Electronic Supplement for Seismic Tomography of the Uppermost Inner Core
To confirm the accuracy of our forward matrix provide set a point for comparison, we first performed a damped least squares inversion of the residual travel times relative to AK135. The inversion is performed on a 3◦× 3◦ grid, and smoothing is applied at a level determined by L-curve test. The resulting model is shown in Figure S1. The isotropic model exhibits strong hemispheric structure in addition to regional variations on the scale of ∼30◦, particularly in the western hemisphere. The heterogeneity in the eastern hemisphere varies from ∼0.3 %dV/V to ∼0.8 %dV/V, while the variation in the west is stronger, about -1.0 %dV/V to 0.1 %dV/V. It should be noted that the locations of near-zero velocity perturbation in the west correspond to the locations without adequate ray coverage. The isotropic hemisphere boundaries each vary strongly with latitude. The boundary beneath Africa has a westward bulge of 3–5◦ at 30◦S, while the Pacific boundary trends westward from 210◦ to 180◦ between the northern and southern ends of the boundary. As a result of lateral smoothing, the transitions between the hemispheres appear to take place over 10–15◦ in longitude. The anisotropic part (b + c) has no discernible hemispheric structure. The damping limits anisotropy primarily to regions with rays traveling in both the axial and equatorial