Application of Sentinel-1 and 2 Imagery for Rapid and Robust Flood Detection: A Case Study of Flood Event in Nepal.

Floods are among the most destructive natural disasters, resulting in significant loss of life and property for millions of people around the world. Extreme flood events in the foothills of the Himalayan ranges and their forelands are closely linked to heavy monsoonal rainfall, steep slopes, and excessive surface runoff from the uphills. Floods hazards in Nepal have become increasingly devasting due to improper land use planning, unplanned settlement distribution, deforestation, land degradation in the upstream watershed, topography, geological setting and climate change. Nepal was hit by an unprecedented late monsoon rainfall, causing widespread landslides and flooding across the country in September 2024y, resulting in significant loss of life and property. This study investigated the use of Sentinel-1 Synthetic Aperture Radar (SAR) with Ground Range Detected (GRD) scenes  for rapid and robust flood detection during the September 2024 flood events in Kathmandu valley and the surrounding areas. The study area is of utmost interest as it comprises diverse geographical setting on the basis of topography and geological setting and these floods events have a significant impact on settlement, infrastructure and other environmental processes. In the study, a standard workflow was applied for the pre-processing of both the products. Based on the application of pre- and post-SAR imagery, this study estimated the extent of flood inundation, highlighting the major impacted area based on pre- and post-land cover map of the study area using machine learning (ML) algorithms and compare the changes with spectral indices. The change detection and Normalized Difference Flood Index (NDFI) was evaluated using threshhold value of temporal Sentinel-1 GRD data. High resolution Google Earth imagery was used for the accuracy assessment of pre flood environment; post flood site data was evaluated from field visit. Greater level of flood impacts were noted both within the Kathmandu valley (Kathmandu. Bhaktapur, Lalitpur district) and outside the valley Banepa, Dhulikhel, Panauti, Namobuddha, Roshi local area of Kaverepalanchok district; Sunkoshi, Golanjor , Phikkal local areas of Sindhuli district of the study area. The overall accuracy of flood inundation mapping was 95 % and the accuracy of land cover map was evaluated about 88 %. A detailed land use/ cover map of the study area was prepared to present the changes post-flood environment using Sentinel -2 Multi-spectral imagery. Further, Permanent water body (PWB) using Normalized Difference Water Index (NDWI) algorithm and Normalized Difference Vegetation Index (NDVI) were prepared for the evaluation of the post-flood impact area . Overall, the analysis inferred that watershed level flooding vulnerability results from natural factors like heavy rainfall and topography, which are further intensified by human activities such as infrastructure development, urbanization and poor land management.