A Bayesian hierarchical modeling approach can improve measurement accuracy of microcystin concentrations

The Bayesian hierarchical model (BHM) is a framework that improves parameter estimation by leveraging information from different sources. In an environmental monitoring program, we often measure important chemical concentrations using calibration-based methods. These methods require fitting a calibration curve repeatedly each time with a small number of standard solutions of known concentrations. This approach is often associated with large estimation uncertainty in the measured concentrations. BHM is a perfect method for reducing calibration curve uncertainty, thereby enhancing the accuracy and stability of the resulting concentration measurements. We demonstrate the effectiveness of a BHM approach by estimating microcystin concentrations from the Lake Erie harmful algal bloom (HAB) monitoring program operated by the Great Lakes Environmental Research Laboratory of the National Oceanic and Atmospheric Administration. We introduced a sequential updating algorithm to implement the BHM framework so that the BHM model can be fit and updated one test at a time. By comparing estimated quality control sample concentrations to their known values, we show that the BHM method yields the best accuracy compared to the currently used methods. Due to the sequential updating approach, the BHM can be readily incorporated into a lab without requiring additional changes to lab procedures, thus offering a key advantage over traditional calibration methods. This advancement could reduce health risks and false-positive shutdowns during HAB events.