Fast 3-D Modeling And Inversion Of Surface Em Data

This work presents an algorithm for three-dimensional modeling and inversion of electromagnetic field data. In recent<br>years, there has been an increase in use of electromagnetic techniques for mineral, oil and geothermal exploration.<br>Since most of the targets are of three-dimensional nature, it is important to consider the full dimensionality of the problem<br>in the interpretation. The forward code is based on a finite difference approximation while the sensitivities needed in the<br>inversion are obtained from the forward problem based on the reciprocity principle. A parametric functional is constructed<br>using the data misfit and a model misfit to an a priori earth model, weighted by a regularization parameter. This functional<br>is then minimized using the conjugate gradient method. The most time consuming parts of an inversion algorithm are the<br>forward problem and calculation of the sensitivities. Also, in the 3-D problem, storage becomes another important issue.<br>First results will be presented, along with methods that will help improve both speed and storage on the way to make the<br>3-D problem tractable.