Development and Evaluation of a Real-Time Flood Forecasting System Based on the K-series Integrated Modeling Framework in the Seomjin River Basin

This study evaluates the applicability of a real-time flood forecasting system developed using K-water’s K-series integrated modeling framework, which couples rainfall-runoff simulation, reservoir operation, river hydraulics, and flood inundation analysis. The system was applied to the extreme flood event that occurred in the Seomjin River Basin, South Korea, in August 2020. Catchment runoff was generated using K-DRUM (K-water Distributed Rainfall Runoff Model), reservoir operations were simulated using K-Oper, and river water levels were computed using K-River. The resulting hydraulic outputs were then processed in K-FIM (K-water Flood Inundation Map) to estimate the extent and depth of the flood inundation. Model performance showed high accuracy, with K-DRUM achieving Nash-Sutcliffe efficiency (NSE) values greater than 0.93 for inflow simulations, and K-River yielding NSE values of 0.82-0.93 for water level predictions. The inundation results from K-FIM also reproduced observed flood traces with reproduction rates of 0.73-0.85. These findings indicate that the K-series integrated modeling framework enables quantitative flood-risk assessment and can support early detection and real-time decision-making during flood events in complex hydrologic systems.