Abstract:
We have formulated dynamic models of processes hypothesized to give rise to observed seismological structure at the base of the mantle. One dimensional seismic models are determined from thermo-chemical convection calculations from which synthetic waveforms are computed. Of the three scenarios studied, (1) a thermal slab, (2) a thermal slab interacting with simple chemical layer, and (3) a thermal slab interacting with a phase transition, the strength of seismic phases computed from (3) best correlates with the global geography of the observed D triplication. Seismic structures predicted by (1) or (2) are inconsistent with the global geography and strength of the triplication. Our successful model motivates a new class of 1-D seismic models with two mild gradients and a small discontinuity. To test this hypothesis, we searched for seismic models that both contain these features and fit D seismic observations. Our preferred model contains a gentle positive gradient initiated 350 km above the CMB and a 1% jump in S-velocity near 200 km. A strong negative gradient begins about 100 km above the CMB, similar to previous studies representing the lower thermal boundary layer. With the reduced first-order discontinuity, the P-wave triplication becomes too small to observe and provides a simple explanation for the observed weakness of the P in proportion to the S triplication.