Flow-induced anisotropy in D": comparing finite-frequency synthetic ScS waveforms with observations

A commonly-made assumption is that the observed anisotropy in D" is caused by the alignment of MgSiO₃ post-perovskite (ppv). However, there are many other potential causes such as the alignment of other mineral phases or of seismically distinct material in a shape-preferred orientation (SPO), and as yet no observations have been able to distinguish these cases. We present initial results of finite-frequency full-waveform modelling of ScS waves traversing a generally anisotropic, heterogeneous lowermost mantle. The input elasticity is derived from a model of texture development in ppv, driven by flow in D". We compare these synthetics with regional, multi-azimuth observations of shear wave splitting in ScS made along the same paths. Synthetics are created down to periods similar to the observations (~10 s), requiring considerable computational expense. Where the observed and modelled splitting agree or disagree, ppv LPO can be considered a compatible or incompatible mechanism, within the assumptions and approximations we make. We have developed an integrated model of mantle flow and the resulting LPO of ppv, derived from seismic and mineral physics observations. Particles are traced through a steady-state flow field, driving a visco-plastic self-consistent model of texture development in ppv. Our model produces fully anisotropic elastic constants which can be used for comparison with global and regional seismic observations of anisotropy in D". Comparisons with global inversions for vertical transverse isotropy, a common assumption, can be made, but in order to accurately account for azimuthal sampling bias, we require detailed knowledge of each raypath used in the inversion. An alternative approach we pursue is to compare individual source-receiver measurements. This has the advantage that there are no constraints on the type of anisotropy we examine. By far the largest difference between the models we test is the deformation mechanism of ppv, something which is still largely unknown. We test three models of dislocation creep in ppv, based on theoretical and experiment considerations, with slip mainly on (100), (010) or (001). No single deformation mechanism can reproduce all the observations suggesting one or more of several things: ppv LPO may not be the dominant mechanism causing the anisotropy; ppv may have several deformation mechanisms depending on the conditions; there may be several causes of D" anisotropy, and ppv LPO may be merely one of them; some of our assumptions or approximations are not sufficiently accurate. As further constraints are placed on mantle flow, and especially the deformation mechanisms of lowermost mantle phases, the number of possibilities will be reduced, and we will move closer to being able to constrain mantle flow directly with measurements and models of D" anisotropy.