Reservoir inflow flood forecasting using near realtime satellite precipitation and rainfall-runoff model for Ta Trach Reservoir

Abstract

Reliable precipitation data are critical for rainfall–runoff modeling and flood forecasting, yet rain gauge networks remain sparse in many developing regions. This study evaluates the suitability of three near-real-time satellite precipitation products—GSMaP-NRT, GPM-IMERG Early Run, and PERSIANN PDIR-Now—for forecasting reservoir inflow to the Ta Trach watershed in central Vietnam. A linear scaling bias-correction method was applied to reduce systematic errors before using the datasets as inputs to the HEC-HMS hydrological model. Performance was assessed using the Nash–Sutcliffe efficiency (NSE), coefficient of determination (R²), and relative volume error (RVE). Bias correction significantly improved rainfall estimates and streamflow simulations for all products. Among them, GPM-IMERG Early Run achieved the highest accuracy, with NSE = 0.85, R² = 0.84, and RVE = 4.95%. These findings demonstrate that bias-corrected GPM-IMERG Early Run provides reliable rainfall forcing for operational reservoir inflow forecasting and supports flood management in data-scarce basins.