PN TOMOGRAPHY WITH LATERAL VARIATIONS OF THE UPPER MANTLE GRADIENT
PHILLIPS, W.S., BEGNAUD, M.L., ROWE, C.A., STECK, L.K., Los Alamos National Laboratory, Los Alamos, NM
87545, wsp@lanl.gov; MYERS, S.C., PASYANOS, M.E., Lawrence Livermore National Laboratory, Livermore, CA
94550, smyers@llnl.gov; BALLARD, S., Sandia National Laboratory, Albuquerque, NM 87185,
sballar@sandia.gov.

The effect of an upper mantle velocity gradient on seismic Pn and P arrival times, measured at distances
of 2.5 to 18 degrees, can be approximated by a simple linear term based on the work of L. -S. Zhao (e.g.
JGR, 98, 1993). This gradient term can be extended to two dimensions for use in tomographic studies. To
demonstrate this, we add a laterally varying upper mantle gradient to the standard Pn time-term
tomography technique, and apply to an Asia data set based on high-quality event locations. We observe
strong lateral variations in upper mantle gradient, ranging to nearly 0.003 s-1. High gradients appear
to be associated with regions of tectonic convergence, both continental and oceanic. The continental
gradient pattern is broad, and is correlated with crustal thickness, but not with upper mantle velocity,
which varies more rapidly. Variance reduction is 66% with respect to Pn tomography without gradients.
Residual levels are high, with respect to previous Asia studies, but this results from different data
quality control procedures, and our use of longer ray paths, rather than from differences in inversion
methods. The use of gradients allows a larger range of path lengths to be used in tomographic inversion,
and increases definition of Pn velocity images in high-gradient regions. In particular, low velocity
anomalies in north-central Tibet and in the Songpan-Ganzi terrain match results obtained by Liang and
Song (GRL, 33, 2006), who added data from temporary deployments to improve coverage in this region of
sparse station density.