Modeling and inversion of electromagnetic data collected over steel casings: An analysis of two controlled field experiments in Colorado

Abstract

Modeling and inversion of electromagnetic (EM) data contaminated by steel infrastructure remains a numerically challenging task. In this study, we collected controlled-source EM (CSEM) field data at two test sites. First, we conducted a CSEM survey over a dry and abandoned well in eastern Colorado. Next, we collected CSEM data over a test well located at the Colorado School of Mines. The purpose of these experiments was to examine methods to invert EM data contaminated by infrastructure effects and recover undistorted subsurface conductivity models. We used a hybrid approach to model casing effects by considering the conductive metal as a distribution of secondary sources with magnitudes calculated using an approximate layered background. In particular, we applied the method of moments algorithm to calculate the magnitude of the secondary electric dipole sources along the casing. Subsequently, we included these secondary sources in a 3D finite-volume-based forward calculation and used the Gauss-Newton method to invert the contaminated field data. As expected, our preliminary inversion results show that by not considering casing effects, significant artifacts are introduced in the recovered models. Furthermore, we show that such artifacts are reduced significantly through the introduction of casing physics in the forward model, enabling the surrounding subsurface conductivity and corresponding geology to be characterized.