Full Waveform Inversion of Transient Electromagnetic Data in the Time Domain

Quasi-seismic imaging is a popular form of electromagnetic (EM) imaging technology. In most cases, it is considered that it will yield more obvious layered characteristics if the EM response is transformed into the virtual wave domain. However, the time profile of the virtual wave cannot effectively show the underground EM structure before migration. In this article, we make an improvement to process the virtual wave, by applying the full waveform inversion (FWI) method, with the steps detailed as follows: 1) we use the truncated singular value decomposition (TSVD) method to transform the transient EM response to the virtual wave field; 2) using the FDTD method, we calculate the virtual wave field according to the resistivity model, then use the shearlet transform (ShT) to remove the direct wave; and 3) we use a least square equation between the virtual wave and transformed virtual wave to express the problem of FWI. Next, we apply one of the steepest descent methods-the Marquardt approach-to obtain the optimal underground conductivity information. Finally, we use the synthetic and field data to show the accuracy and rationality of the method proposed in this article.