The three-dimensional shear velocity structure of the mantle from the inversion of body, surface and higher-mode waveforms

Abstract

SUMMARY We present a 3‐D model of shear heterogeneity (Vsh) in the whole mantle, derived from the inversion of hand‐picked body, surface and higher‐mode waveforms. The forward and inverse problems are formulated using the non‐linear asymptotic coupling theory, the zeroth‐order asymptotic expansion of a Born seismogram computed by normal mode summation and including coupling across mode branches. We invert iteratively for mantle heterogeneity and for centroid moment tensors until convergence. Our model, SAW24B16, is parametrized laterally in spherical harmonics up to degree 24 and radially in 16 cubic b‐splines with knots spaced to reflect the data sampling with depth. The power spectrum of the model is dominated by heterogeneity in the boundary layers at low degrees: degrees 5 and 6 near the surface and degree 2 near the core–mantle boundary. The rest of the lower mantle is dominated by degree 3. We find significant heterogeneity up to spherical harmonic degree 24 in the transition zone and the uppermost part of the lower mantle. Our model displays three slow domains extending continuously between uppermost and lowermost mantle, one originating under Africa and two in the Pacific. The results of our modelling also suggest an interaction near the surface between the anomalies in the Pacific and the network of mid‐ocean ridges in the East Pacific, and between the African anomaly and the low‐velocity zone in the Red Sea, the East African Rift, the Mid‐Indian Ridge and the Mid‐Atlantic Ridge. These anomalies appear to ascend into the upper mantle without altering their shape near the 670 km discontinuity. Fast anomalies accumulate near the 670 km discontinuity under the Southern Kurile, the Japan trench and the Izu arc, with little or no penetration into the lower mantle, in agreement with results from a variety of regional studies. Cross‐sections through subduction zones such as the Marianas, Tonga–Kermadec, Java, the palaeosubduction zone beneath North America and the Peru–Chile trench show continuous fast structures penetrating into the lower mantle, with the depth of penetration varying from region to region. The distribution of both fast and slow anomalies near 670 km thus suggests that the discontinuity does not act as a strong barrier to upgoing and downgoing flow between the upper and the lower mantle.

Short scale heterogeneity in the lowermost mantle: insights from PcP-P and ScS-S data

“…2). As expected, R is largest (2.7) for model SAW24B16 [3], which has stronger velocity rms than other models in the last 500 km of the mantle (solid line). We obtain R = 1.8 at the bottom of the mantle for both SB4L18 [4] and S362D1 [5] (thick dashed line).…”

“…1a,b shows the global distribution of PcP-P travel time residuals, plotted at the surface projection of PcP re£ection points on the core^man-tle boundary (CMB), with an S and a P tomographic model in the background ( [3] and [23], respectively). We selected only data corresponding to epicentral distances v 55 ‡, for which paths of P and PcP are similar, except in the lowermost mantle, so that the travel time residuals should be most sensitive to the structure in the lowermost mantle.…”

“…However, not all of them have negative PcP-P residuals, and there are regions with smaller scale lateral variations than indicated by tomographic maps, which deserve more detailed investigation, as will be discussed below. [3]; (b) P velocity model by [23].…”

Morphology of seismically slow lower-mantle structures

“…When applied to profiles of the entire lower mantle, k-means clustering of five recent tomographic models (Mégnin & Romanowicz 2000;Houser et al 2008;Kustowski et al 2008;Simmons et al 2010;Ritsema et al 2011) retrieves the geographic extents of the LLSVPs, as well as the LLSVP-like, but geographically isolated meso-scale structure called the Perm anomaly (Lekic et al 2012).…”

A global horizontal shear velocity model of the upper mantle from multimode Love wave measurements

“…Trampert & Woodhouse 1995Larson & Ekström 2001;Visser et al 2007;Ekström 2011;Pasyanos et al 2014) and Rayleigh wave maps exist for some higher modes (van Heijst & Woodhouse 1997, 1999Shapiro & Ritzwoller 2002;Beucler et al 2003;Visser et al 2007Visser et al , 2008. Less work has been done on inverting Love wave measurements for β h (z) models (Li & Romanowciz 1996;Mégnin & Romanowicz 2000;Zhou et al 2005) and most existing β h (z) upper-mantle models are constrained using only fundamental mode measurements.…”