deagle77-cloud/forest-fire-risk-chefchaouen

Machine learning and Remote Sensing For Forest Fire Risk Prediction in Chefchaouen, Morocco. # Forest Fire Risk Prediction using Machine Learning and Remote Sensing ## Project Overview This repository presents my Master's research project on forest fire risk prediction in the Province of Chefchaouen, Morocco. The project combines Geographic Information Systems (GIS), Remote Sensing, Google Earth Engine, Python, and Machine Learning techniques to identify areas vulnerable to wildfire and support decision-making for wildfire prevention. ## Objectives - Predict forest fire risk using Machine Learning. - Produce wildfire susceptibility maps. - Analyze environmental and climatic factors influencing wildfire occurrence. - Support sustainable forest management. ## Study Area Chefchaouen Province, Northern Morocco. ## Data Used - Sentinel-2 satellite imagery - MODIS products - Digital Elevation Model (DEM) - Land Cover - Climate variables - Topographic variables (Slope, Aspect, Elevation) ## Methodology - Data preprocessing using Google Earth Engine - Feature extraction - Machine Learning Models: - Random Forest - XGBoost - Gradient Boosting - Accuracy assessment using: - ROC Curve - AUC - Confusion Matrix ## Tools - Google Earth Engine - ArcGIS Pro - QGIS - Python - Scikit-learn - GeoPandas - Rasterio ## Results The project generated wildfire susceptibility maps and evaluated model performance using statistical validation methods. ## Author *Ilyass Fedjikh* M.Sc. Coastal Engineering – Coastal Dynamics & Natural Risks Morocco