ABSTRACT Basic framework of flood monitoring Flood inundation mapping and its impact on land use along the Kosi River in Bihar, India, was carried out by integrating geo-spatial technology, statistical modeling, and field validation. Sentinel-1 SAR data were utilized and a hybrid methodology was implemented combining Pareto scaling normalization and Bayesian probability estimation for flood mapping. Simultaneously, Sentinel-2 optical satellite data was subjected to random forest (RF) Forest classification to map changes in land use/land cover (LULC), with an average classification accuracy of 89%. LULC analysis revealed critical transformations of surface water bodies which expanded by 50.7%, while agricultural land decreased by 2.1% between 2019 and 2024. The built-up region in the study area nearly doubled, from 558.77 to 978.76 km2, highlighting an increasing urban pressure, while it was observed that the area of barren land peaked in 2022 due to flood-borne sedimentation by the Kosi River. The 2020 flood year recorded the highest inundation (1,440.53 km2), which correlated with peak rainfall exceeding 1,800 mm during the monsoon. Interestingly, 2024 experienced a high flood extent (670 km2) despite low rainfall (350 mm), indicating non-climatic triggers impacting the flood extent such as embankment breaches or upstream discharge releases. Topographical analysis using a DEM showed that low-lying southern floodplains were more prone to flooding, while northern uplands exhibited a reduced flood susceptibility due to better drainage. Further ground evidence reinforced these observations by identifying flood-related phenomena including embankment breaches, sedimentation and post-flood land abandonment. The majority of these impacts were concentrated in districts like Supaul, Saharsa and Madhepura.
Hazra et al. (Thu,) studied this question.