Constraining lowermost mantle anisotropy with body waves: a synthetic modelling study

Different mechanisms have been proposed as explanations for seismic anisotropy at the base of the mantle, including crystallographic preferred orientation of various minerals (bridgmanite, post-perovskite and ferropericlase) and shape preferred orientation of elastically distinct materials such as partial melt. Investigations of the mechanism for D '' anisotropy usually yield ambiguous results, as seismic observations rarely (if ever) uniquely constrain a mechanism or orientation and usually rely on significant assumptions to infer flow patterns in the deep mantle. Observations of shear wave splitting and polarities of SdS and PdP reflections off the D '' discontinuity are among our best tools for probing D '' anisotropy; however, currently available data sets cannot constrain one unique scenario among those suggested by the mineral physics literature. In this work, we determine via a forward modelling approach what combinations of body wave phases (e.g. SKS, SKKS and ScS) are required to uniquely constrain a mechanism for D '' anisotropy. We test nine models based on single-crystal and polycrystalline elastic tensors provided by mineral physics studies. Our modelling predicts fast shear wave splitting directions for SKS, SKKS and ScS phases, as well as polarities of P- and S-wave reflections off the D '' interface, for a range of propagation directions, via solution of the Christoffel equation. We run tests using randomly selected synthetic data sets based on a given starting model, controlling the total number of measurements, the azimuthal distribution, and the type of seismic phases. For each synthetic data set, we search over all possible elastic tensors and orientations to determine which are consistent with the synthetic data. Overall, we find it difficult to uniquely constrain the mechanism for anisotropy with a typical number of seismic anisotropy measurements (based on currently available studies) with only one measurement technique (SKS, SKKS, ScS or reflection polarities). However, data sets that include SKS, SKKS and ScS measurements or a combination of shear wave splitting and reflection polarity measurements increase the probability of uniquely constraining the starting model and its orientation. Based on these findings, we identify specific regions (i.e. North America, northwestern Pacific and Australia) of the lowermost mantle with sufficient ray path coverage for a combination of measurement techniques.