Multi-Step Modeling of Receiver Functions and Surface Wave Dispersion
It is argued that a Multiple-Step Procedure in joint modeling of surface
wave group velocity dispersion curves and teleseismic receiver functions should
be used whenever possible for inferring details of crust and upper mantle
structure. This imaging technique is a great complement to images inferred from
seismic tomography. The method relies on an initial grid-search for a simple
crustal structure, and is followed by an inversion. An additional grid search
for shear wave velocity in the mantle qickly converges to a solution that
explains long period deispersion. Finally, a forward modeling of polarization
anisotropy is needed to resolve the fits of both Love and Rayleigh surface wave
dispersion, which, in many cases, is impossible to achieve if only an isotropic
model is considered. The multi-step method thus results in a fit of long period
surface wave dispersion, while preserving the fit to the observed receiver
functions and short period surface wave dispersion. It gives an improved
resolution of lithosphere, achieved in an intuitive way, with the understanding
of which part of the model is responsible for which data.
The grid search for simple crustal structure (STEP 1) is facilitated using a
library of pre-computed receiver functions (RFs) and surface wave group velocity
dispersion curves. An example of a grid-search for receiver functions with 4
variable-thickness layers in the crust and a half-space in the mantle is shown
for station QIZ in China, operated by the Global Seismographic Network.
Grid-search inversion results (for 4 layers in the crust and a half-space in the uppermost mantle, for 1D Vs velocity structure at GSN station QIZ. The observed RFs are shown with thick gray lines, and the synthetic RFs from the best 10 models are shown with yellow thin lines. Best-fitting 1000 models are displayed on the right side utilizing the logarithmic color scheme. PREM is indicated with black thick lines.
|
The multi-step method is illustrated in 4 figures below, which show a gradual
progress in fitting the observed receiver functions and surface wave dispersion
curves for station KBRS in Saudi Arabia. Initial model obtained from a
grid-search (STEP 1) only fits receiver functions well. STEP 2 (linearized
inversion) further improves fits to receiver functions, but does not fit
dispersion curves at intermediate and long periods. Additional grid-search for
structure in the mantle dramatically improves the fit to the long-period
Rayleigh wave dispersion (STEP 3), however the fits to Love wave dispersion are
still poor. A final solution is obtained through STEP 4, in which the
polarization anisotropy is introduced in the crust and upper mantle (dashed
lines) to match both Rayleigh and Love wave dispersion (the final fit of the
dispersion data is shown with dashed lines). | 
STEP
1 |
STEP 2
|
|
STEP 3
|
STEP 4
|
The method is applied to ten stations in the Arabian Peninsula sampling various
tectonic environments including active continental rifting and stable regions
(Tkalčić et al., JGR, 2006). Further variations of the method
have so far been applied to stations in China, Australia and Croatia, with a
combination of grid-search with Monte Carlo and Neighborhood Algorithm
inversions ( Chen et al., JGR, 2010; Stipčević et al.,
in press in GJI; Fontaine et al., to be submitted;
Tkalčić et al., to be submited). Interactive Receiver Function
Forward Modeller (IRFFM) is a Java program for interactive modeling, and it is
freely available from this web site (Tkalčić and Banerjee,
2009).
We are currently working on IRFFM2, a software for an
interactive modeling of receiver functions and surface wave dispersion,
including polarization anisotropy.
Observation of strong polarization anisotropy (up to 12%) in the
crust and upper mantle beneath Arabian Peninsula
In the study of lithospheric structure of Saudi Arabia, the iterative inversion
improves fit to the data by increasing the number of layers in the crust when
necessary. In order to fit the surface wave group velocity for periods greater
than about 50 seconds, we perform a grid search over mantle velocities including
the mantle lid and low-velocity zone (LVZ), keeping the crustal structure fixed
to the values from the previous step. In some cases a clear Love-Rayleigh
discrepancy prevents a simultaneous fit of the group velocities with an
isotropic model. The Love-Rayleigh discrepancy can be resolved by allowing shear
wave transverse isotropy (a.k.a. polarization anisotropy) with a vertical
symmetry axis (Vsh - Vsv differences) in the uppermost mantle.
The resulting shear velocity models confirm rapid crustal thinning of the
Arabian Shield toward the Red Sea, however we do not find strong evidence for
crustal thickening towards the Arabian Platform. Our results suggest that the
mantle lithosphere thickness varies regionally but that the mantle shear
velocities beneath the Arabian Shield and Red Sea coast are generally
anomalously low. Furthermore our results indicate the presence of strong
polarization anisotropy (up to about 12%) in the lithospheric upper mantle, in
the vicinity of, as well as farther away from the Red Sea. Our modeling yields
Vsv > Vsh in the southwestern part of the Arabian Peninsula, consistent with
vertical flow, and Vsh > Vsv in the northwestern part of the Arabian Peninsula
and the continental interior, consistent with horizontal flow, indicating that
the mantle flow pattern is not uniform along the axis of the Red Sea.
Map of station locations (triangles) and the values estimated in the multi-step modeling of receiver functions and surface waves. Line 1: Moho depth, shear wave velocity values in the layer above and the layer below Moho; Line 2: lithospheric lid thickness (NP Ð not pronounced), maximum shear wave velocity in the lid (Vsv or Vsh); Line 3: thickness of the low velocity zone, minimum shear wave velocity in the LVZ (Vsv or Vsh); maximum percentage of Vsh>Vsv transverse isotropy.
|
This is an electronic version of an article published by Journal of Geophysical Research ; Copyright (2006) American Geophysical Union:
Tkalčić, H., M. Pasyanos, A. Rodgers, R. Gok, W. Walter and A. Al-Amri, Multistep method in joint modelling of receiver functions and surface waves: Implication for lithospheric structure of the Arabian Peninsula, Journal of Geophysical Research, B11311, doi:10.1029/2005JB004130, 2006
Download this paper 
also available on the Journal of Geophysical Research online delivery
service, accessible via the journal's Website at:
http://www.agu.org/journals/jb/
back to homepage