Effects of Parameter Uncertainties on Forecast Accuracy of Xinanjiang Model

Due to the incomplete understanding of hydrologic processes and insufficient information of watershed characteristics, uncertainties are inevitably involved in watershed simulation and modeling. Uncertainty analysis is conducted to Xinanjiang model, a widely used conceptual hydrological model in China, to investigate the uncertainties in its forecast performance affected by uncertain parameters. In particular, Latin Hypercube sampling technique is applied to gain insights about stochastic nature of selected model performance indicators and relative importance of parameters. Such information is essential in directing parameter calibration and improving accuracy of Xinanjiang model.