Nonlinear least-squares and universal-kriging estimation of source and ground-water parameters for several types of plumes caused by instantaneous contamination releases

In recent years, a large amount of research has been conducted examining ground-water contamination and the physical properties of aquifers that affect it. Very little of this research has looked at problems specific to contamination plumes that arise from one or more instantaneous releases of a contaminant into an aquifer. In this dissertation, two statistical methods for estimating aquifer and source parameters associated with pulse-type contamination plumes are examined. The first method is based on nonlinear least squares, and is applied to plumes arising from a single contaminant release that are observed once, plumes arising from a single contaminant release that are observed more than once, and plumes that arise from several releases that are observed once. The second method is based on universal kriging and is applied only to plumes that arise from single releases that are observed once. For each plume model and method, a statistical model is developed, a computer program to perform the parameter estimation is discussed, and the program and model are evaluated using data derived from the statistical models and using contaminant transport models. Finally, all models and methods are tested using a set of data derived from the study of a real aquifer located at Borden Air Base in Ontario, Canada.