A Statistical Boundary Layer Model for the Mantle D"-Region

The D"-region above the core-mantle boundary (CMB) plays critical roles in the dynamics of both the mantle and the core. However, its observed complexity over a broad range of spatial scales has defied simple interpretations in terms of purely thermal or compositional mantle heterogeneity. We have extended a statistical boundary layer model, originally proposed by Howard (1964), to include (1) arbitrary time dependent mantle flow, (2) compositional stratification due to variations in iron content, and (3) the post-perovskite phase transformation, which gives time and spatial average structural and dynamical properties in the D"-region. The primary observational constraints we use include profiles of seismic shear wave and bulk sound wave velocity, variance, and skewness, and their correlation, the depth of seismic reflectors interpreted as phase transitions, and hotspot-derived estimates of the lower mantle plume flux. We find that purely thermal or compositional boundary layer models do not satisfy these constraints, nor do models with a thermal boundary layer superimposed on a chemical layer with uniform composition. Instead, a spatially heterogeneous dense layer with some internal chemical stratification is preferred. In addition, the temperature may contribute to seismic shear-wave tomography indirectly, via depth variation in the post-perovskite (pPv) phase lens. Subject to uncertainties in the lower mantle and core adiabats, best-fitting models predict total CMB heat flow of 15 TW with ± 2 TW lateral variation, for the commonly-used lowermost mantle thermal conductivity of 10 Wm-1K-1. Reference: Howard, L. N. 1964. Convection at high Rayleigh numbers. In Applied Mechanics, Proc. 11th Congress of Applied Mathematics, ed. H. Görtler, 1109-1115.