NEW APPROACHES IN THE INTERPRETATION OF DEEP EM SOUNDING DATA ALONG THE

1

• NEW APPROACHES IN THE INTERPRETATION OF
  DEEP EM SOUNDING DATA ALONG THE NARYN
  TRANSECT IN KYRGYZ TIAN-SHAN
• NARYN Work Group
• Moscow State University (Golubtsova N.S.,
  Pushkarev P.Yu.)
• GEMRC, Inst. of Physics of the Earth RAS,
  Troitsk, Russia (Sokolova E.Yu., Baglaenko N.V.,
  Martanus E.R., Varentsov I.M.)
• Scientific Station, United Inst. of High
  Temperatures RAS, Bishkek, Kyrgizia (Rybin A.K.,
  Batalev V.Yu., Safronov I.V., Schelochkov G.G.)
• presented by Elena Sokolova

2
NARYN MT/GDS transect
The NARYN transect of EM soundings, crossing for
700 km the Kyrgyz and China Tian Shan, became a
promising object for investigation of the deep
conductivity structure of this active region. The
data collection acquired by Kyrgyz and American
teams includes 19 long period MT sites (large
circles) simultaneously observed in 3 groups with
perfect quality by LIMS equipment (S.K. Parks,
1999-2000) and tens of local AMTs (small dark
circles, CES-2 soundings, Scientific Station of
UIHT RAS, Bishkek).
?
A valuable experience of EM data interpretation
on the transect has been obtained in (Rybin et
al., 2002) and (Bielinski et al., 2003). The
presented paper describes new developments in
the rational complex of NARYN sounding data
analysis, which are elaborated to increase the
resolving power of the EM method and finally to
get new assumptions on the deep geoelectric
structure and geodynamics of the Tian Shan
region. The primary step done to achieve higher
resolution consisted in compiling of a new data
ensemble, adequate in precision, comprehension
and dimensionality to the modern requirements of
the profile inversion quality. The new methods
of synchronous sounding data processing
(Varentsov et al., 2003) were applied to
construct the horizontal magnetic tensor (M)
responses and to verify and improve available
long period impedance (Z) and tippers (Wz)
estimates, which together with conditional local
prospecting Z and Wz data have completed the
broadband multi-component profile data set. The
dimensionality and principal directions analysis
for these transfer operators were done with a
help of the phase tensor (Caldwell et al., 2003)
and horizontal magnetic tensor (Varentsov,
Sokolova, 2004) decompositions. It was approved
that the most part of the impedance and
horizontal magnetic tensor data are well
satisfied to 2D approach in the whole frequency
range considered, while the tipper data above
2,5-3 hour seriously deviate from the general
direction of EM field polarization correspondent
to the sub-latitude regional tectonics. The
general strategy of the 2D inversion implies the
combination of the interactive successive partial
inversions approach (Dmitriev et al., 2003) and
joint weighted multi-component inversion
(Varentsov, 2002), verifying and complimenting
each other. The both approaches are based on the
assumption of the priority of the geomagnetic and
impedance phase data role in the inversion
course.
3
1. Outlook of the previous EM studies
N
S
The trial and error fitting of the real tipper
data 0.1-1600 s (broadband soundings with CES-2
instruments in the sites shown by numbers).
The favor of the performed interpretation it
has revealed for the first time the correlation
of the regional crustal conductors with the low
velocity zones detected by seismic tomography,
that can be reasonably explained by fluids
saturating deep horizons and fault zones.
Disadvantages restricted frequency range and
component ensemble.
Fig. 1.1. Geoelectric and seismic cross-sections
along a profile crossing Kyrgyz Tien Shan 300 km
east from NARYN line (Trapeznikov et al., 1997)
a the resistivity profile (shown inside the
model are resistivities in Ohm?m NL Nikolaevs
faults line, AIF Atbashi-Inylchek faults) b
the Vp velocity (in km/s) profile from seismic
tomography according to (Roecker et al., 1993),
the low velocity zones are hatched.
N
S
NL AIF

• The successive partial inversions of long period
  (20-20000s) bimodal MV and MT data (14 LIMS
  stations, 1999) with automatic regularized
  inversion code (Varentsov, 2002).
• The main advantage of the approach the priority
  of the MV data, getting free from the distorting
  influence of near-surface inhomogeneties with the
  lowering frequencies and clearing the information
  on the deep structures.
• The rough block approximation of the conductivity
  section (40 blocks) and low period range limited
  the resolution.

Fig. 1.2. Geoelectric and seismic cross-sections
along the NARYN profile (Rybin et al., 2001).
a,b see the legend for Fig.1.1
(Rodi and Mackies, 2001) regularized bimodal
inversion of 19 MT and MV long period LIMS
(1999-2000) and 30 broadband CES-2 soundings
data looks like the most comprehensive
approach. However, resulted mosaic structure
of high and low conductive bodies still contains
artifacts and instabilities of this preliminary
inverse problem solution, starting from a
homogeneous conductivity distribution in the
upper 150 km.
Fig.1.3. Geoelectric cross-sections along the
NARYN profile (Bielinski et al.,2003).
4
2. New approaches to NARYN EM data processing
site 407
The original tipper and impedance estimates on
LIMS data were obtained with a standard equipment
software and adapted A. Chave code (Rybin et al.,
2000) . Basing on the analysis of multi-site
data records we reprocessed the observations
estimating local impedances Z and tippers Wz as
well as synchronous geomagnetic operators (Fig.4)
in Remote Reference and Multi-Remote Reference
modes (Varentsov et al., 2003). Multi-RR stacking
offered an extra statistic dimension and improved
the stability of responses at the longest
periods. In general, NARYN data are
characterized by low level of noises, but in a
few sites a new sorting strategy based on the
control of horizontal magnetic field spatial
variability (RRMC method,Varentsov, Sokolova,
This Workshop) increased the accuracy of the
estimates.
Fig. 2.1. The comparison of the original Z
processing results (1999) (Rybin et al., 2000)
for site 407 with the new ones (2004), obtained
with a help of PRC_MTMV system in RRMC mode
(Varentsov et al., 2003) upper panel - the
impedances (left main right additional) and
the lower panels the corresponding phases.
Fig. 2.3. The apparent resistivity curves for
long period MT sounding site 401-414 (Ro_xy, NS
upper left panel and Ro_yx, EW lower left
one) and correspondent main impedance phases
(right panels) (PRC_MTMV processing system (2004,
Varentsov et al., 2003).

• Fig. 2.2. The comparison of the original Wz
  processing results (1999) for site 406 (upper
  panel) and site 404 (lower panel) with the
  corresponding new ones (2004_WZ tipper 2004_SZ
  synchronous tipper with the base site 410).

5
3. MT/GDS data set on NARYN transect
local transfer operators

• Fig.3. Pseudosections of local transfer
  operators along NARYN profile.
• LIMS and CES-2 data 0.1-32680 s phases of Z_det
  (Rybin et al., 2000) (a)
• LIMS and Broad Band data 0.1-32680 s amplitudes
  of the real and imaginary induction vectors
  (Rybin et al., 2000) (b)
• LIMS data 20-32680 s phases of the main
  impedances (Rybin et al., 2000) (c) and
  processing results of NARYN WG_2004 (d).

Fig. 3, 4 illustrate the representativeness of
NARYN data set and reflect a wide spectrum of
inhomogeneties on the different structural levels
of geoelectric section.
a
b
c
d
6
4. Extension of the data set inter-station
transfer operators
411 410
Horizontal M H(r) M(r,r)H(r) , and vertical
Sz (regional tipper) Hz(r)Sz(r,r)H(r)
synchronous operators were reliably estimated for
the periods up to 46000 s for 12 LIMS sites with
RRMC technique (Varentsov, Sokolova, This
Workshop) for two reference bases 410 ( for 10
sites of southern synchronous array) and 411
(for 4 northern ones).
Fig.4.1. Pseudosections of amplitudes (top) and
phases (bottom) of Mxx component of the
horizontal magnetic tenso (HMT)r.

• Fig.4.2. Pseudosection of HMT invariant AMP_Mma
  (amplitude of M tensor component in max
  direction) according to Swift (top), and
  regional tipper Sz amplitudes of the real
  induction vectors (middle) and imaginary ones
  (bottom).

7

• 5. Invariant analysis

N
S
Fig.5.2. Pseudosections of the maximal impedance
phases (left) and impedance Skew (right)
amplitude (top) and phase tensor (bottom)
transformations.
Recently suggested scheme of phase tensor and
horizontal magnetic tensor (HMT) decomposition
(Caldwell et al., 2004 Varentsov, Sokolova,
This Worckshop) became the preferable instruments
of the invariant analysis . Their results are
essentially less dependent from the galvanic
effects (compare with conventional amplitude
transformations in Fig. 5.1-5.2). The impedance
data are rather well satisfied to 2D approach in
the whole frequency range considered, while the
tipper data above 2,5-3 hour deviate from the
general sub-latitude polarization correspondent
to the regional tectonics. The HMT principle
directions are fitting well to correspondent
impedance phase tensor invariants. Strikes are
almost profile-perpendicular skews in most of
sites are quite low for periods from hundreds of
seconds till 4-6h and justify two-dimensionality
of impedance and HMT responses in a broad period
range.
Fig.5.1. Vector diagram of max (blue) and min
(red) impedances according to (Eggers, 1982)
(top panel) and (Caldwell et al., 2004)
(bottom). Horizontal axes show the periods and
the vertical ones sounding sites along NARYN
profile from the North to the South. The linear
scale length are also shown on the bottom of the
figures.
Fig.5.3. Real (blue) and imaginary (red)
induction vectors (left) and anomalous horizontal
magnetic perturbation vector diagrams (right).
The legend similar to Fig. 5.1.
8
6. Problems of the data set at long periods
T32768 s
Fig. 6.2. The map of real (black) and imaginary
(red) induction vectors at NARYN profile and in
the surrounding Kyrgyz Tian Shan area.
Fig.6.1. The frequency distribution of the real
(blue) and imaginary (red) induction vectors (top
panel) and graphs of the amplitudes (middle
panel) and azimuths (bottom) of Real ones for 7
GDS sites in the region of Kyrgyz Tian Shan (look
at the map of NARYN profile to find sites
location).

• The deviation of the induction vectors (Tgt 2,5-3
  hour) from the general sub-meridional direction
  and their regular sub-latitude pointing at daily
  harmonics is most probably produced by Sq
  source effect. It prevents the usage of long
  period (Tgt2.5 h) tipper data in the conventional
  2D inversion. The presence of significant
  vertical component in the external field (like in
  the case of polar source of BEAR array sounding
  (Sokolova et al., This Workshop) put also under
  question the perspectives of long period NARYN
  MT data interpretation in the traditional
  plane-wave paradigm despite their obvious 2D
  character, regular behavior and even approaching
  the MV global curve (s. 403, for example). The
  same may be addressed to HMT data at Sq
  harmonics. The question definitely needs further
  investigation. The estimation of the generalized
  (6-component) impedance tensor (Dmitriev,
  Berdichevsky, 2002) could be the next step to
  clarify the situation with long period NARYN
  data interpretation.

Site 403
Site 412
Fig. 6.3. The frequency distributions of the real
(blue) and imaginary (red) induction vectors
(bottom row of plots) and MT curves (amplitude,
top panels, and phases, middle panels) for sites
403 (left) and 412 (right). For s. 403 the
comparison of new MT processing results (RRMC
method, Varentsov and Sokolova, This WS) with the
previous ones (Rybin et al. 2000) are shown,
while for s. 412 the cloud (just simply line)
of estimates for different spectral windows. In
the vector plot for s. 403 Sz and Wz (RRMC and
SS, single site estimates) are compared, while
for s. 412 only RRMC estimates of Wz and Sz
vectors.
9
7. Inversion with the priority of the geomagnetic
data
a
b
A
The main advantage of MV data is that with
lowering frequency the near-surface distortions
of the magnetic field attenuate and vanish, and
hence do not spoil the information on the deep
structures. So, we avoid the diffi-culty
associated with static shift of the apparent
resisti-vity curves, and the geoelectric
const-ructions become more reliable. Our
experience demon-strates the efficiency of an
interpretation scheme based on successive partial
inversions with the priority of geomag-netic data
and reasonably chosen staring model.
d
c
e
f
Fig. 7.1. The tests simulating the integrated
interpretation of MV and MT data for Tian
Shan- type models in the course of successive
partial inversions (SPI) for different data
component. (a) - true model and (b) - blocky
starting model for the SPI-interpretation. (c )
resistivity section obtained from tipper
inversion with REBOCC code with starting model
xxx (resistivity isolines in Omm) (d) results
of blocky inversion of tippers with starting
model (b) with (Varentsov, 2002) code
(resistivity values inside blocks in Omm) (e)
the same results for transverse ap. resistivity
inversion (f) longitudinal impedance phases
and (g) combined data ensemble inversion.
g
Fig.7.2. Resistivity model (Omm) along NARYN
profile inversion of the observedl Re Wz data (
13 periods from 25. to 1600 s at 19 sites) with
REBOCC code (6 iterations, RMC0.5). The results
are shown for (a) homogeneous half space (100
Omm) starting model (b) layered Earth with
gradient interlayer boundaries, including crustal
conductor at depths 30-50 km.
10
Conclusions Basing on the improved transfer
operators estimates and results of their
invariant analysis, the quasi two-dimensional
multi-component data ensemble for profile
bi-modal inversion on NARYN transect have been
formed. The ensemble incorporates amplitude and
phase impedance, tipper and horizontal magnetic
tensor components. Each data component are
supplied with the mask of weights, reflecting the
objective measure of its deviation from 2D (skew
and angle parameters) and accuracy. The general
weighting scheme suggests the priority of the
phase invariants in the whole frequency diapason
and geomagnetic data - at the short and middle
period range, where they are more protected from
disturbing subsurface galvanic effects. The
application of different 2D inversion strategies
is in the progress, including the inversion of
pure geomagnetic ensembles, the successive
inversion of single components, and the use of
static shift correcting routines within bi-modal
multi-component inversion solutions. The first
experiments with separate TM/TE impedance and
single tipper inversions confirm the importance
of geomagnetic responses in the resolution of
deep structures.The progress in the NARYN EM
data inversion we connect with the application of
the powerful stabilized inversion technique
using the piece-wise continuous conductivity
structure approximation (Varentsov, 2002
Varentsov, 2004) and combining the advantages of
the traditional block parameterization with the
abilities of modern scanning schemes for the
arbitrary conductivity distribution in the
selected windows of the section.
Acknowledgements To the fruitful atmosphere of
the NARYN WG collaboration. This study was
supported by grant RFBR 04-05-64970

• References
• Berdichevsky, M.N., Dmitriev V.I., et al., 2003,
  On magnetovariational sounding new
  possibilities, Phys. Solid Earth, 39 (09),
  701-727.
• Bielinski R.A., et al., 2003, Lithospheric
  heterogeneity in the Kyrgyz Tien Shan imaged by
  magnetotelluric studies. GRL, V. 30, N15, 1806.
• Caldwell T.G. et al., 2004, The magnetotelluric
  phase tensor. GJI, 158, 457-469.
• Dmitriev, V.I., Berdichevsky, M.N., 1979, The
  fundamental model of magnetotelluric
  prospecting, IEEE Proc., 67,1034-1044.
• Dmitriev V.I., Berdichevsky, M.N., 2002, A
  generalized impedance model, Izvestya, Phys.
  Solid Earth,38 (10), 897-903.
• Eggers D.E. An Eigenstate formulation of the
  magnetotelluric impedance tensor.Geophysics.
  1982. V.47. ?. 1204 - 1214.
• Roecker S.W. et al., 1993, Three-dimensional
  Elastic Wave Velocity Structure of the Western
  and Central Tien Shan, JGR. V. 98. ? B9., 15779
  15795.
• Rybin A., et al. 2000, Magnetotelluric and
  magnetovariation studies of the Estern Kyrgyz
  Tien-Shan, Abstracts of Inernational Workshop
  Geodynamics of the Tien-Shan, Bishkek,
  Kyrgyzstan, p.70-71.
• Sokolova E.Yu., Varentsov Iv.M., 2004. The
  Effective Source for the BEAR EM Array Sounding
  on the Baltic Shield. This Workshop.
• Trapesnikov et al., 1997, Magnetotelluric
  soundings in the mountains of the Kyirgyz Tien
  Shan, Physics of the Earth, 1, 3-20.
• Varentsov Iv.M., 2002, A General Approach to the
  MT Data Inversion in a piecewise-continuous
  medium, Izvestya, Phys. Solid Earth, 38(11),
  913-934.
• Varentsov Iv.M. et al., 2003. System of EM Field
  Transfer Operators for the BEAR Array of
  Simultaneous Soundings Methods and Results.
  Izvestya, Phys. Solid Earth, 39(2),118-148.
• Varentsov Iv.M., Sokolova E.Yu., BEAR WG, 2003.
  Diagnostics and Suppression of Auroral
  Distortions in the Transfer Operators of the EM
  Field in the BEAR Experiment. Izvestya, Phys.
  Solid Earth, 39(4), 283-307.
• Varentsov Iv.M., Sokolova E.Yu., 2004. The
  Multi-site Estimation of MT/GDS Transfer
  Functions with Horizontal Magnetic Control. This
  Workshop.
• Varentsov Iv.M., EMTESZ-Pomerania WG, 2004. The
  Estimation and Analysis of Horizontal Magnetic
  Inter-station Transfer Functions in the
  EMTESZ-Pomerania project. This Workshop.