It is critical to resolve the strength, shape and extent of
seismic velocity heterogeneity in the lower mantle in order to
understand how
it relates to the physical and chemical processes that take place in
this region (see *Lay et al.*, 1998 for a review).
The deep mantle beneath the Central Pacific is thought to experience some
vigorous convection, and may be responsible for the largest hotspot observed
at the surface of the earth. Tomographic models of this region consistently
show a large slow anomaly that extends several hundred kilometers
above the core-mantle boundary (CMB) (e.g. *Li and Romanowicz*, 1996).
However, while these models give a reasonable description of the
position of anomalous domains, they do not fully explain
travel times sensitive to structure in that region (e.g. *Bréger et al.*, 1998).
*Bréger and Romanowicz* (1998) showed that it was
possible to use tomographic studies as a starting point for forward modeling
of travel times, and proposed a revised model of the
lowermost mantle beneath the Central Pacific characterized by a large
slow S-velocity anomaly () extending
above D", and adjacent, on its eastern side, to a smaller domain of
very fast S velocities (faster by up to 4 to 5%). They used S-SKS and SKKS-SKS differential residuals,
which have little sensitivity to heterogeneity in the mid-mantle, and thus poor vertical
resolution. In the present study, we apply the forward modeling approach of
*Bréger and Romanowicz* (1998) to an extended dataset which
includes both S-SKS and ScS-S differential residuals
(Figure 25.1) in an attempt to verify our model and constrain it better at
mid-mantle depths.

We use the differential travel time residuals of *Garnero
and Helmberger* (1993), as well as the measurements
of *Vinnik et al.* (1998), and the ScS-S differential
travel time dataset of *Russell et al.* (1998).
Residuals are computed with respect to PREM reference model (*Dziewonski and Anderson*, 1981).
ScS and S arrivals are measured on the transverse component and SKS ones on the radial component. The uncertainty
on measurements is on the order of 1 to 2s. Dispersive effects are
expected to affect measurements of Sdiff travel times so that this
uncertainty should be higher at larger epicentral distances.
The compiled dataset that we use here provides a particularly good
coverage of the lowermost mantle beneath the Central Pacific (Figure 25.2).
We adopt the approach of *Vinnik et al.* (1998), *Bréger et
al.* (1998), and
*Bréger and Romanowicz* (1998), and
analyze the variations of the residuals as a function of epicentral
distance
when the station or the event is fixed, along narrow linear "corridors".

In an attempt to foward model the observed residual variations, we first
computed the residual values predicted by several recent tomographic S-velocity
models, using the 1D raypaths predicted by PREM.
The model that gave the closest match to both S-SKS and ScS-S
observations (SKS12, *Liu and Dziewonski*, 1994) was
chosen as a starting model. This initial model was progressively
perturbed in two ways: (1) by keeping the shape of the heterogeneity
and increasing its strength, and (2) by applying small vertical and
horizontal shifts of no more than 200km. The perturbations which
were applied to the model are obviously non-unique. There are some
trade-offs, particularly between the shape, position, vertical extent,
and S-heterogeneity in this domain. In fact, our approach leads
to a whole family of models rather than one single `best' model.
The features which are common to all the models can be considered
as robust, and are discussed in what follows (Figure 25.2).

We found that the data require a large slow domain extending from the CMB to several hundred kilometers above the CMB, where the S-velocity anomaly is strong and reaches -2 to -4%, which is significantly larger than predicted by SKS12 (Figure 25.2). Predicted residuals fit the observations better when the model is modified up to 1200 km above the CMB, that is, when the slow anomalous domain extends well above the top of D". This does not preclude the possibility of explaining the observations by a more complex structure localized in the last 300 to 500 km of the mantle. However, such a scenario seems unlikely because it would imply a structure that is less compatible with existing tomographic models. On the other hand, we note that differential residuals appear to rapidly lose their sensitivity as heterogeneity gets shallower, so that we cannot preclude that the anomalous domain extends even higher above the CMB.

A few S-SKS profiles for a fixed earthquake show a sudden
increase at small epicentral distance (86 to 90),
followed by a decrease of the residuals at distances
from about 90 to 96 (*Bréger and Romanowicz*, 1998). Increasing residuals
can be easily explained by the fact that S waves spend
more time in the slow region with increasing distance, and therefore
accumulate an increasingly large delay. However, it is impossible
to explain decreasing residuals with the slow structure alone,
and introducing a fast region becomes necessary. This region
needs to be localized in the last 400 km of the mantle in order
to start affecting the S-wave at epicentral distances larger than about 90.
The exact strength, position, and velocity anomaly of this
region are difficult to determine precisely considering that
it is sampled by only a few profiles, but we estimate that it
is a fairly localized domain, with lateral
dimensions on the order of 500 km by 500 km, and extending vertically
to about 400 km above the CMB respectively,
and corresponds to a S-velocity anomaly on the order of 3 to 4%.

Bréger, L., and B. Romanowicz, Three-dimensional structure at the base
of the mantle beneath the Central Pacific,
*Science, 282*, 718-720, 1998.

Bréger, L., B. Romanowicz, and L. Vinnik, Test of tomographic
models of D" using differential travel time data,
*Geophys. Res. Lett., 25*, 5-8, 1998.

Dziewonski, A. M., and D. L. Anderson,
Preliminary Reference Earth Model,
*Phys. Earth Planet., Int 25*, 297-356, 1981.

Garnero, E., and D. V. Helmberger,
Travel times of S and SKS: implications
for three-dimensional lower mantle structure beneath the central Pacific,
*J. Geophys Res., 98*, 8225-8241, 1993.

Lay, T., Q. Williams, and E.J. Garnero,
The core-mantle boundary layer and deep Earth dynamics,
*Nature, 392*, 461-468, 1998.

Li, X.-D., and B. Romanowicz, Global mantle shear velocity model
developed using nonlinear asymptotic coupling theory,
*J. Geophys Res., 101*, 22245-22272, 1996.

Liu, X.-F., W.-J. Su, and A. M. Dziewonski,
Improved resolution of the lowermost mantle shear wave velocity structure
obtained using SKS-S data (abstract),
*Eos. Trans. AGU, 75*, 232, 1994.

Russell, S.A., T. Lay, and E. J. Garnero,
Seismic evidence for small-scale dynamics in the lowermost mantle at the root of the Hawaiian hotspot,
*Nature, 396*, 255-258, 1998.

Vinnik, L., L. Breger, and B. Romanowicz,
Anisotropic structures at the base of the Earth's mantle,
*Nature, 393*, 564-567, 1998.

Questions or comments? Send e-mail: www@seismo.berkeley.edu

© 2001, The Regents of the University of California.