Constraints on the correlation of P- and S-wave velocity heterogeneity in the mantle from P, PP, PPP and PKPab traveltimes



We investigate the correlation of large-scale P- and S-velocity heterogeneity in the mantle by determining how well 106,000 compressional P, PP, PPP, and PKPab traveltimes can be explained by S-wave velocity model S20RTS (scaled using a depth dependent factor) and by a model in which the lateral P-velocity variations are different. We first assess the assumption that P-wave traveltimes can be explained by a model in which lateral P-velocity variations (δvP) are identical to S-velocity variations (δvS) in model S20RTS. For a given depth, we project δvS from S20RTS into model S2P using a depth-dependent scaling factor inline image defined as: inline image. We find, by grid search, that the highest reduction of data variance is obtained when inline image increases linearly from 1.25 at the surface to 3.0 at the core–mantle boundary. A comparison of S-wave (+SS) and P-wave (+PP) traveltimes for identical source–receiver pairs also indicates that inline image increases with depth. Significantly higher variance reduction is not obtained when inline image is parametrized with an additional degree of freedom. Therefore, the precise shape of inline image cannot be constrained by our data.

P- and PP-wave traveltime anomalies with respect to the scaled model S2P yield coherent geographic variations. This indicates that there are large-scale lateral P-velocity variations in the mantle that are different from those in model S2P. We estimate these variations by inverting P-wave traveltime anomalies with respect to model S2P for a degree 12 model of P-velocity heterogeneity. This model, P12s2p, indicates where in the well-sampled mantle regions we need to modify model S2P to further improve the fit to the traveltime data. Anomalies in P12s2p exist throughout the mantle. It is, therefore, not obvious that compositional heterogeneity is prominent in the lower 1000 km of the mantle only, as suggested previously. Low P-wave velocities in the upper mantle beneath oceans are the strongest anomalies in P12s2p and explain better the delayed traveltimes of PP-wave phases with oceanic surface refection points. Lower mantle anomalies include high and low P-velocity structures beneath eastern Asia and North America, respectively. The high P-velocity anomaly in the lower mantle beneath the central Pacific is consistent with the assertion made by other researchers that large-scale lower mantle upwellings are not purely thermal in origin.