Probabilistic Prediction of Microcystin Concentrations in Lake Erie

Session: Harmful Algal Blooms and Their Toxicity: Remote Sensing and Modeling Approaches (1)

Craig Stow, NOAA Great Lakes Environmental Research Laboratory, craig.stow@noaa.gov
Freya Rowland, Cooperative Institute for Great Lakes Research, University of Michigan, frowland@umich.edu
Song Qian, University of Toledo, song.qian@utoledo.edu
Thomas Johengen, CILER, University of Michigan, johengen@umich.edu
Mark Rowe, NOAA GLERL, mark.rowe@noaa.gov
Qianqian Liu, Grand Valley State Univ., liuqianqian0622@gmail.com
Eric Anderson, NOAA/GLERL, eric.j.anderson@noaa.gov

Abstract

NOAA’s Great Lakes Environmental Research Laboratory and the Cooperative Institute for Great Lakes Research have maintained a weekly monitoring network in western Lake Erie since 2008 to assess the seasonal progression of harmful algal blooms.  The network has included approximately eight fixed stations, and occasional “bloom-chasing” expeditions, at which a suite of water quality characteristics, including microcystin concentration, have been measured. These data display a consistent relationship between microcystin and chlorophyll a concentrations, which we have quantified using a Bayesian hierarchical modeling framework. The model is a piecewise linear (aka “hockey stick”) structure with components to include spatial differences and changes with time. Analogous data from Saginaw Bay exhibit similar structure and can be included in the Bayesian framework to increase predictive precision. The model provides a basis for the probabilistic prediction of microcystin concentration, conditional on laboratory measured chlorophyll. Additionally, the posterior parameter distribution can be used as prior information to constrain predictions using satellite-based chlorophyll measurements with the goal of short-term toxin prediction using remotely-sensed observations.