Machine Learning for Modeling Vegetation Restoration of Forests Using Satellite Images

Forest fires cause widespread destruction, altering vegetation and land globally, leaving lasting impacts. This study focuses on modeling vegetation recovery in the Zagros forests of western Iran post wildfires, using climatic and environmental factors. The study employs Landsat images to gauge vegetation cover and burn severity, using regression techniques to estimate environmental variables. Several machine learning methods are applied for modeling. Results demonstrate the superior precision and accuracy of the gradient boosting method in reconstructing post-fire vegetation recovery processes.