On the assessment of the impact of reducing parameters and identification of parameter uncertainties for a hydrologic model with applications to ungauged basins

In this paper, we investigate model parameter uncertainties associated with hydrological process parameterizations and their impacts on model simulations in the Three-Layer Variable Infiltration Capacity (VIC-3L) land surface model. We introduce an alternative subsurface flow parameterization into VIC-3L to reduce the impacts of model parameter uncertainties on model simulations by reducing the number of model parameters that need to be estimated through a calibration process. The new subsurface flow parameterization is based on the concepts of kinematic wave and hydrologic similarity, and has one parameter for calibration. Results from the 12 MOPEX (Model Parameter Estimation Experiment) basins obtained by applying the VIC-3L model with the new subsurface flow formulation show that the performance of the new parameterization is comparable to the original subsurface flow formulation, which has three parameters for calibration. In addition, a probabilistic approach based on Monte Carlo simulations is used to evaluate model performance and uncertainties associated with model parameters over different ranges of streamflow. Studies based on the 12 MOPEX watersheds show that compared to the parameter associated with the new subsurface flow parameterization, the VIC shape parameter (i.e. the b parameter that represents the shape of the heterogeneity distribution of effective soil moisture capacity over a study area) has a larger impact on model simulations and could introduce more uncertainty if not estimated appropriately. Furthermore, investigations on the b parameter suggest that the ensembles (i.e. the mean response and its bounds) from the Monte Carlo simulations could provide reasonable predictions and uncertainty estimates of streamflows, which have important implications for applications to ungauged basins. The study also shows that appropriate reduction of the number of model parameters is an effective approach to reduce the impacts of parameter uncertainties on model simulations. This is more so for applications to ungauged basins or basins with limited data available for calibration. The new subsurface flow parameterization and the probabilistic uncertainty analysis approach are general and can be applied to other modeling studies.