| The lithosphere and the D"-region are two boundary layers of mantle convection, which are responsible for controlling the rate and spatial distribution of heat transport in the Earth. My objective is to increase the understanding of the mechanisms of mantle convective motions associated with plate tectonics to shape the complex structure of the D" region, and thereby improve the estimate of the magnitude and variation of heat flux across the CMB.;There is broad agreement that plate tectonics is the expression of mantle convection, however the physics of plate tectonics is still poorly understood. At the subduction zones, slabs of lithosphere undergo a bending deformation that resists tectonic plate motions, although the magnitude of this resistance is not known because of poor constraints on slab strength. However, because slab bending slows the relatively rapid motions of oceanic plates, observed plate motions can be used to constrain the importance of bending. I have estimated the slab pull force and the bending resistance globally for 207 subduction zone transects using new measurements of the bending curvature determined from slab seismicity. Predicting plate motions using a global mantle flow model, I then constrain the viscosity of the bending slab to be at most ∼ 300 times more viscous than the upper mantle. I find that stronger slabs would be intolerably slowed by the bending deformation. Weaker slabs, however, cannot transmit a pull force sufficient to explain rapid trenchward plate motions, unless slabs stretch faster than seismically observed rates of ∼ 10 -15 s-1. The constrained bending viscosity (∼ 2 x 1023 Pa s) is larger than previous estimates that yielded similar or larger bending resistance (here ∼ 25% of forces). This apparent discrepancy occurs because slabs bend more gently than previously thought, with an average radius of curvature of 390 km that permits subduction of strong slabs. This ability of slabs to undergo gentle bending may be essential for plate tectonics on Earth.;The D"-region above the core-mantle boundary (CMB) plays critical roles in the dynamics of both the mantle and the core, however, the complexity of this region observed over a broad range of spatial scales defies simple interpretations as either purely thermal or purely chemical heterogeneity. I formulate a one-dimensional, time-dependent boundary layer model for the D"-region, which provides statistical properties of the dynamics and seismic heterogeneity, by coupling thermal, chemical, and phase (TCP) variability in the layer. I assume a Gaussian-like time-dependent mantle flow, compositional stratification due to variations in iron content, heat flow variations due to changes in the local temperature gradient, and a post-perovskite phase transformation. I then compute a range of TCP boundary layer model cases that fit the observed seismic shear wave velocity heterogeneity from 50 km to 300 km above the CMB and find that they are consistent with an average CMB heat flow near 13 TW with +/- 3 TW variations, CMB temperature of 4000 K, a large positive post-perovskite Clapeyron slope, and an average heat transport of about 3 TW associated with deep mantle plumes.;In order to simulate complex time-dependent mantle convection flow, I perform a series of 3-D numerical calculations involving non-Newtonian rheology, solid-state phase transitions, and multi-composition as well as imposing historical plate motion back to 120 Ma as surface boundary condition. Synthetic seismic tomography is created from the resultant geodynamical temperature and composition fields, which I then compare to the global seismic tomography models SAW642AN, TX2008, PRI-S05 and S20RTS, statistically and visually. The best-fitting models match the seismic tomography in terms of their RMS variation, Gaussian-like frequency distribution, and spherical harmonic degree-2 pattern for large low velocity regions. Mantle heat flow at the CMB and at the surface are about 13.1 ∼ 14.7 TW and 31 TW, respectively. The Urey ratio is in reasonable range of 0.36 ∼ 0.58. Mantle convection constrained by seismic tomography, can explain the thermal and chemical structure in the mantle D"-region, as well as the average magnitude and the spatio-temporal variation of heat transport across CMB. |
