Electronic Thesis and Dissertation Repository


Master of Engineering Science


Civil and Environmental Engineering


Professor Slobodan P. Simonovic


Regional frequency analysis (RFA) is used to obtain reliable estimates of local precipitation events for a variety of applications in water resources engineering. The focus of the presented research is on an initial step of the RFA process; that is the formation of precipitation regions (also referred to as regionalization). The aim of this study is to dissect the regionalization procedure into its individual components that require subjective user input, and to evaluate their respective influences on the results. All assessments are conducted in two of Canada's climate regions; namely the Prairie and Great Lakes-St. Lawrence lowlands. Additionally, a new fuzzy clustering approach to regionalization that uses optimization is proposed. It is evident that the outcomes are sensitive to the choice of the regionalization method, the number of regions into which the sites of the study area are partitioned, the climate site attributes and the temporal resolution of the precipitation data. Recommendations for the selection of such factors are provided based on their application.