A global vertically polarized upper mantle azimuthal anisotropy model
(Jump to Model Downloads)
Here we present two new models of 2Ψ azimuthal seismic anisotropy through the upper mantle and transition zone, computed using a large dataset of ~750,000 vertical component (Rayleigh-wave) seismogram fits. These models were constructed using the automated multi-mode inversion of surface and S wave forms across the broad period band of 11 to 450 seconds (with the best sampling in the 20 to 350 second range). In these new global models, regularization of anisotropy is implemented to more uniformly recover the amplitude and orientation of anisotropy, including near the poles.
The smooth model, SL2016svA, robustly captures the dominant long-wavelength upper mantle structure, demonstrated through its relatively good vertically-averaged correlation coefficients with other azimuthal seismic models and SKS splitting. This model has previously been used in the studies by Becker et al 2014 (EPSL) and Becker et al 2015 (GRL)–then referred to as SL2013svA–comparing global mantle flow models with sesimic anisotropy. The rough model, SL2016svAr, demonstrates the shorter wavelength structure capable of being captured globally, particularly in regions with elevated station density.
Slices through SL2016svA (left panels) and SL2016svAr (right panels) at six depths from the crust to the transition zone. Perturbations in peak-to-peak amplitude are indicated in percent (from the reference), with the absolute maximum and minimum and RMS amplitude indicated at the bottom left and right of each panel, respectively. The red sticks denote the orientation of azimuthal anisotropy; their length scales with the anisotropy amplitude.
These two new models were compared qualitatively and quantitatively with other recently published global 2Ψ azimuthal anisotropy models, including LH2008a (Lebedev & van der Hilst, 2008; Becker et al, 2012), YB13sv (Yuan & Beghein, 2013), 3D2015-07Sva (Debayle et al 2016), and DR2012a (Debayle & Ricard, 2013).
Horizontal slices at depths of 50, 100 and 200 km through six azimuthal anisotropy models: LH2008a (Lebedev & van der Hilst 2008; Becker et al 2012), SL2016svA, SL2016svAr, YB13sv (Yuan & Beghein 2013), 3D2015-07Sva (Debayle et al 2016) and DR2012a (Debayle & Ricard 2013). The same colour scale for peak-to-peak anisotropy in percent is used in each panel (shown at the bottom). The red sticks denote the fast propagation direction as well as the anisotropy amplitude (relative lengths). The maximum amplitude and RMS amplitude of each panel are indicated beneath it.
Spectral correlations between the two new models, SL2016svA and SL2016svAr, and LH2008a (Lebedev & van der Hilst, 2008; Becker et al, 2012). The centre panels show the correlation coefficient as a function of spherical harmonic degree and depth, with the colour key given by the colour scale at the bottom. The right sub-panels show the azimuthal anisotropy RMS amplitude, logarithmically-scaled, for the two models being compared. The left sub-panels indicate the average radial correlation functions for degrees up to and including 8 (orange) and 20 (purple); their radially averaged correlations coefficients are indicated in the titles of each frame.
Spectral correlations of SL2016svA and SL2016svAr with YB13sv (Yuan & Beghein, 2013), 3D2015-07Sva (Debayle et al, 2016), and DR2012a (Debayle & Ricard, 2013). Plotting conventions follow those of previous figure.
We have provided a single download which contains both SL2016svA and SL2016svAr. Contained within is a README file along with 2 versions of each model. The first version (SL2016svA_n-k.mod/SL2016svAr_n-k.mod) is provided laterally and vertically on the exact model knots solved for in the inversion. The second version (SL2016svA_g-k.mod/SL2016svAr_g-k.mod) is interpolated vertically to additional depths.
If using the model, please reference:
Schaeffer, A. J., S. Lebedev & T. W. Becker. Azimuthal seismic anisotropy in the Earth’s upper mantle and the thickness of tectonic plates. Geophys. J. Int., 207, pp 901-933, 2016. doi:10.1093/gji/ggw309 [Link]
Version 1.1 (October 2016):
Andrew J. Schaeffer 