Enhancing Incomplete Runoff Records With WRF ‐Simulated Rainfall for Water Resources Management

ABSTRACT

Rainfall–runoff modeling supports the prediction of potential floods, but accurate estimates depend on high‐quality observations that many watersheds lack, leaving incomplete and gapped runoff records. This study explores whether simulated runoff driven by rainfall generated under plausible atmospheric conditions can help fill such gaps. We focused on the extensive 6648 km ² Chungju Dam watershed in South Korea, using the Weather Research and Forecasting (WRF) model. Seventeen distinct rainfall scenarios were produced for each annual flood event using various atmospheric configurations, and the Génie Rural à 4 paramètres Horaires (GR4H) model was used to simulate runoff for comparison with observations. The 17 scenarios bracketed the observed peak‐flow magnitudes and timings within the ensemble range, although the WRF rainfall systematically underestimated event totals and peak intensity. Frequency analysis showed that observed quantiles remained within the WRF‐generated range across return periods, but this containment reflects the ensemble spread rather than unbiased skill. These findings suggest that WRF‐generated rainfall can support water‐resources planning in watersheds with incomplete runoff records, provided the WRF rainfall bias is characterized or corrected. This was demonstrated for a large, monsoon‐dominated, comparatively well‐observed watershed; smaller, data‐sparse, or convection‐dominated basins require further validation.