Geoscience Reference
In-Depth Information
from roughly north-south in the northern part
of the arc to roughly east-west further south. The
anisotropic regions were of the order of 100 km
in extent and implied a 4% difference in shear-
wave velocities in the mantle wedge above the
slab.
In another early paper, Fukao (1984) stud-
ied ScS splitting from a deep-focus event in the
Kuriles recorded in Japan. The uniformity in
polarization across the Japanese arc is remark-
able. The faster ScS phase had a consistent polar-
ization of north-northwest -- south-southeast and
an average time advance of 0
than waves that exit the slab earlier. Actually,
the anisotropy implied by the vertical ScS waves
just gives the difference in shear-wave velocities
in that (arbitrary) direction and is not a mea-
sure of the total azimuthal S-wave velocity vari-
ation, which can be much larger. The presence
of near-source anisotropy can give results similar
to those caused by a long cold isotropic slab and
can cause artifacts in global tomographic mod-
els, e.g. fast blue bands in the lower mantle on
strike with the deep focus earthquakes used in
the inversion.
The mineral assemblage in the deeper parts
of the slab are different from those responsi-
ble for the anisotropy in the plate and in the
shallower parts of the slab. The orientation of
high-pressure phases is possibly controlled both
by the ambient stress field and the orienta-
tion of the 'seed' low-pressure phases. Results
to date are consistent with the fast crystallo-
graphic axes being in the plane of the slab. The
most anisotropic minerals at various depths are
olivine (
4s over the
slower ScS wave. The splitting could occur any-
where along the wave path, but the consistency
of the results over the arc and the difference
from the direct S results, from events beneath
Japan, suggests that the splitting occurs in the
vicinity of the source. The fast polarization direc-
tion is nearly parallel to the dip direction of the
Kurile slab and the fast P-wave direction of the
Pacific plate in the vicinity of the Kurile Trench.
The stations are approximately along the strike
direction of the deep Kurile slab. All of this sug-
gests that the splitting occurs in the slab beneath
the earthquake. This earthquake has been given
various depths ranging from 515 to 544 km,
the uncertainty possibly resulting from deep-
slab anisotropy. If the slab extends to 100 km
beneath the event, the observed splitting could
be explained by 5% anisotropy. This event shows a
strong S-wave residual pattern with the fast direc-
tions along the strike direction. The residuals
vary by about 6 s. The waves showing the earliest
arrival times spend more time in the slab than
the nearly vertical ScS waves. They also travel in
different azimuths. If the fast shear-velocity direc-
tions are in the plane of the slab, this will add
to the effect caused by low temperatures in the
slab. Thus, a large azimuthal effect can accumu-
late along a relatively short travel distance in the
slab. If the slab is 5% faster due to temperature
and 5% anisotropic, then rays travelling 300 km
alongthestrikedirectionwillarrive6searlier
.
8
±
0
.
400 km), modified-spinel (400--500 km)
and MgSiO
3
-ilmenite (
<
500 km). The last phase
is not expected to be stable at the higher tem-
peratures in normal mantle, being replaced by
the more isotropic garnet-like phase majorite.
Thus, the deep slab cannot be modeled as sim-
ply a colder version of normal mantle. It differs
in mineralogy and therefore in intrinsic veloc-
ity and anisotropy. Ilmenite is one of the most
anisotropic of mantle minerals (Figure 20.13),
especially for shear waves. If it behaves in aggre-
gate as do ice and calcite, which are similar
structures, then a cold slab can be expected to
be extremely anisotropic. Seismic waves which
travel along the South American slab for large
distances on their way to North American sta-
tions can be expected to arrive very early, espe-
cially at shield stations. Some of this effect may
be mapped as fast bands in the deep man-
tle under North America and the Atlantic and
will
>
be
interpreted
in
terms
of
deep slab
penetration
.
