Assessing the Transferability of a Basin-Scale HEC-RAS Rain-on-Grid Model Across Independent Hurricane Events

Reliable simulation of hurricane-induced flooding is challenging because rainfall–runoff generation, river flow, and floodplain inundation interact across large basins. This study evaluated the cross-event transferability of a fully distributed, two-dimensional HEC-RAS Rain-on-Grid (RoG) model for the Tar–Pamlico River Basin in eastern North Carolina. The model integrates terrain, land cover, soil-infiltration properties, and gridded rainfall within a hydrodynamic framework. It was calibrated using Hurricane Matthew (2016) and was independently validated for Hurricane Florence (2018) using unchanged parameters. Performance was assessed against observed discharge, flood depth, and inundation extent. Across streamflow gages, calibration yielded Nash–Sutcliffe efficiency (NSE) values of 0.73–0.99, coefficients of determination (R2) of 0.78–0.99, and ratios of root mean square error to observation standard deviation (RSR) of 0.10–0.51. Independent validation produced NSE values of 0.63–0.98, R2 values of 0.81–0.99, and RSR values of 0.15–0.60, indicating consistent performance despite differing hurricane rainfall and flood responses. The validated model was subsequently applied to a 500-year design storm to generate basin-scale flood-hazard information. The findings demonstrate that a calibrated basin-scale RoG model can retain useful predictive capability across independent hurricane events and support flood-hazard assessment and emergency planning in hurricane-prone river basins.