Inversion Strategy - Models and Methods of Magnetotellurics - page 523

model. Long-period MV and MT curves ( T 10000 s) are interpreted with the

programs IGF-MT2D (Novozhynski and Pushkarev, 2001) and II2DC (Varentsov,
2002), implementing the misfit minimization in the class of blocky media with
a fixed geometry of blocks. The interpretation is conducted in the hypotheses
test mode. We examine three hypotheses of the Cascadia subduction zone which
involve (1) dehydration in the continental crust (EMSLAB-I model, Fig. 12.52),
(2) dehydration in the continental crust and development of a continental par-
tially melted asthenosphere (EMSLAB-II model, Fig. 12.53), and (3) dehydration
in the continental crust and development of a continental asthenosphere with a
subvertical magmatic zone of ascending melting (predictive CASCADIA model,
Fig. 12.44).
The two-dimensional interpretation model (START model) is shown in
Fig. 12.61 with starting values of resistivities. The oceanic water resistivity is
taken as 0.3 Ohm

= 1 −

m. The seafloor topography and thicknesses of bottom sedi-
ments, as well as sediments of the accretionary prism and shelf, are specified
from the bathymetric and sedimentary thickness maps. The depth to the oceanic
asthensphere is defined as 37 km in conformity with the models CASCADIA,
EMSLAB-I, and EMSLAB-II. The surface of the subducting oceanic plate is recon-
structed from seismic data (Trehu et al . , 1994) and seismic tomography imagery
(Weaver and Michaelson, 1985; Rasmussen and Humphries, 1988). The structure
of the continental volcanic-sedimentary cover is determined from the 1D inver-
sion of short-period MT curves. The downgoing plate as well as continental crust
and mantle are divided into uniform, resistive blocks (10 3 -10 4

·

m) whose
distribution allows a free choice of crustal and mantle structures consistent with
the three hypotheses under the question. The hypothesis that best fits observa-
tions is selected automatically in the process of resistivity optimization and misfit
minimization.
Let us use an interpretation scheme in which the magnetovariational inversion
plays a leading role, whereas the magnetotelluric inversion serves to check and
edit the MV results. The main advantage of this scheme is that, with lowering
frequency, the magnetovariational tipper becomes free from the distorting effects of
near-surface heterogeneities. It is clear that, in this way, we substantially improve the
reliability of geoelectric information burdened with galvanic distortions of apparent
resistivities.
Magnetovariational and magnetotelluric data obtained at 15 stations on the Lin-
coln line ( T

Ohm ·

1-10000 s) are successively interpreted on three levels (Fig. 12.62).
Here, we follow the same algorithm of partial inversions as in Section 12.6
describing the experiments on integrated interpretation of synthetic MV and
MT data.
Level I. Inversion of Re W zy and Im W zy (TE-mode). The START model was
taken as the starting model. The MV inversion yields the TP model shown in
Fig. 12.63. The model misfits are given in Table 12.2, where

=

Im W zy
are the misfits of the real and imaginary tippers (rms deviations of the model val-
ues from the observed values), while

Re W zy and

max Re W zy −

min Re W zy and

Re W zy =

max Im W zy −

min Im W zy characterize the maximum variation in

Im W zy =