A Determining the Suitable Algorithm to Calibrate SWAT Model for Daily-Runoff Simulation: A Case Study of Bazoft Watershed, Southwestern Iran

Distributed models such as soil and water assessment tool (SWAT) are increasingly being used for an optimum watershed management, especially, for management of the water resources and thus it is important for these models to pass a careful calibration and uncertainty analysis. To fulfil this demand, the potential use of three calibration procedures including Particle Swarm Optimization (PSO), Sequential Uncertainty Fitting algorithm (SUFI-2), and Generalized Likelihood Uncertainty Estimation (GLUE) was investigated to calibrate SWAT model for dailly runoff simulation in a mountainus watershed (the Bazoft watershed, central Iran). Two indices including coefficient of determination (R2) and Nash-Sutcliffe (NS) were used to quantify the accuracy and efficiency of calibration-uncertainty performance. Using PSO algorithm for the calibration of the model resulted in a better runoff simulation compared to that of using GLUE and SUFI-2 algorithms. The NS value for the calibrated-SWAT model, using the PSO algorithm, was 0.62 while it was 0.48 and 0.52 for the calibrated models using the GLUE and SUFI-2 algorithms, respectively. It can be concluded that using PSO algorithm to calibrate SWAT model for dailly runoff simulation in watersheds with limited input data (such as that in our case) may lead to more suitable calibration results.