Abstract

Large proportion of the world’s readily available water supply is at risk due to the rapidly increasing populations of certain types of harmful algae. During the photosynthesis, species like blue-green algae and cyanobacteria consume nutrients and produce toxins that have potential adverse effects to humans and animals.
This thesis focuses on developing a GIS-based statistical approach to explore the water quality parameters facilitating the algae bloom, and to geographically map the extent and spread of these parameters to enable tracking and prediction of potential algae outbreaks.
The relationship between Chlorophyll-a, which represents the concentration of algae biomass, and the water quality parameters such as depth, phosphorus, nitrogen, alkalinity, suspended solids, pH, temperature, electrical conductivity, dissolved oxygen and secchi depth is analyzed though correlation matrix then by utilizing modeling techniques including multiple linear, nonlinear regression, neural network and data mining prediction models are developed to quantify the contribution from essential water quality parameters to eutrophication.
The developed GIS and statistical analysis approaches have been applied to the Lake Champlain. The performance for the developed statistical, neural network and data mining chlorophyll-a models has been examined through the comparison with the observed field data and through statistical error analysis. Two new techniques have been examined in this thesis study. First, data mining has helped to reveal the nonlinear behavior of algae growth in some parts of the case study area. Second, the GIS spatial analysis is employed to visualize the spread and extent of the water quality parameters and the algae chlorophyll-a, which graphically present the location-based impact of eutrophication on important lake water resources. For example, the analysis of the GIS-based impact maps suggests that the algae is affecting the Vermont section of Lake Champlain mainly the Northern and Southern section. The developed models suggest that algae production is affected by nutrients particularly phosphorus. When phosphorus is encountered at low to mild concentrations, the nutrient is linearly affecting algae production, however, at extreme concentrations of the nutrient the relationship between nutrient and algae production become nonlinear. The developed GIS model along with the statistical analysis applied on lake Champlain suggest that Extreme levels of Nitrogen in north and Chloride in the South caused deviations in the models prediction accuracy