LAUSR

Calibration is an important step in most hydrological modeling processes because it helps produce reasonable results. This study aims to investigate the seasonal sensitivity of streamflow parameters and to evaluate the ability of a season-based multi-objective approach to calibrate the Soil and Water Assessment Tool (SWAT) model. The primary goal was achieved through an integrated approach. A variance-based global sensitivity technique, Sobol’ method, was used to evaluate the seasonal sensitivity of streamflow parameters. For the multi-objective approach, three objective functions were considered: the Nash–Sutcliffe efficiency, Nash–Sutcliffe efficiency of logarithmic transformed discharge, and relative bias. The model performances of the season-based multi-objective approach MOO(II), based on these functions and flow duration curves during wet and dry seasons, were compared to three other methods: SOO(I), a conventional single-objective approach to the entire series; SOO(II), a season-based single-objective approach; and MOO(I), a multi-objective approach for the entire series. The four methods were assessed using the SWAT model to predict daily discharge in the Jinjiang basin in southeastern China. The results showed that sensitivity of model parameters varied between the wet and dry seasons. The seasonal calibration approaches, MOO(II) and SOO(II), showed significantly better simulation performances during the dry season while the multi-objective approaches produced more accurate simulations of different aspects of the hydrograph, including peak and low flows and overall water balance, compared to the single-objective methods. MOO(II) captured the seasonal variation of hydrological processes best, compared to the other methods, and the parameter values it identified demonstrated significant seasonal variations.