Is anelasticity hiding chemical variations in the lowermost mantle?

Ab initio calculations have now provided us with estimates of almost all the necessary thermo-elastic parameters with which to evaluate whether chemistry, temperature or combinations of both are responsible for large local variations in shear velocites observed within the lower mantle and D". For instance the ab intio estimates for the high temperature and pressure elastic constants are known for the Mg and Al endmembers of perovskite and post-perovskite. In addition the zero Kelvin elastic constants are known for the ferrous and ferric bearing endmembers of both phases, and include the effect of the high- and low-spin states. At present, the high-temperature elasticity of the Fe-bearing phases is still not known, and currently one has to assume that the temperature dependence is similar to the Mg end-member. Taken at face value, lateral variations in shear velocities of about 3% can be explained by temperatures variations of about 750 K in the lower mantle and perhaps 1000 K in D".Since these large variations in temperature seem unlikely over short distances, the favoured explanation is that they have a chemical as well as thermal origin. For instance, similar variations in velocity can be obtained from about a 12% enrichment Fe in the perovskite phase or 15% in postperovskite. Since we have ab initio estimates of density as a function of composition and temperature it is also possible to calculate how a combination of temperature and composition affects the density. For instance perovskite with about 4% Fe enrichment and is 500 K hotter than ambient, would satisfy a 3% shear velocity increase while remaining neutrally buoyant. However, analyses like this ignore anelasticity, and as Karato (1993) and Karato and Karki (2001) have shown, anelasticity can significantly change parameters such as dlogVs/dT and dlogρ/dlogVs, perhaps by as much as 100%. This would mean that only half of the temperature contrast would be needed to produce the same velocity contrast, and therefore, less Fe enrichment in areas which are hot and Fe enriched. This motivates our re-examination of the issue to see whether we can place bounds on the likely effect that anelasticity has on derivatives like dlogVs/dT for perovskite and post-perovskite, and, therefore whether the ab initio results can be used to constrain the relative contributions of temperature and composition to large seismic anomalies.