A new look at polarity information in D" reflections

Polarities of seismic reflection of P and S-waves at the discontinuity at the top of  D" are usually assumed to indicate the sign of the velocity contrast across the D" reflector. For reflections in paleo-subduction regions the S-wave reflections off D" (SdS) are the same as ScS and S, indicating a positive velocity contrast at the reflector. In recent years, an opposite polarity of PdP waves (P-reflection at the D" discontinuity) has been observed in some regions, partly dependent on travel direction, partly dependent on distance. This would indicate a velocity reduction in P-waves where a velocity increase is detected in S-waves. This phenomenon can be explained with the presence of post-perovskite below the top of D", but azimuthal dependence of PdP polarities can be better explained with anisotropy. Here we re-analyse PdP and SdS wave polarities and, when modelling the polarities and amplitudes using Zoeppritz equations, we find that a ratio of dVs/dVp= R of larger than 3 reverses polarities of P-waves in the absence of anisotropy, i.e. we find a polarity of PdP that would point to a velocity decrease while modelling a velocity increase. The S-polarity stays the same as S and ScS and does not change even with large R. Values of R up to 4.1 have been reported recently, so these cases do exist in the lower mantle. Using a set of 1 million models with varying minerals and processes across the boundary, we carry out a statistical analysis (Linear Discriminant Analysis, LDA) and find that there is a marked difference in mantle mineralogy to explain R values larger and smaller than 3, respectively. The regime of cases with R-value larger than 3 is mostly due to an increase in MgO and post-perovskite across the discontinuity. In regions where high R is observed, alternate explanations of lowermost mantle composition versus anisotropy can then be tested by measuring polarities in different azimuths.