Radial 1-D seismic structures in the deep mantle in mantle convection simulations with self-consistently calculated mineralogy

Numerical thermo-chemical mantle convection simulations in a spherical annulus geometry with self-consistently calculated mineralogy and mineral physics are used to predict detailed deep mantle seismic structures, particularly local radial profiles of shear wave velocity (Vs) and bulk sound velocity (Vb). The predicted structures are compared to seismological observations and used to guide the interpretation of seismic observations and to test the model. The mantle composition is described as a mixture of MORB (Mid-Oceanic-Ridge-Basalt) and harzburgitic end-members in the Na2O-CaO-FeO-MgO-Al2O3-SiO2 system. To assess the influence of chemical variability, four different sets of end-member compositions are evaluated. Results confirm that the post-perovskite (pPv) phase causes anti-correlated S wave and bulk sound velocities in the deep mantle, due to pPv being fast in Vs but slow in bulk sound velocity. Local 1-D seismic profiles display great lateral variability, and often have multiple discontinuities in the deep mantle due to MORB layers in folded slabs, with a positive Vs anomaly and negative bulk sound anomaly, or the perovskite-pPv phase transition. The pPv transition is not visible inside piles of segregated MORB because of the high temperature and small velocity contrast of pPv in MORB. Piles of segregated MORB are seismically slow in both Vs and bulk sound despite being expected to be fast in Vs, because they are hotter than the surrounding material. Anelasticity has a significant influence on profiles of Vs only in the lower thermal boundary layer, which corresponds to below 2600 to 2800 km depth depending on region, where temperatures are higher than the extrapolated adiabat. These results indicate the importance of using a joint geodynamical-mineralogical approach to predict and aid in the interpretation of deep mantle seismic structure, because interpretations based on seismology and mineral physics alone may be misleading and do not capture the strong lateral variability in 1-D structure obtained here: for example, multiple reflections arising from folded slabs and the precise balance between thermal and compositional influences on seismic structure.