Remote sensing–derived land use land cover for hydrological modelling of flood hydrographs in the Mun River Basin, Thailand

Event-based hydrological models in tropical basins often rely on static land use land cover (LULC) inputs, limiting their ability to represent rapidly changing watershed conditions. This study integrates multi-temporal, satellite-derived LULC data from Google Earth Engine (GEE) with the Hydrologic Engineering Center–Hydrologic Modeling System (HEC-HMS) to improve rainfall–runoff simulations in the Mun River Basin, Thailand. Annual LULC maps (2016–2024) were generated from Sentinel-2 imagery using a Random Forest (RF) classifier, achieving overall accuracies of 0.80–0.91 and Kappa values of 0.71–0.87. The most accurate map (2017) was combined with soil data to derive Curve Number (CN) values for model parameterization. HECHMS was calibrated using four major flood events and validated with three independent events across six stations, showing strong performance in both periods. The findings demonstrate that incorporating updated, cloud-based LULC inputs significantly improves flood hydrograph simulation in rapidly changing tropical basins.