Uncertainty Assessment in Reservoir Water Quality Modeling: Implication for Model Improvement

Abstract

As reservoir simulation models become more widely used, there is greater need for uncertainty assessment in water quality modeling. In comparison with the modeling of quantity phenomena, such as hydrological modeling, water quality modeling involves additional uncertainties in the modeling of pollutant loadings and the transport and fate of contaminants in receiving waters. In this work, a general and flexible method based on generalized likelihood uncertainty estimation (GLUE) is used to estimate the uncertainty in reservoir water quality modeling that arises from parameter uncertainty and error in model inputs. A one-dimensional model which was set up to simulate the hydrothermal and water quality of Pepacton Reservoir, part of the New York City water supply system, was used to demonstrate the method. Obtained results show that most model parameters and inputs follow wide non-Gaussian distributions, indicating they are of high uncertainty. The results also show that uncertainty is low for the simulated water temperatures of the epilimnion and hypolimnion, and dissolved oxygen (DO) of the epilimnion. Unfortunately, the simulation uncertainty for total phosphorus and chlorophyll