The ecosystems of Andean scrub (561.25 km2), agricultural zone (277.01 km2) and humid puna grassland (230.37 km2) present the largest areas with ‘very high’ probability of occurrence of forest fires. GIS and MaxEnt are useful and important tools in making preventive decisions and fighting forest fires.