Algorithm for stable solution of inverse problem of magnetotelluric sounding

Abstract

When solving inverse problems of magnetotelluric sounding (MTS), equivalent solutions appear, noticeably different from each other. But the solution of a direct problem under a given medium model and boundary conditions is the only one, the response of the medium to the source of the electromagnetic field is unique. An algorithm is considered that leads to an accurate solution of the test problem when striving for zero inconsistencies in input and model data. Several starting medium models and two optimization methods are used: a nonlinear least squares method with calculations of the sensitivity matrix and a method based on metaheuristic algorithms used when target functions have several local minima. Using numerical calculations, a stable solution of the inverse MTS problem for the 3D-medium model was obtained.