Abstract:
|
Acton Lake, a freshwater reservoir in Hueston Woods State Park (Ohio), has experienced poor water quality and high phytoplankton biomass. The abundance of phytoplankton has been regularly collected from the lake since 1994. Over the years, changes in farming practices have influenced the water quality of the lake. New methods for modeling proportional time series are introduced. A likelihood ratio and Bayesian Hidden Markov Model are developed to find change points in a proportional time series. The new method is then applied to the Acton Lake data to detect changes in the proportion of phytoplankton taxa.
|