Three-dimensional magnetotelluric inversion using a heterogeneous smoothness-constrained least-squares method

Negi, Tateyuki*; Mizunaga, Hideki*; Asamori, Koichi; Umeda, Koji

This paper presents a fast algorithm for electromagnetic inversion. The algorithm is distinct for the level of accuracy it attains when bypassing the sensitivity matrix update. A common sensitivity matrix for homogeneous half-space is used in all of the iterations. Instead of a fixed sensitivity matrix, the Laplace filter coefficients at each element are updated based on the spatial differences of resistivity calculated by a reconstructed structure derived from the former iteration. This substitution is expected not only to reduce the computation time required for large-scale inversions, such as those for three-dimensional surveys, but also to allow reconstruction for sharp boundaries in resistivity structures. This algorithm was applied to three-dimensional magnetotelluric inversion in order to confirm the effectiveness. This paper reports the results of several synthetic examples using different resistivity models and conditions.