Global Wildfire Susceptibility Mapping Based on Machine Learning Models

Wildfires are a major natural hazard that lead to deforestation, carbon emissions, and loss of human animal lives every year. Effective predictions wildfire occurrence burned areas essential forest management firefighting. In this paper we apply various machine learning (ML) methods on 0.25° monthly resolution global dataset wildfires. We test the prediction accuracies four different fire classifiers: random (RF), eXtreme Gradient Boosting (XGBoost), multilayer perceptron (MLP) neural network, logistic regression. Our best ML model predicts with over 90% accuracy, compared approximately 70% using then train regression models predict size obtain an MAE score 3.13 km2, 7.48 km2 linear To our knowledge, is first study be conducted in such dataset. use developed shed light influence factors areas. suggest building upon these results create ML-based weather indices.