An adaptive mesh method for electromagnetic inverse problems

Abstract

Electromagnetic data can be used to recover the electrical conductivity of the subsurface of the earth. In order to extract quantitative value from an exploration survey, the data must be inverted. When working with traditional 3D regular rectangular meshes, the number of cells required to accurately discretize the earth can be prohibitively large. This is especially true for large irregularly shaped surveys or regions of severe topography. Using an adaptive mesh structure, such as an Oc-Tree, high model resolution can be given where it is needed, while cells can be made coarse where possible. We develop an electromagnetic inverse algorithm on a semi-structured Oc-Tree mesh. The algorithm is applied to a synthetic example of controlled source frequency domain and to a field ZTEM data example.