Predictive Modeling of E. coli Levels at Urban Beaches along Lake Michigan
Mentor 1
Todd R Miller
Mentor 2
Gabriella Pinter
Location
Union Wisconsin Room
Start Date
27-4-2018 1:00 PM
Description
Sewage contamination of natural water bodies is a serious problem inhibiting usage of recreational aquatic environments such as beaches. The bacterium Escherichia coli is often found in the feces of warm blooded animals, so high levels of E. coli in water suggest contamination with sewage. As such, the United States Environmental Protection Agency uses E. coli as an indicator to assess the risk of acquiring a gastrointestinal illness due to using recreational waters. Both growth-dependent and growth-independent molecular assays require hours to measure E. coli in water. Therefore, significant delays exist between the time of sampling and time when warnings are issued to the public. Thus, results are not indicative of the current conditions. In this study, we explored the use of statistical models to predict current conditions based on real-time limnological and meteorological conditions. To do this we sampled Lake Michigan’s water over the summer of 2017 in order to monitor E. coli concentrations at popular beaches. We then created predictive statistical models with computer software (Virtual Beach v3.0) using E. coli data from previous years, and paired it with hourly recorded environmental conditions obtained from a nearby buoy, and the Great Lakes Forecasting System. We will present a comparison between measured data and our models, and discuss potential changes that can be made to improve predictions.
Predictive Modeling of E. coli Levels at Urban Beaches along Lake Michigan
Union Wisconsin Room
Sewage contamination of natural water bodies is a serious problem inhibiting usage of recreational aquatic environments such as beaches. The bacterium Escherichia coli is often found in the feces of warm blooded animals, so high levels of E. coli in water suggest contamination with sewage. As such, the United States Environmental Protection Agency uses E. coli as an indicator to assess the risk of acquiring a gastrointestinal illness due to using recreational waters. Both growth-dependent and growth-independent molecular assays require hours to measure E. coli in water. Therefore, significant delays exist between the time of sampling and time when warnings are issued to the public. Thus, results are not indicative of the current conditions. In this study, we explored the use of statistical models to predict current conditions based on real-time limnological and meteorological conditions. To do this we sampled Lake Michigan’s water over the summer of 2017 in order to monitor E. coli concentrations at popular beaches. We then created predictive statistical models with computer software (Virtual Beach v3.0) using E. coli data from previous years, and paired it with hourly recorded environmental conditions obtained from a nearby buoy, and the Great Lakes Forecasting System. We will present a comparison between measured data and our models, and discuss potential changes that can be made to improve predictions.
