Date of Award

8-31-2020

Publication Type

Doctoral Thesis

Degree Name

Ph.D.

Department

Civil and Environmental Engineering

Keywords

AEM3D, E. coli, Ecological Modelling, Hydrodynamic Modelling, Predictive Modelling, TUFLOW-FV

Supervisor

Rajesh Seth

Supervisor

Luis F. Leon

Rights

info:eu-repo/semantics/embargoedAccess

Abstract

Lake St. Clair is a freshwater lake in the Lake Huron to Erie corridor in the Great Lakes Basin. Millions of people in Canada and the United States rely on that water source for drinking, fishing and recreational purposes. Lake St. Clair’s watershed is heavily impacted by human activity, which can result in contamination of its waters by fecal matter of human or animal origin containing waterborne pathogens, and thus pose a direct threat to human health. Common sources of such pollution include combined sewer overflows, wastewater treatment plant bypasses, and agricultural application of manure derived from animal fecal waste. Several such sources are present in Windsor Essex County (WEC), Ontario, Canada, which is located along the southern edge of Lake St. Clair. Two popular public beaches and drinking water intakes are located in the nearshore region adjacent to the southern edge. Fecal microbial pollution is currently monitoring using fecal indicator bacteria (FIB), such as Escherichia coli (E. coli). Monitoring methods have several limitations including their inability to predict water quality in real-time or in advance, or to identify potential sources of contamination for more effective management. Mathematical models are tools that can be very effective and complementary to monitoring in overcoming its limitations. Model predictions can be real-time or near real-time and also help to identify or exonerate potential sources of microbial pollution. In the current study, two types of modelling approaches that are commonly being used in the assessment of microbial contamination in beach waters and lakes were investigated: statistical modelling based on multiple linear regression (MLR) and hydrodynamic-ecological modelling. The statistical MLR models developed for Sandpoint Beach in Lake St. Clair showed higher accuracy in the range 64-78%, for predicting both exceedance and non-exceedance of the applicable standard, as compared to 54% accuracy obtained using the current method based on E. coli measurements. Amongst the MLR models developed, an increase of about 5-14% in model performance was observed when qualitative sky weather condition was included. Results with mechanistic structured grid high-resolution AEM3D model developed for Lake St. Clair showed that four major tributaries (Thames, Sydenham, St. Clair and Clinton River) are unlikely to be responsible for the E. coli exceedances of provincial guideline observed at Sandpoint Beach. Amongst the major tributaries, predicted E. coli concentrations were dominated by the contribution of St. Clair River for most of Lake St. Clair. The maximum predicted E. coli concentration from the combined input of the major tributaries was less than 100 CFU/100 ml for most of the lake and less than 10 CFU/100 ml at Sandpoint Beach. Predicted E. coli were significantly affected by varying water temperature and sunlight result in the temporal and diurnal dynamics of microbial water quality in Lake St. Clair. About 12–148% differences in predicted E. coli concentrations were observed at six drinking water intakes located in Lake St. Clair when time-variable decay rates were used instead of a constant decay rate. Also, on average nighttime E. coli predictions were 21–68% higher at these water intakes, as compared to daytime levels. Results from the AEM3D model showed that while the flow contribution of eight smaller tributaries in Windsor Essex Region to the lake is insignificant (less than 0.2%), their contribution to the adjacent nearshore region along the southern edge of Lake St. Clair could be quite significant. Within about 1 km from the shoreline of this nearshore region, flow contributions from the small tributaries were estimated in the range between 18-35%, while their contribution to E. coli concentration was estimated to be more than 80%. Results with mechanistic unstructured grid TUFLOW-FV/AED2+ lakewide model and with a finer mesh nested model over a 2 km region surrounding Belle River showed differences of up to a factor of four in predicted E. coli concentrations at adjacent Lakeview Park West Beach (LP Beach). The differences reduced to a factor of up to 1.3 at nearby Lakeshore WTP intake located about one km away from shore. While the average contribution of the Belle River to E. coli concentrations at Lakeshore WTP intake was predicted to be <20%, the contribution increased to >80% when higher concentrations (10-35 CFU/ 100 ml) were predicted. The results also indicate that the construction of the marina may have contributed to some increase in E. coli concentrations at LP Beach from the external sources considered. However, construction of a new 150 m jetty in 2018, in place of the 25 m jetty separating Belle River from LP Beach, is expected to reduce the E. coli concentrations at LP Beach from the same sources by about 80%.

Share

COinS