Smooth and Sparse Inversion of EMI Data from Multi-Configuration Measurements

In this study, we deal with the inversion of frequency-domain electromagnetic data collected with devices with different configurations (varying inter-coil spacing, frequency, height from the ground). More specifically, we present the results of the application of a Gauss-Newton inversion algorithm based on a non-linear forward model onto several synthetic resistivity and magnetic permeability vertical profiles. In addition, we shortly discuss the inclusion into this inversion scheme of a quite novel stabilizing term, based on minimum (gradient) support, and promoting sparse reconstructions. We demonstrate the effectiveness of this sparse inversion algorithm on synthetic and real datasets.