Joint inversion of multiple geophysical data and its application to geology differentiation

by Jiajia Sun

Institution: Colorado State University
Year: 2015
Keywords: Inversion (Geophysics); Algorithms; Cluster analysis; Induced polarization; A priori
Record ID: 2061257
Full text PDF: http://digitool.library.colostate.edu:80/R/?func=dbin-jump-full&object_id=459606


Geophysical inversion has been an important and effective tool in both mineral and petroleum industries. The recovered images from geophysical inversions are interpreted by geologists to understand the subsurface geological structures and to guide further exploration activities such as spotting drill holes. Geophysical inversion, however, does not always provide a good subsurface image that reliably reflects the structural and petrophysical properties of the target due to many reasons. My research focuses on improving the fidelity of geophysical inverted models and consequently, the geology differentiation results. To achieve that, I developed three new joint inversion algorithms. I developed a clustering inversion algorithm that jointly inverts geophysical data and statistical petrophysical data through the use of fuzzy c-means (FCM) clustering technique in a deterministic inversion framework. The clustering inversion allows both geophysical and petrophysical data to directly contribute to the construction of a physical property model through the minimization of a single objective function. The geology differentiation is accomplished simultaneously with the geophysical inversion during the minimization process. No post-inversion analysis is needed any more. I also extended the clustering inversion algorithm to joint inversion of seismic and gravity data constrained by a priori petrophysical information. The jointly inverted velocity and density models can predict not only the observed seismic traveltime and gravity data but also the a priori petrophysical data. By using some variants of the FCM clustering, I further generalized the joint clustering inversion algorithm to more general petrophysical scenarios that one might encounter in reality. I also developed a new 4D inversion algorithm that jointly inverts the induced polarization data measured at different time channels. I introduced time regularization to encourage smoothness in the chargeabilities in time dimension. I also designed a norm fitting strategy to make the chargeability models behave consistently with the observed IP data. Geology differentiation is accomplished by applying the FCM clustering directly to the recovered chargeability decay curves. My research provides the necessary tools and algorithms for jointly inverting geophysical and petrophysical data, for integrating geology interpretation into geophysical inversion, and ultimately, for enhanced geophysical images and geology differentiation results.