MUTUALLY CONSISTENT ESTIMATES OF UPPER-MANTLE COMPOSITION FROM SEISMIC VELOCITY CONTRASTS AT 400 KM DEPTH

Seismologically determined properties of the 400 km discontinuity may be compared to experimentally determined properties of the associated phase transformation in order to place constraints upon upper mantle bulk composition. Disagreement among previous studies is commonly ascribed to differences in elastic equations of state (especially to assumptions about pressure and temperature derivatives) between studies. However, much of the disparity between studies is actually due to the selection of different seismic data functionals (P-wave velocity, S-wave velocity, etc.) for comparison to mineral elasticity calculations, rather than to the differences in elasticity data sets and equations of state. Within any given study, bulk sound velocity comparisons generally yield more olivine-rich compositional estimates than do P-wave velocity comparisons, which in tum indicate more olivine than S-wave velocities. Indeed, such variation in compositional estimates within a given study (arising from choice of data functional) exceeds the variation between studies (arising from elastic equation of state approximations). It can be argued that bulk sound velocities are better constrained seismologically than densities and, being independent of assumptions about shear moduli, should provide more reliable compositional estimates than P- or S-wave velocities. Using recently measured bulk and shear moduli equations of state, mutually consistent estimates of upper mantle olivine content can be obtained from P-wave, S-wave, and bulk sound velocity contrasts at 400 km only if partial derivative ln mu/partial derivative T of beta has a value of about -2 x 10(-4) K-1, yielding approximately 52% olivine by volume. A value of partial derivative ln mu/partial derivative T smaller in magnitude would require reassessment of several underlying assumptions.