TY - JOUR

T1 - Three-dimensional joint inversion for magnetotelluric resistivity and static shift distributions in complex media.

AU - Sasaki, Yutaka

AU - Meju, Maxwell

N1 - Copyright 2006 by the American Geophysical Union. The magnetotelluric method is used for studying fundamental flow processes in the crust and mantle but suffers from problems caused by geological heterogeneity. We develop methods for accurately accounting for multi-scale heterogeneity, leading to accurate 3D resistivity imaging from surface to mantle depth. Methodological developments (2nd author), implementation (2nd author). RAE_import_type : Journal article RAE_uoa_type : Earth Systems and Environmental Sciences

PY - 2006/5

Y1 - 2006/5

N2 - Accurate interpretation of magnetotelluric (MT) data in the presence of static shift arising from near-surface inhomogeneities is an unresolved problem in three-dimensional (3-D) inversion. While it is well known in 1-D and 2-D studies that static shift can lead to erroneous interpretation, how static shift can influence the result of 3-D inversion is not fully understood and is relevant to improved subsurface analysis. Using the synthetic data generated from 3-D models with randomly distributed heterogeneous overburden and elongate homogeneous overburden that are consistent with geological observations, this paper examines the effects of near-surface inhomogeneity on the accuracy of 3-D inversion models. It is found that small-scale and shallow depth structures are severely distorted while the large-scale structure is marginally distorted in 3-D inversion not accounting for static shift; thus the erroneous near-surface structure does degrade the reconstruction of smaller-scale structure at any depth. However, 3-D joint inversion for resistivity and static shift significantly reduces the artifacts caused by static shifts and improves the overall resolution, irrespective of whether a zero-sum or Gaussian distribution of static shifts is assumed. The 3-D joint inversion approach works equally well for situations where the shallow bodies are of small size or long enough to allow some induction such that the effects of near-surface inhomogeneity are manifested as a frequency-dependent shift rather than a constant shift.

AB - Accurate interpretation of magnetotelluric (MT) data in the presence of static shift arising from near-surface inhomogeneities is an unresolved problem in three-dimensional (3-D) inversion. While it is well known in 1-D and 2-D studies that static shift can lead to erroneous interpretation, how static shift can influence the result of 3-D inversion is not fully understood and is relevant to improved subsurface analysis. Using the synthetic data generated from 3-D models with randomly distributed heterogeneous overburden and elongate homogeneous overburden that are consistent with geological observations, this paper examines the effects of near-surface inhomogeneity on the accuracy of 3-D inversion models. It is found that small-scale and shallow depth structures are severely distorted while the large-scale structure is marginally distorted in 3-D inversion not accounting for static shift; thus the erroneous near-surface structure does degrade the reconstruction of smaller-scale structure at any depth. However, 3-D joint inversion for resistivity and static shift significantly reduces the artifacts caused by static shifts and improves the overall resolution, irrespective of whether a zero-sum or Gaussian distribution of static shifts is assumed. The 3-D joint inversion approach works equally well for situations where the shallow bodies are of small size or long enough to allow some induction such that the effects of near-surface inhomogeneity are manifested as a frequency-dependent shift rather than a constant shift.

KW - magnetotellurics

KW - static shift

KW - 3-D inversion.

U2 - 10.1029/2005JB004009

DO - 10.1029/2005JB004009

M3 - Journal article

VL - 111

JO - Journal of Geophysical Research: Solid Earth

JF - Journal of Geophysical Research: Solid Earth

IS - B5

M1 - B05101

ER -