Enhancing Flood Simulation in Data-Limited Glacial River Basins through Hybrid Modeling and Multi-Source Remote Sensing Data

Due to the scarcity of observational data and intricate precipitation–runoff relationship, individually applying physically based hydrological models machine learning (ML) techniques presents challenges in accurately predicting floods within data-scarce glacial river basins. To address this challenge, study introduces an innovative hybrid model that synergistically harnesses strengths multi-source remote sensing data, a (i.e., Spatial Processes Hydrology (SPHY)), ML techniques. This novel approach employs MODIS snow cover sensing-derived glacier mass balance calibrate SPHY model. The primarily generates baseflow, rain runoff, snowmelt melt runoff. These outputs are then utilized as extra inputs for models, which consist Random Forest (RF), Gradient Boosting (GDBT), Long Short-Term Memory (LSTM), Deep Neural Network (DNN), Support Vector Machine (SVM) Transformer (TF). reconstruct relationship between streamflow. performance these six is comprehensively explored Manas River basin Central Asia. findings underscore SPHY-RF performs better simulating daily streamflow flood events than other five models. Compared model, significantly reduces RMSE (55.6%) PBIAS (62.5%) streamflow, well (65.8%) (73.51%) floods. By utilizing bootstrap sampling, 95% uncertainty interval established, effectively covering 87.65% events. Significantly, substantially improves simulation struggles capture, indicating its potential enhance accuracy prediction offers framework robust forecasting basins, offering opportunities explore extreme warming climate.