The Effects of Inhomogeneities on Magnetotellurics

Abstract

The magnetotelluric (MT) method is a low-frequency, electromagnetic (EM), geophysical method which is widely used for investigating the conductivity structure of the Earth's crust and upper mantle. Although MT often seeks to determine the gross conductivity structure and the variation of conductivity with depth, it suffers from the ubiquitous presence of EM scatterers of all scales. The thesis advances the understanding of the effects of multiple and small-scale inhomogeneities on the MT impedance tensor. One of the most pressing difficulties at present in MT are distortions of the EM fields by small-scale, 3-D inhomogeneities which are often found near the surface of the Earth. The thesis investigates the effects of these inhomogeneities on the observed impedance tensor. These studies show that for a great many cases the impedance tensor conforms to a particular factorization, each part of which is related either to the local inhomogeneities or to the large scale structure. Each factor is associated with a different physical effect. This factorization is used to explain the necessary requirements for a useful parametrization of the data contained within the impedance tensor. The development of such a complete, useful decomposition is begun with a method for dealing with the most significant effect of small-scale surface bodies; namely galvanic or frequency-independent distortion of the horizontal electric fields. In addition, analytic and statistical solutions are provided to a number of relevant low-frequency EM scattering problems which model the response of inhomogeneities. An analytic solution for two semi-infinite slabs over a basement is developed in the thesis. This extends existing solutions to this problem so as to include all frequencies and arbitrary basement conductivities. For the particular case of an insulating or perfectly-conducting basement, the above method is extended to provide the EM response for a model of a quasi-anisotropic layer. This model was used to study the benefits of extensive spatial sampling of the electric field in complex media and the presence of fictitious conducting layers when data due to large lateral inhomogeneities are interpreted one-dimensionally. The study showed, for this particular model, that the use of large electrode spacing could be beneficial if used with care. The anisotropy model was also employed to investigate the question of what bulk properties the MT method samples. A statistical technique is used for investigating the response of very fine-bedding in MT demonstrating that such structures can have a relatively significant effect of up to ten degrees on the impedance phase. These results are corroborated by a deterministic model.