Integral equation modeling of three‐dimensional magnetotelluric response

Abstract

We have adapted a three‐dimensional (3-D) volume integral equation algorithm to magnetotelluric (MT) modeling. Incorporating an integro‐difference scheme increases accuracy somewhat. Utilizing the two symmetry planes of a buried prismatic body and a normally incident plane wave source greatly reduces required computation time and storage. Convergence checks and comparisons with one‐dimensional (1-D) and two‐dimensional (2-D) models indicate that our results are valid. We show theoretical surface anomalies due to a 3-D prismatic conductive body buried in a half‐space earth. Instead of studying the electric and magnetic fields, we have obtained impedance tensor and magnetic transfer functions by imposing two different source polarizations. Manipulation of the impedance tensor and magnetic transfer functions yields the following MT quantities: apparent resistivity and phase, impedance polar diagrams, tipper direction and magnitude, principal directions, skew, and ellipticity. With our preliminary analyses of these MT quantities, we have found that three‐dimensionality is usually revealed by all of them. Furthermore, we have recognized two pairs of complementary parameters: apparent resistivity and phase, and skew and ellipticity. Because of surface charges at conductivity boundaries, low‐frequency 3-D responses are much different from 1-D and 2-D responses. Thus, in many cases 3-D models are required for interpreting MT data. Although an overall 3-D MT interpretation is still not practical due to high computer costs and the complicated structure of the true earth, combined 2-D and 3-D modeling can be applied to yield a gross 3-D structure, which is composed of a cross‐section and its strike extent. In doing so, we suggest that the cross‐section be obtained from higher frequency 2-D E‐perpendicular [Formula: see text] mode modeling, and that the strike extent be derived by matching with lower frequency E‐parallel [Formula: see text] mode results due to corresponding 3-D models. In addition, we have indicated that some simple 3-D features, e.g., location above conductive zone, corners, and symmetry lines, can easily be recognized from the surface MT response.