Evaluating Monthly Flow Prediction Based on SWAT and Support Vector Regression Coupled with Discrete Wavelet Transform

Reliable and accurate streamflow prediction plays a critical role in watershed water resources planning management. We developed new hybrid SWAT-WSVR model based on 12 hydrological sites the Illinois River (IRW), U.S., that integrated Soil Water Assessment Tool (SWAT) with Support Vector Regression (SVR) calibration method coupled discrete wavelet transforms (DWT) to better support modeling watersheds limited data availability. Wavelet components of simulated from SWAT-Calibration Uncertainty Procedure (SWAT-CUP) precipitation time series were used as inputs SVR build SWAT-WSVR. examined performance potential compared it observations, SWAT-CUP, SWAT-SVR using statistical metrics, Taylor diagrams, hydrography. The results showed average RMSE-observation’s standard deviation ratio (RSR), Nash–Sutcliffe efficiency (NSE), percent bias (PBIAS), root mean square error (RMSE) is 0.02, 1.00, −0.15, 0.27 m3 s−1 0.14, 0.98, −1.88, 2.91 validation sites, respectively. Compared other two models, proposed possessed lower discrepancy higher accuracy. rank overall three SWAT-based models during whole study period was > SWAT-CUP. supplies an additional approach can improve accuracy SWAT simulation data.