Joint inversion of Multi-frequency Electromagnetic Induction and Seismic Refraction Data For Improved Near Surface Characterization

Abstract:

We present a joint inversion routine between multi-frequency Electromagnetic (EM) induction and seismic refraction data that is based on using both cross-gradients and disconnect constraints. The joint inverse problem was solved using an iterative nonlinear least-squares formulation. Following each iteration, the cross gradient constraint enforces structural similarities between the EM and seismic models, whereas the disconnect constraint enforces sharp boundaries between different strata within the EM model. The locations of boundaries within the EM model are assumed to be consistent with the locations of user-defined velocity contours in the seismic model. We tested our method on a challenging synthetic EM and seismic model scenario that contains water-bearing zones as well as positively and negatively correlated model parameter values. In addition, we applied our method to GEM-2 and seismic refraction field data sets acquired along a 28-m-long profile in Laramie (WY), and we precisely recorded the locations where ground surface resistivity and velocity likely changes along that line. Unlike the results of separate EM and seismic inversions and the results of joint inversion based only on a cross-gradient constraint, our method was able to detect the water-bearing zones. In addition, it better captured ground surface changes in the field data set.