Anisotropy

As discussed in section 2.1.3, the development of the Lehmann discontinuity is hypothesised to be the result of a change of the anisotropy structure due to a change in the creep mechanism. To test whether this mechanism can be the explanation for the L, the effect of the anisotropy on SS phases and its precursors must be studied.
One way to study the anisotropy of the mantle material is to investigate shear wave splitting. Anisotropy causes shear waves to split into two pulses with different polarizations, one travelling faster than the other, with the differential travel time accumulating with the path length traversed in the anisotropic region and the amount of anisotropy.
The shear wave splitting is studied by long-period or broad-band 3-component stations. YKA is only equipped with 4 broadband, 3-component stations with an aperture of 10 km. The much larger temporary Tw~st array is better suited to study the long-period SS and S$^d$S waves.
The upper mantle anisotropy is concentrated in the top 300 km of the mantle (Nishimura and Forsyth, 1989). Most of the rest of the upper mantle is highly isotropic. By using deep focus earthquakes, the source side anisotropy can be neglected. The upper mantle anisotropy beneath the Tw~st array station has been studied (Kay et al., 1999), and its signal can be removed from the data. The splitting parameters for all Tw~st array stations are listed in Table A.2.
SS phases travelling through the anisotropic upper mantle near the surface reflection point show large splitting times (Wolfe and Silver, 1998) even after the data were corrected for upper mantle anisotropy. The S$^d$S phases are used to study the depth structure of the anisotropy. If the precursor shows no splitting, the anisotropic layer is above the discontinuity. If the anisotropy is located above the Lehmann discontinuity, the change of the anisotropy structure is a likely explanation for this discontinuity.
After the correction for the upper mantle anisotropy beneath the stations, 4th-root vespagrams for the R and T components are computed and examined for S$^d$S precursors. The slowness and backazimuth resolution of the Tw~st array is sufficient to resolve S$^d$S. The time windows of S$^d$S (d $\le$ 450 km) are studied using the particle motion of the R and T components. The particle motion of SS is circular polarized as a result of the splitting. If S$^d$S shows linear polarization of the particle motion the phase does not travel through anisotropic regions other than the area beneath the station. Additionally, the time shifts between R and T components are controlled by the superposition of the traces.
As discussed in section 4.3, only a few events with reflection points in the Pacific were recorded at Tw~st. Shallow events were also used to obtain a larger dataset. The effect of the source side anisotropy for the shallow events is still in the data after the upper mantle correction for the mantle directly beneath the station. The extent of this effect can be checked using particle motions of the direct S phase. Qualitatively, this study can give an idea of the depth structure of the upper mantle anisotropy and the origin of the Lehmann discontinuity.