You are here:
Model Uncertainty Analysis and the Margin of Safety
Citation:
Hantush, Mohamed M., H. Zhang, R. Camacho-Rincon, E. Ahmadisharaf, AND Y. Mohamoud. Model Uncertainty Analysis and the Margin of Safety. Chapter 9, Total Maximum Daily Load Development and Implementation: Models, Methods, and Resources. American Society of Civil Engineers (ASCE), Reston, VA, , 271-306, (2022).
Impact/Purpose:
This book chapter summarizes stat-of-the-science and state-of-the-art in methods for the estimation uncertainty of water quality models and the selection of margin of safety in TMDL development. States may find these methods useful in determining TMDLs while dealing with uncertainty in water quality data and models.
Description:
This chapter covers the state-of-the-practice on the selection of margin of safety (MOS) in total maximum daily loads (TMDL) and the state-of-the-art on model uncertainty estimation and risk-based MOS determination. It provides a summary of sources of uncertainty, approaches for MOS, a survey of MOS types implemented in practice, and advanced probabilistic methods that practitioners may find helpful in future TMDL development. Uncertainty analysis quantifies predictive capability of a model and therefore a necessary step after calibration and validation of the model. An explicit inclusion of the MOS is performed by setting aside a fraction of the calculated assimilative loading capacity to account for uncertainty in the TMDL study. Several TMDL studies have considered statistical models including linear regression. Bayesian methods utilize Bayes theorem to formally modify the prior assumptions on model parameters and to update the probability distributions of the parameters and model outputs as additional information becomes available.
