Master of Science (MS), Bowling Green State University, 2015, Applied Statistics (ASOR)

This paper examines hazardous algae blooms in Lake Erie, focusing on previously created predictive statistical models, and creating different predictive models based on two proxy measurements for hazardous algae bloom occurrences – dissolved oxygen and chlorophyll-a. While prior models have used different proxies for hazardous algae blooms, including remote sensing and boat tows, the study presented here examines whether different proxies, a larger dataset, and different independent variables create valid hazardous algae bloom predictive models and/or improve upon prior forecasting methods. More specifically, since there is no single definition for hazardous algae blooms, and no one agreed upon metric to measure them, this study examines whether the chosen proxies are suitable proxies for hazardous algae blooms in Lake Erie, using linear regression and ANOVA analyses to create a number of different models. The results from these models indicate that both dissolved oxygen and chlorophyll-a are suitable proxies for hazardous algae bloom occurrences. Further, the modeling results confirmed the Lake indicators that are the greatest contributors to hazardous algae blooms, and confirmed prior research that the Lake had changed in terms of hazardous algae bloom growth and occurrence after the mid-1990s. Following these results, the paper examines the public policy response to recent blooms. Combining the results from this and prior studies, the public policy response was scrutinized, and the paper concludes that more will likely need to be done in the future to mitigate bloom occurrences and severity.

Committee: Nancy Boudreau Ph.D. (Advisor); John Chen Ph.D. (Committee Member); Sheila Roberts Ph.D. (Committee Member) Subjects: Applied Mathematics; Environmental Law; Public Policy; Statistics