Stochastic Modeling And Uncertainty Assessment For Watershed Water Quality Management PDF Download

Complex watershed water quality models have been increasingly used to support Total Maximum Daily Load (TMDL) development. However, systematic approaches for addressing the significant simulation uncertainty are lacking. For TMDLs supported by complex watershed models, defining the margin of safety (MOS) component through a rigorous uncertainty analysis remains a significant challenge. This study aimed to develop (1) a systematic approach of uncertainty analysis for complex watershed water quality models in the watershed management context; and (2) a framework for defining the MOS with an explicit consideration of uncertainty and degree of protection. A global sensitivity analysis technique was first applied to select critical model parameters. A framework for sources of uncertainty and their interactions was built. Based on this framework, Generalized Likelihood Uncertainty Estimation (GLUE) was initially evaluated as a potential approach for conducting stochastic simulation and uncertainty analysis for complex watershed models. The limitations of GLUE became evident, which led to the development of a new Bayesian approach, Management Objectives Constrained Analysis of Uncertainty (MOCAU). The concept Compliance of Confidence (CC) was then introduced to bridge the gap between modeling uncertainty and MOS. An optimization model was also developed for cost-minimized TMDLs. This study used WARMF as an example of a complex watershed model and constructed a synthetic watershed for developing and testing methodologies. The methodologies were also implemented to study the diazinon TMDL in the Newport Bay watershed (southern California). This research contributes to the theory of stochastic watershed water quality modeling, as well as to the practices of managing watershed water quality.