Hydrological models play a crucial role in water resources management, but their simulation results are affected by various uncertainties, such as aleatory and epistemic uncertainties. To enhance model accuracy and applicability, this study comprehensively investigated parameter sensitivity and uncertainty in the Distributed Xin’anjiang Model, employing the Nash–Sutcliffe efficiency coefficient (NSE), relative error (RE), and Kling–Gupta efficiency (KGE) as objective functions. Daily hydrometeorological data spanning from 2016 to 2020 were collected for the catchment area upstream of the Xixian Hydrological Station in the Upper Huai River Basin. Firstly, the data set was partitioned into a calibration period (2016–2019) and a validation period (2020). After calibration, the simulation results of the model closely matched the measured data. Subsequently, the Sobol sensitivity analysis method was employed to assess the reliability of the Generalized Likelihood Uncertainty Estimation (GLUE) method results, determining the sensitive parameters and their posterior distribution ranges under different objective functions. Finally, we compared the impact on interval prediction accuracy when sensitive parameters were assigned prior distribution ranges versus posterior distribution ranges under different objective functions. The results indicated that parameters KC represents the ratio of evapotranspiration, and IM denotes the percentage of impervious areas in the catchment remain sensitive regardless of the objective function used. Adopting the posterior distribution ranges of sensitive parameters can improve the precision of uncertainty intervals, although their specific performance varies. During high-flow periods, when the objective function was NSE, the uncertainty intervals achieved the highest precision, with the percentage of observations bracketed by the unit confidence interval (PUCI) and the mean prediction interval center deviation (MPICD) values of 0.25 and 77.68 m3/s, respectively. However, KGE demonstrated significant advantages in capturing the hydrological dynamics during flood peaks and their adjacent periods, where its accuracy was notably superior to that of other objective functions. The results of the uncertainty and sensitivity analyses help reduce uncertainties in the modeling process and provide theoretical guidance for applying and calibrating the Distributed Xin’anjiang Model in hydrologically complex regions with high flood risks.
Sun et al. (Sat,) studied this question.