Improving Semi-Distributed Rainfall-Runoff Model Accuracy in the Prasae Basin Using High-Resolution Radar Rainfall Data

Abstract

The reliability of rainfall input data is a crucial factor influencing the accuracy and precision of rainfall-runoff modelling. High-resolution spatial and temporal radar rainfall data offer a promising alternative for integration with semi-distributed rainfall-runoff models that simulate flow hydrograph at the sub-catchment scale. This investigation employs the URBS model to simulate runoff at the Z.11 runoff station, located at the Khlong Prasae in the Prasae River Basin, Rayong Province. The study aims to compare the accuracy of runoff simulations using different rainfall input data: rain gauge rainfall data versus radar rainfall data. The rain gauge rainfall data were analyzed from an automatic rain gauge network and interpolated into spatial rainfall using the Inverse Distance Weighting method to derive sub-basin rainfall data. Meanwhile, the radar rainfall data were obtained through a composite method incorporating data from the Sattahip and Samut Songkhram radar stations. The results of model calibration and validation with runoff data of six events during the years 2022 and 2023, indicate that runoff estimation using radar rainfall data superior accuracy compared to rain gauge rainfall data, which are spatially limited. This improvement is reflected by statistical indices that are consistently higher than those obtained from rain gauge rainfall data across all cases, with NSE values ranging from 0.67 to 0.95 and KGE values ranging from 0.60 to 0.96, indicating good to very good performance. Furthermore, the parameter sets derived from radar rainfall data exhibit lower variability than obtained from rain gauge data, demonstrating that radar rainfall data can significantly enhance the accuracy and efficiency of runoff simulations.