Deciphering Flood Inundation Zones Using Multitemporal SAR Datasets for Central Brahmaputra River Basin
摘要
Floods in the Brahmaputra River basin are a recurring hazard, leading to significant loss of life, property, and agricultural productivity. This study aimed to map flood inundation in the middle Brahmaputra River basin using Sentinel-1A Synthetic Aperture Radar (SAR) data for the years 2021 and 2022. The primary objective was to assess the spatial extent of flood-affected areas and the impact on different land cover types. Flood-specific indices, including the Normalized Difference Flood Index (NDFI) and Normalized Difference Vegetation Flood Index (NDFVI), were applied to delineate water bodies and flooded areas. Sentinel-1 images were selected based on flood events recorded in the region, processed using radiometric terrain correction, and classified into water and non-water categories. The results revealed two significant flood events in each year. In 2021, the first event on July 15th affected 2740.59 km2, while the second event on September 1st and 3rd inundated 2122.15 km2 and 1012.39 km2, respectively. In 2022, the first event, spanning May 23rd, 25th, and June 6th, affected a total area of 2754.07 km2. The second major event, occurring between June 16th and July 24th, inundated a significantly larger area of 6928.37 km2, with the largest inundation of 5218.88 km2 on June 16th. The analysis showed that agricultural lands were the most affected land cover type, with over 70% of flooded areas impacting crops. Other affected areas included scrubland, barren land, forest, and settlements. The study’s results demonstrate the effectiveness of Sentinel-1 data and flood-specific indices for rapid and accurate flood inundation mapping. By using SAR data, floods could be monitored in near real-time, even under cloudy conditions, providing crucial information for disaster management and mitigation efforts. The methods applied here can be integrated into regional flood risk assessments, with the potential for automating flood detection using machine learning techniques in future research.