Three-dimensional seismic anisotropy in the Pacific upper mantle from inversion of a surface-wave dispersion data set

Abstract

SUMMARY We present a new, 3-D model of seismic velocity and anisotropy in the Pacific upper mantle, PAC13E. We invert a data set of single-station surface-wave phase-anomaly measurements sensitive only to Pacific structure for the full set of 13 anisotropic parameters that describe surface-wave anisotropy. Realistic scaling relationships for surface-wave azimuthal anisotropy are calculated from petrological information about the oceanic upper mantle and are used to help constrain the model. The strong age dependence in the oceanic velocities associated with plate cooling is also used as a priori information to constrain the model. We find strong radial anisotropy with vSH > vSV in the upper mantle; the signal peaks at depths of 100–160 km. We observe an age dependence in the depth of peak anisotropy and the thickness of the anisotropic layer, which both increase with seafloor age, but see little age dependence in the depth to the top of the radially anisotropic layer. We also find strong azimuthal anisotropy, which typically peaks in the asthenosphere. The azimuthal anisotropy at asthenospheric depths aligns better with absolute-plate-motion directions while the anisotropy within the lithosphere aligns better with palaeospreading directions. The relative strengths of radial and azimuthal anisotropy are consistent with A-type olivine fabric. Our findings are generally consistent with an explanation in which corner flow at the ridge leads to the development and freezing-in of anisotropy in the lithosphere, and shear between the lithosphere and underlying asthenosphere leads to anisotropy beneath the plate. We also observe large regions within the Pacific basin where the orientation of anisotropy and the absolute-plate-motion direction differ; this disagreement suggests the presence of shear in the asthenosphere that is not aligned with absolute-plate-motion directions. Azimuthal-anisotropy orientation rotates with depth; the depth of the maximum vertical gradient in the fast-axis orientation tends to be age dependent and agrees well with a thermally controlled lithosphere–asthenosphere boundary. We observe that azimuthal-anisotropy strength at shallow depths depends on half-spreading rate, with higher spreading rates associated with stronger anisotropy. Our model implies that corner flow is more efficient at aligning olivine to form lattice-preferred orientation anisotropy fabrics in the asthenosphere when the spreading rate at the ridge is higher.