Multi-Year Assessment of Flood Exposure on Aman Rice Cultivation Using Sentinel-1 SAR and Sentinel-2 Data in Kurigram District, Bangladesh

Abstract

Bangladesh is highly vulnerable to flooding which poses a serious threat to rice production systems. Aman rice cultivation is highly dependent on monsoon rainfall and seasonal inundation. This study developed an integrated remote sensing and GIS-based framework to assess flood exposure of Aman rice cultivation in Kurigram District, Bangladesh, during 2020-2025. Sentinel-1 Synthetic Aperture Radar (SAR) and Sentinel-2 optical data were integrated within the Google Earth Engine (GEE) platform to generate annual Aman rice distribution and flood inundation maps. The SAR backscatter, spectral bands and vegetation indices were used for rice mapping with a Random Forest (RF) classifier, while Sentinel-1 SAR-based change detection was used for flood mapping. The annual classifications produced overall accuracies of 74.73–85.56% and Kappa coefficients of 0.495–0.709, indicating variable classification agreement among the study years. Annual Aman rice areas varied between 949.51 and 1,486.73 km², with cultivation concentrated mainly in low-lying floodplain regions. Flood exposure showed considerable temporal variation. The highest exposure was observed in 2020, affecting 13.54% of the mapped Aman rice area, followed by increasing exposure during 2023-2025. The findings demonstrate that agricultural flood exposure is determined not only by the total extent of inundation but also by its spatial correspondence with cultivated land. This framework demonstrates the potential of integrating crop mapping and flood monitoring for agricultural exposure analysis.