Calibration and validation of SWAT model using PSO algorithm for the simulation of runoff and sediment in a mountainous watershed

Soil erosion is a major environmental threat to sustainability and agricultural productivity leading to reduction of soil fertility, loss of nutrients, and declines of crop yields in arable lands. In this study, the suitability of SWAT model for simulating the daily runoff and sediment in the Bazoft watershed (one of the main sub-basins of the Karun river basin in central Iran) was investigated. For this purpose, three different climate data sets including climate data obtained from the Bazoft watershed (Bazoft watershed data), the climate research unit data (CRU data), and a combination of the CRU and Bazoft watershed data (Bazoft-CRU data) were first investigated. Then, the potential use of PSO algorithm for calibrating and validating of the model was evaluated. The constructed SWAT model using a combination of the CRU data and the Bazoft climate data had a better runoff simulation performance in the study area. The obtained Nash-Sutcliff (NS) coefficient for the constructed SWAT model using the Bazoft-CRU data was 0.38 while it was 0.31 and 0.21 for the proposed SWAT model using the Bazoft watershed data and the CRU data, respectively. The obtained r-factor and p-factor values in the model calibration period using the PSO algorithm were 1.25 and 0.81, respectively. The R2 and NS coefficients for the validation period were also 0.64 and 0.60, respectively. Furthermore, in the sediment simulation step, the r-factor and p-factor values for the calibration period were 0.69 and 0.85, respectively, and they were 0.63 and 0.80 for the validation period, respectively.