A Bayesian approach for estimating bacterial nonpoint source loading in an estuary with limited observations

Area of Science:

• Environmental modeling
• Estuarine water quality
• Bayesian statistics

Background:

• Deterministic models struggle with bacterial nonpoint source uncertainty, leading to inaccurate estuarine concentration simulations.
• Accurate estimation of bacterial loading is crucial for effective watershed management and achieving water quality standards.

Purpose of the Study:

• To develop and apply a Bayesian approach for quantifying uncertainty in bacterial nonpoint source loading estimation within estuarine models.
• To improve the simulation of bacteria concentration in estuaries by incorporating in-stream observations.
• To establish a framework for estimating allowable bacterial loads to meet water quality objectives.

Main Methods:

• Incorporated a Bayesian approach into a tidally averaged estuarine model.
• Utilized Bayes' theorem to create a joint probability distribution for nonpoint source loadings based on estuarine bacteria observations.
• Implemented the approach on a finite difference model to handle estuarine geometry variations and non-linear transport.

Main Results:

• The Bayesian approach successfully estimated nonpoint source loads within acceptable error ranges, even with limited observations.
• Spatial correlations in estuarine observations led to error compensation between adjacent watersheds, improving overall load estimation.
• The method demonstrated feasibility in estimating bacteria sources and developing allowable loads for Holdens Creek.

Conclusions:

• The Bayesian method provides an efficient and robust methodology for assessing nonpoint source contributions in watershed management.
• This approach effectively addresses uncertainty and error issues inherent in estuarine bacterial simulations.
• The study highlights the advantage of using Bayesian inference for improving water quality modeling and management in estuarine systems.