Complex Seismic Anisotropy at the Edges of a Very-low Velocity Province in the Lowermost Mantle

A prominent very-low velocity province (VLVP) in the lowermost mantle is revealed, and has been extensively mapped out in recent seismic studies (e.g., Wang and Wen, 2004). Seismic evidence unambiguously indicates that the VLVP is compositionally distinct, and its seismic structure can be best explained by partial melting driven by a compositional change produced in the early Earth's history (Wen, 2001; Wen et. al, 2001; Wang and Wen, 2004). In this presentation, we study the seismic anisotropic behavior inside the VLVP and its surrounding area using SKS and SKKS waveform data. We collect 272 deep earthquakes recorded by more than 80 stations in the Kaapvaal seismic array in southern Africa from 1997 to 1999. Based on the data quality, we choose SKS and SKKS waveform data for 16 earthquakes to measure the anisotropic parameters: the fast polarization direction and the splitting time, using the method of Silver and Chan (1991). A total of 162 high-quality measurements are obtained based on the statistics analysis of shear wave splitting results. The obtained anisotropy exhibits different patterns for the SKS and SKKS phases sampling inside the VLVP and at the edges of the VLVP. When the SKS and SKKS phases sample inside the VLVP, their fast polarization directions exhibit a pattern that strongly correlates with stations, gradually changing from 11°N~to 80°N~across the seismic array from south to north and rotating back to the North direction over short distances for several northernmost stations. The anisotropy pattern obtained from the analysis of the SKKS phases is the same as that from the SKS phases. However, when the SKS and SKKS phases sample at the edges of the VLVP, the measured anisotropy exhibits a very complex pattern. The obtained fast polarization directions change rapidly over a small distance, and they no longer correlate with stations; the measurements obtained from the SKS analysis also differ with those from the SKKS analysis. As the SKS and SKKS phases have similar propagation paths in the lithosphere beneath the array, but different sampling points near the core mantle boundary. The anisotropy in the lithosphere should have a similar influence on SKS and SKKS phases. Therefore, the similar anisotropy obtained from the SKS and SKKS phases sampling inside the VLVP and its correlation with seismic stations suggest that the observed anisotropy variation across the seismic array is mainly due to the anisotropy in the lithosphere beneath the Kaapvaal seismic array, and the interior of the VLVP is isotropic or weakly anisotropic. On the other hand, for the SKS and SKKS phases sampling at the edges of the VLVP, the observed complex anisotropy pattern and the lack of correlation between the results from the SKS and SKKS analyses indicate that part of that anisotropy has to originate from the lowermost mantle near the exit points of these phases at the core mantle boundary, revealing a complex flow pattern at the edges of the VLVP.