Integrating Environmental Variables and Machine Learning for Wildfire Susceptibility Prediction in Portugal

Wildfires constitute a significant ecological disturbance within Mediterranean ecosystems, exerting profound effects on forest dynamics, biodiversity, and land management practices. The development of precise susceptibility mapping is essential to inform prevention strategies, optimize resource allocation, and promote sustainable forest management by increasing fire pressure. This study employed and compared four machine learning classifiers—Random Forest, Classification and Regression Trees (CART), Gradient Boosting, and Extreme Gradient Boosting (XGBoost)—to model wildfire susceptibility across Portugal. Six environmental and anthropogenic predictors were utilized: vegetation indices, land use/land cover, slope, elevation, wind speed, and distance to settlements. The results indicated that vegetation-related variables, particularly NDVI and land cover, were the most significant determinants of fire occurrence, followed by slope and wind speed, thus underscoring the role of biophysical conditions in shaping fire regimes. Among the evaluated models, XGBoost demonstrated the highest predictive performance (overall accuracy = 92.98%, AUC = 0.98), surpassing or equalling the other ensemble methods. The resulting susceptibility maps identified the northern and central interior regions as the most fire-prone, consistent with historical fire records. Our findings underscore the efficacy of ensemble machine learning techniques in capturing complex fire–environment interactions and provide spatially explicit information that can enhance fire prevention planning, support conservation priorities, and guide adaptive forest management in Mediterranean landscapes.