Recent years have seen a large number of observations of azimuthal
anisotropy in the upper mantle from the splitting of teleseismic shear-wave
arrivals. These observations are very useful because they can be made from single
seismic stations, they provide greater horizontal resolution than surface
wave studies, and they sample deeper into the mantle than regional refraction
studies. Splitting in SKS and SKKS arrivals (mantle shear waves that travel as
compressional waves in the core) is particularly easy to interpret since these waves are
radially polarized upon leaving the liquid outer core. Silver and Chan
[1991] examined these phases at 21 global seismic stations, and detected
splitting at most of them, with the delay times between the shear waves varying
from 0.6 s to 1.7 s. They noted an apparent correlation between the orientation of
the anisotropy and the last substantial tectonic episode in the
lithosphere, a probable result of strain-aligned olivine crystals in the uppermost mantle.
Makeyeva et al. [1992] studied SKS and SKKS splitting in the Tien Shan
region of central Asia. They found a general correlation between the fast
anisotropy direction and the axis of the mountain belt, but a 90
shift in the
anisotropy near an inferred region of thermally driven convection.
Kuo and Forsyth [1992] detected SKS splitting at several sites on oceanic islands. Savage and Silver [1993] examined splitting in SKS, SKKS and S arrivals at stations in the western United States and found a complicated pattern of anisotropy, which, in some cases, required two anisotropic layers of differing orientation for successful modeling. Portable seismic arrays can give greater coverage in specific regions than fixed seismic networks, provided sufficient events are recorded during the deployment. Splitting observations from portable experiments were reported by Silver and Kaneshima [1993] for a 1500 km transect across North America and by Shih et al. [1991] above deep earthquakes in Colombia. Teleseismic detection of near-source anisotropy was reported by Kaneshima and Silver [1992] below intermediate focus earthquakes in Peru and Kamchatka from S wave splitting, and by Fischer and Yang [1994] in the Kuril-Kamchatka subduction region from splitting in S and sS waves (S is the direct shear wave, sS is the near-source surface reflection from a deep event).
Attempts to resolve anisotropy near SS (the first surface multiple of S) bouncepoints have produced mixed results. Woodward and Masters [1991] found little indication in long-period SS - S times for azimuthal upper mantle anisotropy in the Atlantic, while Sheehan and Solomon [1992], analyzing similar data, found evidence for anisotropy. Yang and Fischer [1994] attempted to resolve shear-wave splitting in SS arrivals and found some indication of upper mantle anisotropy beneath the Atlantic, with the inferred orientation for the central North Atlantic in approximate agreement with the global anisotropy model of Montagner and Tanimoto [1991]. This global heterogeneity and anisotropy model is based on an inversion of surface-wave phase velocities, and contains significant anisotropy down to depths of about 300 km. However, the degree to which current surface-wave phase-velocity studies can resolve anisotropy is unclear, due to limitations in the path coverage and tradeoffs between anisotropy and heterogeneity in the inversions. Citing a lack of resolution, Zhang and Tanimoto [1993] chose not to include anisotropy in their surface-wave modeling of upper mantle structure.
A different approach to resolving upper mantle anisotropy involves observations of quasi-Love and quasi-Rayleigh waves, anomalous arrivals that are not pure Love or Rayleigh waves that are generated by coupling between the wave types [ Park and Yu, 1992, 1993; Yu and Park, 1993; Park, 1993]. The coupling is most readily explained by models of azimuthal anisotropy rather than lateral heterogeneity. An advantage of these observations is that, like the SKS splitting discussed above, information about anisotropy can be obtained from a single source-receiver path.