Open AccessInversion for regional 2‐D resistivity structure in the presence of galvanic scatterers
Author(s) -
DeGrootHedlin C.
Publication year - 1995
Publication title -
geophysical journal international
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.302
H-Index - 168
eISSN - 1365-246X
pISSN - 0956-540X
DOI - 10.1111/j.1365-246x.1995.tb06843.x
Subject(s) - magnetotellurics , occam , inversion (geology) , matrix decomposition , geology , electrical impedance , electrical resistivity and conductivity , inverse problem , geophysics , algorithm , mathematical analysis , mathematics , tectonics , physics , computer science , seismology , eigenvalues and eigenvectors , quantum mechanics , programming language
SUMMARY Magnetotelluric (MT) field data are often distorted by small‐scale, near‐surface conductivity anomalies. In this paper, the Occam regularized inversion technique is expanded to solve simultaneously for the galvanic distortion matrix, regional strike, and resistivity parameters that correspond to the smoothest regional 2‐D resistivity model fitting the data. This involves fitting all four complex components of the MT impedance tensor. By solving for smooth structure along with the strike and decomposition parameters, rather than the usual two‐step decomposition/inversion process, we avoid inferring conductivity features that result only from an erroneous choice of strike and decomposition parameters. Furthermore, we ensure that the recovered regional impedance tensor is physically realistic. A method is presented for determining subsets of the data corresponding to the assumed physical model, that of 2‐D structure at inductive scale‐lengths, overlain by galvanic scatterers. This method is based on the linear dependence of the elements of the distorted MT tensor. Tests of the extended Occam algorithm are presented for both synthetic data and a subset of the BC87 data. Results show that, for appropriate subsets of the data, the determination of strike and decomposition parameters can be fully automated with the extended Occam algorithm.