Characterization of Parameter Uncertainty in SWAT Model using MCMC Bayesian Framework: The Case of Naryn River Basin

Abstract

Analysis of parameter uncertainty in distributed watershed model is a worldwide challenge. In this study, The Differential Evolution Adaptive Metropolis (DREAM) technique is developed to analyse the uncertainty of Soil and Water Assessment Tool (SWAT) model parameters. SWAT is used for providing the basic hydrologic simulation, DREAM algorithm is employed to approximate the posterior distributions of model parameters with Bayesian inference. DREAM is then used to capture the uncertainty and implications of parameters in the Naryn River Basin (in Central Asia). The posterior distribution of parameters is obtained. Results shows that: (i) the posterior sampling results of DREAM algorithm are satisfactory; (ii) concentrated precipitation during rainy season generates more runoff; (iii) more precipitation exists in the form of snowfall.