The combination of uncertainties on elastic and anelastic properties of mantle minerals together with long period seismic data and observed global attenuation measurements provide an important constraint on the nature of the upper mantle above 400km depth.
In a previous paper, we found that adiabatic pyrolite is not compatible with seismic observations. One of the most striking features that is not reconcilable with such simple structure is the high gradient we found globally below 250km (Cammarano and Romanowicz, 2007). Here we refine our interpretation by adding testing of predicted structures against observations and investigating the effects of two mantle thermoelastic models.
We found that a purely thermal interpretation would be possible only
for low values of activation volume in order to be compatible with the measurements and would imply a cold UM () and a negative thermal gradient below 250km, on average, at the global scale.
A compositional explanation, more dynamically feasible in our opinion, would be more consistent with an adiabatic (potential T of 1300) thermal structure and predicts a significant enrichment in garnet component with depth. By using the Stixrude 2007 model, we estimate a doubling of MORB component with respect to pyrolite around 350km depth. Test on the dynamical evolution of this C or T structure and other studies on 3-D Q structure and including constrain from density are required to clearly discriminate between the two possibilities.