3D magnetic inversion in highly magnetic environments using an octree mesh discretization

Abstract

Standard techniques for inverting magnetic field data are marginalized when the susceptibility is high and when the magnetized bodies have considerable structure. A common example is a Banded Iron Formation where the causative body is highly elongated, folded, and has susceptibility greater than unity. In such cases the effects of self-demagnetization must be included in the inversion, which can be accomplished by working with the full Maxwell?s equations for magnetostatic fields. This problem has previously been addressed in the literature but there are still challenges with respect to obtaining a numerically robust and efficient inversion algorithm. In our paper we use a finite volume discretization of the equations and an adaptive octree mesh. The octree mesh greatly reduces the number of active cells compared to a regular mesh, which leads to a decrease of the storage requirement as well as a substantial speed up of the inversion. Synthetic and field examples are presented to illustrate the effectiveness of our method.