Structural information in the inverse problem

Marios Karaoulis, Timothy H. Larson, Ismail Ahmed, A. Revil, Jason F. Thomason

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Data integration in geophysics provides additional information to elucidate subsurface structure and reduce non-uniqueness of inverted models. There are several strategies for incorporating data integration into numerical models. A traditional approach is to use this information as a priori knowledge, as an initial model or layer boundary specified on the mesh before the inversion. Although in some cases this approach has proven effective, the information is not directly incorporated into the inverse problem and might be lost in the final model. Another strategy is through joint inversion, where data and models are inverted simultaneous. The data integration comes from the joint operator as structural similarity or petrophysical equations. There are some limitations with this approach. In particular, structural similarity doesn't take into account different sensitivity patterns which differ for different geophysical methods, e.g. in a crosswell configuration electrical resistivity tomography is sensitive close to the borehole region while seismic waves are sensitive towards the center part. Moreover different methods have differing resolution. Therefore, a single joint operator might not be effective in all cases. In this work we demonstrate the use of image-guided inversion, where structural information is taken directly from a high resolution geophysical image (e.g. ground penetrating radar or seismic reflection) or from a geological cross-section. This structural information is introduced into the inverse problem through a weighted smoothing matrix, where it correlates and favors formations related to a specific structural feature and not just uniformly across the entire model. Both sharp and smooth features can be imaged and the recovered models can have a more realistic distribution of values. As an example of the method we use migrated seismic reflection images to extract the structural information and resistivity imaging to recover the resistivity distributions. We check the validity of this approach using synthetic models and real seismic and resistivity data collected along the same profile lines in Illinois.
Original languageEnglish (US)
Title of host publicationAmerican Geophysical Union Fall Meeting
Place of PublicationWashington, DC
PublisherAmerican Geophysical Union
Volume2013
StatePublished - 2013

Keywords

  • ISGS

Fingerprint

Dive into the research topics of 'Structural information in the inverse problem'. Together they form a unique fingerprint.

Cite this