Field Estimation of Manning’s Roughness and Its Application to CSA Discharge Computation during Flood Events

River discharge is a key indicator for water resources management and flood forecasting; however, the traditional single stage–discharge rating curve used for its estimation produces systematic errors under unsteady flow conditions due to hysteresis. In this study, field-measured Manning’s roughness coefficients (n) are first estimated at Naju Bridge on the Yeongsan River by combining H-ADCP–measured discharges with water-surface slopes derived from upstream and downstream stage observations, using the continuous slope–area (CSA) framework in inverse form. The resulting 10-min n time series for the 2020 flood events is then segmented by stage to represent cross-sectional controls on roughness. These stage-wise n segments are subsequently applied to the CSA method to compute discharge time series for the 2019 and 2021 flood events, and the estimates are validated against observations. The estimated n exhibits a consistent stage-dependent pattern, including a rapid decrease at low stages, convergence at intermediate stages, an inflection point near the onset of rapid cross-sectional expansion, and an increase at high stages, reaching n ≈ 0.08–0.09 during extreme floods—values higher than those from conventional empirical formulas and design criteria. Using the measured and stage-segmented n, the CSA-based discharge estimates successfully reproduce hysteresis across six flood events, achieving R² ≥ 0.94 and peak error ≤ 3%, although nRMSE exceeds 10% under low-flow conditions. Overall, applying field-derived roughness substantially improves CSA discharge estimation and supports the practical use of roughness monitoring and CSA-based computation in rivers subject to unsteady flow.