Development of a Real-Time Hydrologic Model for Early Warning of Underpass Flooding

This study developed a real-time flood early-warning model for the ○○○ underpass in Gwacheon using 1-minute rainfall data. Runoff generation was represented by the SCS-CN loss function and the Clark unit hydrograph, while downstream translation and attenuation were modeled using Lag & K routing to compactly represent travel time delays and peak damping among road sub-catchments. To reflect the uncertainty arising from blockage and capacity reduction by leaves/debris, the storm-inlet efficiency was introduced as a scenario coefficient ranging from 0 to 1. Under configurations comparable to the commercial HEC-HMS model, the proposed model exhibited comparable performance in terms of RMSE, peak discharge, time-to-peak, and total inflow, confirming its reliability. For the real flooding event, calibration targeting the arrival time corresponding to a 10-cm water depth reproduced observations most closely at 85% inlet efficiency; with more conservative assumptions (75-80%), the arrival time occurred earlier, enabling the formulation of operational guidelines from a flood-forecast lead-time perspective. The proposed model proved its practical utility as an operational tool for rapidly estimating the expected arrival times of automated gate closures and establishing warning thresholds for underpasses.