MATHEMATICAL MODELS TO PREDICT Escherichia coli CONCENTRATIONS AT WISCONSIN BEACHES (MULTIPLE YEAR COMPARISIONS)

Water quality management;Water quality Mathematical models;Lakes Monitoring;Escherichia coli

Manchirala, SreenivasMcDermott, Colleen ;

University of Wisconsin

关键词: Water quality management;    Water quality Mathematical models;    Lakes Monitoring;    Escherichia coli;   

Others  :  https://minds.wisconsin.edu/bitstream/handle/1793/46730/Manchirala%2c_Sreenivas_Thesis_%28With_TP%29.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: University of Wisconsin

【 摘 要 】

Beach water quality is determined using Escherichia coli as an indicator organism for fecal contamination. Traditional methods may result in improper beach closures and openings, as they base their decisions on previous day E. coli concentrations. To overcome these problems mathematical models were developed using various environmental and water quality variables, such as water temperature, wave height, wind speed and direction, turbidity, and rainfalls to predict E. coli concentrations. Mathematical models were developed using ?Virtual Beach? (VB) software, which uses multiple linear regressions to determine the relationship between E. coli concentrations and environmental variables. Data were collected from eleven Wisconsin beaches in Ashland, Bayfield, and Door counties during summer 2008. Mathematical models were developed using 2008 and the 2007 ? 2008 combined data, which were compared with 2007 models. Mathematical models were used to calculate the variability in loge E. coli concentrations in terms of adjusted R2 (coefficient of determination). The Kreher Park Beach ? 2008 model was the only one, which predicted loge E. coli concentrations 100% accurately, without false positives or false negatives. Single variable models were developed for Ephraim Park, Lakeside Park, and Otumba Park beaches for the 2008 recreational season. The combination of years of data (2007 & 2008) did not improve the fitness of the mathematical models (decreased adjusted R2) for all the beaches except the Kreher Park Beach. Explanatory variables that were included in the mathematical models were unique for each beach. Therefore, it was concluded that the mathematical models were beach specific and that combination of years of data did not necessarily result in more robust mathematical models, due to annual changes in swim season physical parameters.