Waterlogging and Drainage Congestion in the Kosi Fan of theHimalayan Foreland

This study uses optical and radar satellite images (i.e., Landsat and Sentinel-1 & 2) to monitor seasonal waterlogging in the Kosi Fan from the 1987-2021 period. We used Google Earth Engine (GEE) platform to process the satellite images. We applied Random Forest (RF) classifier to classify the image pixels that correspond to waterlogging from optical and microwave images. The optical images detect the waterlogging patches more accurately (70-80%) as compared to the radar images (50-65%). We observed that the waterlogging patches located along the road and stream networks show a high probability of occurrence. We have computed the probability of occurrence of waterlogging patches near the road and stream network intersection. The result indicates a high correlation of the occurrences of waterlogging patches in the proximity of structural interventions (rail, road, network embankment, etc) on the Kosi Fan.