Assessment of Rainfall and Flood Prediction Accuracy in the Seomjin-River Basin Using the Local Data Assimilation Prediction System

This study aimed to assess the feasibility of utilizing rainfall predictions from numerical forecasting systems for managing frequent localized flash floods caused by climate change. A flood event in the Seomjin River Basin in August 2020 was selected to evaluate the accuracy of rainfall predictions from the Local Data Assimilation Prediction System (LDAPS) and verify the reliability of discharge predictions using the forecasted rainfall. The forecasted rainfall data were combined at 6-, 12-, 24, and 48-h intervals for validation. The 6-h combination demonstrated the highest reproducibility, while the 12-hour combination provided satisfactory results. Thus, discharge modeling using LDAPS provides relatively reliable results for short-term predictions; however, long-term predictions require improvement. The findings of this study can facilitate the evaluation of the potential of LDAPS for flood prediction and management.