Regression and Inverse Analyses in Regional Ground-Water Modeling

Abstract

Numerical models of ground-water flow within the regional aquifer underlying Lambton County, Ontario, Canada, are constructed by the conjunctive application of methods of regression and inverse analyses. Regression analysis of physiographic and hydraulic head data reveals a distinct relation between ground-water levels and ground-surface topography that is used to condition the aquifer models that are subjected to inverse analysis. Inverse analysis determines the variation of hydraulic head along the perimeter of the region and the distribution of ground-water recharge and discharge within the region that optimally replicate the observed hydraulic head data. The use of physiographic data as a substitute for geologic data in the construction of the aquifer models is defended on the basis of the constraints that apply to the investigation and the opportunity to invoke hydrogeologic judgment in the evaluation of the results. Interpretation of the results of the analyses reveals important characteristics of the hydrogeology of Lambton County, including an area of elevated ground-water recharge and the partitioning of ground-water discharge to the Saint Clair River.