AFFORESTATION DEFORESTAION ANALYSIS USING MACHINE LEARNING

Deforestation in the Amazon rainforest poses significant environmental and ecological challenges, necessitating accurate and scalable monitoring solutions. This research utilizes Google Earth Engine (GEE) and machine learning to analyze forest cover changes in a 50,000 km² study area within the Amazon. By leveraging Landsat 8 (2015) and Sentinel-2 (2019) satellite imagery, a Random Forest classifier is applied to classify land cover and detect forest loss and gain over a four-year period. Key preprocessing steps include cloud masking, spectral index computation (NDVI), and spatial resampling to enhance classification accuracy. The study integrates scikit-learn, Pandas, and visualization tools for performance evaluation, including confusion matrices, accuracy assessments, and Cohen’s Kappa statistics. Results indicate the effectiveness of Random Forest in identifying deforestation patterns, providing a reliable method for large-scale environmental monitoring. The findings contribute to data-driven conservation efforts and underscore the importance of cloud-based geospatial analysis in mitigating deforestation.