Deep Upper Mantle Anisotropy from SKS Splitting Delay Times and Higher Mode Surface Waves
Abstract:The splitting of SKS waves is widely used to study mantle anisotropy and to get insight on Earth's dynamic interior. However, while SKS splitting is an unambiguous signature of anisotropy, it cannot be used directly to infer the depth distribution of the anisotropy because the dealy times between the two polarization directions result from integrated measurements along quasi-vertical paths between the core-mantle boundary and the surface. SKS splitting data are generally interpreted in terms of olivine lattice preferred orientation in the uppermost 250 km of the mantle. There is, nevertheless, growing evidence from regional studies that the splitting may contain significant contributions from greater depths.
In addition, SKS delay times are strongly under predicted by tomographic models of azimuthal anisotropy derived from fundamentqal mode surface waves. These models typically display anisotropy only in the 250 km of the mantle because fundamental mode surface wave data have little sensitivity to structure below that depth. For example, we calculated predictions of SKS splitting using the global azimuthal anisotropy model of Debayle et al. (2005), and found a global delay times distributon with a median of 0.4 s, compared to 1.1 s for the global station-averaged SKS splitting database compiled by Becker et al. (2012). This discrepancy may indicate that seismic anisotropy is present at greater depths.
In this work, we analyzed a new global 3-D model of azimuthal anisotropy (Yuan and Beghein, 2013) that was obtained from the inversion of higher modes surface wave phase velocity data with unprecedented sensitivity to the top 1000 km of the mantle. While this model is comparable to previous models in the uppermost mantle, it also displays anisotropy amplitudes of about 1% below 250 km depth and a complex 3-D pattern of fast axes directions down to 800 km depth. The predicted SKS delay times calculated using this new model have a median of 0.75 s, which is in much closer agreement with SKS measurements than other tomographic models.
This result shows that seismic anisotropy below 250 km contributes significantly to SKS delay times, and future geodynamics interpretation.