The GSN station ARU (Arti, Russia) is located on the easternmost edge
of the East European platform (Figure 16), in the Uralian foredeep
[Ivanov et al. 1975; Zonenshain et al. 1984] that accumulated
sediments in a passive continental margin setting through most of the
Paleozoic. The continental collision at the final stage of Uralian
orogeny formed an extensive thrust-sheet complex to the east of ARU,
and may also have deformed the sediments underlying ARU.
[Levin and Park, 1997a] studied crustal anisotropy beneath ARU
using the anisotropic receiver function method.
This method, which utilizes P-S conversions in layered media, is
sensitive to anisotropy adjacent to interfaces where the conversions
occur.
Levin and Park, [1997a] show that receiver function data are consistent
with a vertically-stratified earth structure in which both the
uppermost and the lower crust are anisotropic, with hexagonal symmetry
and strong tilt in the symmetry axes. Unlike typical models of
seismic anisotropy in the mantle, the crustal model for ARU
contains a layer of anisotropy with a slow symmetry axis. A conceptual
model that would exhibit this type of anisotropy is a layer of
imbricated peridotite and metapelite lenses in the region of the
crust-mantle transition, as described by Quick et al. [1995] in the
Ivrea deep crustal exposure in the Alps.
Upper mantle anisotropy under ARU has previously been studied by
Helffrich et al, [1994], who report a mean fast axis strike of 
and a splitting time 
s.
We have compiled splitting data from SKS and other core-refracted
phases for this station (Figure 17). Several populations of
fast-axis strikes are evident. For example, events from the
northwest have westerly strike, while those from the east have more
northwesterly strike. The best-fitting one-layer anisotropic earth
model (Figure 18, top), with fast-axis at 
azimuth, does not reproduce this pattern well. A combination of the
Levin and Park [1997a] crustal model with the mantle model of
Helffrich et al. [1994], does better (not shown), but does poorly at
the ENE back azimuths.
A model with a second mantle layer improves the fit significantly
(Figure 18, bottom).
This best-fitting model (Table 3) has a 58 km upper mantle
layer with a fast-axis striking 
atop a 140 km
layer with fast-axis plunging 40 
to the east (Figure 19).
Figure 20 illustrates a conceptual model of anisotropic layering under
ARU. Lower crust and uppermost mantle under the Uralian foredeep are
characterized by the common direction of anisotropy 
,
which is significantly oblique to the trend of Urals. It may be
related to the deformation within the East European platform that predates the formation of the orogen. The
anisotropy-inducing fabric in the lower part of the lithosphere is
aligned nearly normal to the strike of the Uralian Orogen.