Comparison of three global optimization algorithms for calibration of the Xinanjiang model parameters

Abstract

The Xinanjiang model, a conceptual rainfall-runoff (CRR) model with distributed parameters, has been successfully and widely applied to flood forecasting of large basins in humid and semi-humid regions of China. With an increasing demand for timely and accurate forecasts in hydrology, how to obtain more appropriate parameters for CRR models has long been an important topic. These models have a large number of parameters which cannot be directly obtained from measurable quantities of catchments characteristics. In this study, three different optimization methods are used to calibrate the Xinanjiang streamflow model: genetic algorithm (GA), shuffled complex evolution of the University of Arizona (SCE-UA) and the recently developed shuffled complex evolution Metropolis algorithm of the University of Arizona (SCEM-UA), using streamflow data of the Shuangpai Reservoir in China. Two different time steps of 1 and 3 hr are used in the analysis. The results indicate that the SCEM-UA algorithm can infer the most probable parameter set and furnish useful information about the nature of the response surface in the vicinity of the optimum. Moreover, there is larger uncertainty for 1 hr forecasting than for 3 hr forecasting. This is significant in assessing risks in likely applications of Xinanjiang models.