Loading...
Understanding geophysical information in magnetotelluric and magnetic data: an investigation through multiparameter and multiphysics inversion
Darrh, Andréa
Darrh, Andréa
Citations
Altmetric:
Advisor
Editor
Date
Date Issued
2023
Date Submitted
Collections
Research Projects
Organizational Units
Journal Issue
Embargo Expires
2024-10-18
Abstract
The emphasis of geophysical inversions has changed significantly over recent time. Instead of inverting for one scalar physical property, we often are required to include more complex representations of our model and physics in the forward model to capture more realistic representations of the subsurface. In the same vein, multiphysical joint inversions are now implemented where individual inversions were primarily used, again with the intention of recovering a more realistic subsurface by leveraging complementary sensitivities within the multiphysics approach. In this thesis, I first examine the effect that dipping electrical anisotropy has on 1D MT data and inversions, quantifying the impact that electrical anisotropy has on the impedance through the eccentricity of polar plots. I also demonstrate that the use of an L1 norm improves the recovery of the true depth extent and horizontal electrical conductivities of an anisotropic layer. Inversions using an isotropic L2 norm and an anisotropic L2 and L1 norm display the effectiveness of each inversion formulation at recovering an anisotropic layer, with the L1 inversion resulting in the best recovery of the anisotropic layer. I then expand the inversion to a 3D, isotropic, joint inversion of MT and magnetic data to investigate the ability of a joint inversion coupled using physical property information to improve recovered electrical conductivity and susceptibility models. The simultaneous joint inversion formulation showed that, by coupling physical property information through a fuzzy c-means (FCM) clustering term, the recovery of physical property values was improved and, as a consequence, the subsurface structure is better defined. Lastly, I examine the assumptions behind different criteria used to select final models from a joint inversion and introduce a new criterion called the dell-manifold criterion. This new criterion finds the maximum point of curvature on a 2D Tikhonov surface to find the optimal recovered models for a joint inversion without needing a prior accurate noise estimate.
Associated Publications
Rights
Copyright of the original work is retained by the author.