Estimation of River High Flow Discharges Using Friction-Slope Method and Hybrid Models

Accurately estimating river water flow during floods is crucial to water resource management, dam reservoir operation, and flood mitigation strategies. Although hydrological models for flood prediction have improved, they still face constraints and make inaccurate forecasts. Hydraulic models face uncertainties related to riverbed Manning roughness coefficient and energy slope. This study employs a novel Friction-Slope (α parameter) method to estimate flood discharge. Investigation focuses on three alluvial rivers in Golestan, Iran. The computation method uses the Manning formula and accounts for river energy slope and riverbed Manning roughness coefficient. The α parameter is calculated using easy-to-access river cross-section variables: flow depth, area, and hydraulic radius. SVR-PSO, SVR-GWO, and SVR-RSM hybrid methods are used to achieve this. Calculated river flow discharges are compared to measured data. Statistical evaluation criteria like R 2 , MAE, RMSE, and conformity index determined the hybrid models' optimal structures. The SVR-RSM model had the highest accuracy during testing, with an R 2 value of 0.97, MAE of 0.22, RMSE of 1.66, and d of 0.99. Once the α parameter was determined using the RSM-SVR model, river flow discharges were calculated and compared to observed values. The testing phase produced the most accurate results, with R 2 = 0.88, MAE = 0.15, RMSE = 0.41, and d = 0.98.