Sentinel-2 image classification of mangrove populations change from cyclone Amphan to locate communities in the Sundarbans most affected

Increasing anthropogenic activities have accelerated the degradation of mangrove populations in the Sundarbans, leading to reduced buffer zones between residential areas and flood zones. This study analyzed the impacts of cyclone Amphan on communities and mangroves in the Sundarbans through remote sensing and machine learning classification. Spatial data for Cyclone Amphan was collected from Sentinel-2 Level 2A satellite imagery using Google Earth Pro. The imagery was processed in Google Earth Engine (GEE), where indices such as NDVI, NDMI, and NDWI were calculated. A Random Forest classifier was trained, with 70% of the data allocated for training and 30% for validation of model performance. The FAO Forest Proximate People (FPP) 1.0 dataset was then analyzed by overlaying communities located within 1 kilometer of the mangrove loss. The classification reached a producer accuracy, user accuracy, and F1 score of 0.99 for the mangrove class. The analysis showed a net loss of 201.99 square kilometers of mangroves from the cyclone. Further integrating the FPP layer and visualizing all results on a GEE map revealed the locations of communities most vulnerable to the impacts of future cyclones. The findings confirm that cyclone Amphan caused significant mangrove loss in the Sundarbans, exacerbating the vulnerability of nearby communities. Image classification effectively captured cyclone-induced changes, though limitations such as spectral resolution constraints and the lack of ground validation remain. Strengthening conservation efforts is critical to protecting mangroves as a natural defense measure against the increasing intensity of cyclones.