Improving Watershed Modeling with Aggregation of the Main Hydrological Components in SWAT Model

More accurate knowledge of the water balance components in the watersheds will help a better understanding of hydrological behavior of the watersheds. In this study, the effect of two components, river discharges and actual evapotranspiration, were investigated in modeling of the Neyshabour watershed (9500 km2) using SWAT (Soil Water Assessment Tool). SWAT was calibrated first with the river discharges values and then with the combination of river discharges and actual evapotranspiration obtained based on remote sensing. The model performance was validated in predicting these components for a separate time period. The sensitivity analysis results showed that parameters such as available water content, soil evaporation compensation coefficient, and plant uptake compensation factor were most sensitive to changes of input model parameters. The results showed as well that in three hydrometric stations of Andarab, Kharve Majmoo, and Hossein Abad the root mean square error for river discharges predictions on the validation period varied between 0.06 to 0.19 m3/s and 0.02 to 0.09 m3/s respectively for the models calibrated based on river discharges and combination of river discharges and actual evapotranspiration. The use of both mentioned components, i.e. river discharge and actual evapotranspiration, was prominently affected the predicting actual evapotranspiration. The root mean square error for actual evapotranspiration prediction in three selected subbasins varied between 10 to 18.52 mm/month and 6.84 to 7.82 mm/month for the models calibrated based on river discharges and combination of river discharges and actual evapotranspiration, respectively. Therefore, simultaneous use of river discharge and actual evapotranspiration improved the accuracy of modeling.