Various hydrological models have been applied to the management of water resources and water quality. However, parameter uncertainty is of perpetual interest in the application of hydrological models. In this context, the HSPF model was constructed and calibrated using monthly observed stream data from 1998 to 2010 in the Chaohe River watershed, northeast of Beijing. Specifically, the sensitivity and uncertainty of the model parameters were investigated by the GLUE algorithm with the PEST platform. The major results were illustrated as follows:① the hydrological simulation shows good performance with Nash-Sutcliffe efficiency of 0.84 and 0.55 in the period of calibration and validation, respectively; ② the parameters were divided into three categories:global sensitive parameters (LZSN, INFILT, IRC, and AGWRC), regional sensitive parameters (UZSN), and non-sensitive parameters (DEEPFR, BASETP, AGWEPT, INTFW, and CEPSC); ③ strong correlations were detected within the sensitive parameters, which further involved significant negative correlations (LZSN~INFILT, INFILT~UZSN, and UZSN~AGWRC) and a positive correlation (LZSN~UZSN) and (UZSN~AGWRC); ④ the equifinality for different parameters was found in the HSPF model, indicating that parameter sets determine the simulation performance rather than individual parameters; ⑤ among various external factors, precipitation was identified as the most important condition for simulation uncertainty; and ⑥ the temporal difference in simulation performance was considered using annual, seasonal, and monthly scales with simulation precisions of 81.80%, 78.70%, and 80.56%, implying that the annual scale might be the optimal simulation period with higher accuracy. This research result is useful for the application and localization of the HSPF model.
Keywords: Chaohe River watershed; HSPF model; hydrological simulation; multiple temporal scales; uncertainty.