Part one. A magnetotelluric study of the High Cascades Graben in Central Oregon. Part two. Program 3Dfeem--a multidimensional electromagnetic finite element model

This dissertation consists of two parts, the application of the magnetotelluric method to a study of the High Cascades in Central Oregon and the development of a multidimensional finite element program to model the vector electromagnetic response of the earth.;Part one gives the results of a study of the High Cascades Graben involving the acquisition and interpretation of 40 stations of wide band magnetotelluric (MT) data along three east-west lines in the vicinity of Mt. Jefferson and the Three Sisters volcanoes. Data were modeled using one-dimensional inversions of rotationally invariant parameters followed by conventional two-dimensional modeling using existing programs. The response of final two-dimensional models matched observations to within the error associated with observations, except where strong three-dimensional effects were evident in the data set. Conductive and resistive layers in the modeled upper crust were interpreted in terms of the known and inferred geologic structure of the range. Lateral contrasts in conductivity thought to be associated with graben-bounding structure were observed in "pseudosections" of the raw data and in the final models. A thin resistive layer associated with recent, relatively unaltered graben-filling basaltic andesites was also noted. A pervasive conductor detected in the lower crust deepens to the east, which may indicate concomitant eastward crustal thickening.;Part two describes program 3Dfeem (three dimensional finite element electromagnetic model), developed to model the response of a given assemblage of conductors to a plane-wave or electric dipole source. This application of the finite element method is based on the principle of minimization of energy; solution vector fields are sought that minimize an expression giving the energy in the fields and their sources at discrete points within the model volume. Solution fields are considered to be a superposition of "primary" fields (the response of a one-dimensional earth in the presence of specified source(s)) and "anomalous" fields resulting from heterogeneities within the one-dimensional model. Results from two and three-dimensional conductivity configurations are seen to be in favorable agreement with those from other computer models. Additionally, a comparison of solution fields from 3Dfeem and a two-dimensional analytic solution indicates that, with appropriate mesh spacing, 3Dfeem may give field estimates that are accurate to within about 8 percent. Adaptation of algorithms developed in EMCAL to decrease mesh spacing in the vicinity of heterogeneities ensures that the error criteria are met. Key portions of the annotated code are attached as an appendix.