One‐dimensional inversion of natural source magnetotelluric observations

Abstract

Natural source magnetotelluric (MT) data are inverted to find a subsurface resistivity which is isotropic and a continuous function of depth. The inversion consists of two parts. In the first, our concern is to construct a resistivity structure whose responses are acceptably close to the observations, that is, the measured amplitudes and/or phases. This construction proceeds iteratively and is realized through an automated algorithm which is characterized by numerical stability and rapid convergence. In the second part, we attempt to investigate the degree of nonuniqueness inherent in the problem. This is accomplished by considering only spatial averages of the resistivity because all models linearly close to the constructed model will have the same average resistivity. The predictions of this statement are examined quantitatively in an example where the averages from five different models which reproduce the observations are compared. In an example using data from a moderately complex layered model, we have inverted jointly and separately the amplitudes and phases and compared their resolution. Finally, data from the Snake River Plain‐Yellowstone area are inverted and the results compared with those from a parametric inversion.