Identifying major sources of uncertainty in watershed water quality modeling: an application of the Deterministic Equivalent Modeling Method (DEMM)

Watershed water quality models are widely used in management practices such as Total Maximum Daily Loads (TMDLs). However, the modeling often involves significant uncertainty, especially in addressing non-conventional pollutants on which both knowledge and data are very limited. In this study, the Deterministic Equivalent Modeling Method (DEMM), incorporating the Probabilistic Collocation Method (PCM), is used as an efficient alternative to conventional Monte Carlo Simulation (MCS) for uncertainty analysis. DEMM is one of the Response Surface Methods (RSMs) which calculates uncertainty in output variables based on the direct effect of every uncertain input parameter. This study aims to 1) examine the applicability of DEMM to complex watershed models; 2) develop strategies for identifying major uncertainty sources in watershed water quality modeling. A case study of watershed diazinon (an organophosphorus pesticide) pollution modeling is explored. The results demonstrate that the stochastic response of the output variables to the uncertain input parameters can be adequately approximated by DEMM. A low-order DEMM can save a great amount of CPU time, compared to MCS. Also, DEMM can be used for parameter sensitivity analysis and preliminary model validation without a full calibration of the watershed model. Overall, if designed appropriately, DEMM can be used to identify major sources of uncertainty in watershed water quality modeling, and provides useful information on improving the modeling.