Source: Dissertation Abstracts International, Volume: 77-06(E), Section: B.

Adviser: Maureen D. Long.

Subduction systems are vitally important to plate tectonics and mantle convection, but questions remain about many aspects of subduction dynamics, particularly the nature of sub-slab mantle flow. Observations of seismic anisotropy can shed light on the pattern of mantle flow in subduction systems. Understanding the sub-slab mantle is the focus of my first three chapters as well as my last chapter.

In Chapters 1 and 2, I examine the sub-slab anisotropy beneath the Caribbean, Scotia, Central America, Alaska-Aleutians, Sumatra, Ryukyu, and IzuBonin-Japan-Kurile subduction systems. I find that measured fast splitting directions in these regions generally fall into two broad categories, aligning either with the strike of the trench or with the motion of the subducting slab relative to the overriding plate. In theses systems, there is a correlation between fast direction and age of the subducting lithosphere; older lithosphere (> 95 Ma) is associated with trench parallel splitting while younger lithosphere (< 95 Ma) is associated with plate-motion parallel fast splitting directions.

In Chapter 3, I compile measurements from recent studies of source-side splitting studies to test the predictions of a number of recently proposed conceptual models for the dynamics of the sub-slab mantle. I find that a model in which fast splitting directions are determined by slab age matches the observations better than either the 3D-return flow or radial anisotropic models. Based on this observation, I propose that the sub-slab mantle is characterized by two distinct anisotropic and mantle flow regimes. Beneath younger lithosphere (< 95 Ma), I propose the sub-slab mantle is characterized by 2-D entrained flow resulting in an entrained mantle layer. Beneath older lithosphere (> 95 Ma), the entrained layer is thin and effectively serves as decoupling layer; the dynamics of the sub-slab region beneath old lithosphere is therefore dominated by three-dimensional return flow.

In Chapters 4 and 5, I focus on the dynamics of lowermost mantle. Shear wave splitting of SK(K)S phases is often used to examine upper mantle anisotropy. In specific cases, however, splitting of these phases may reflect anisotropy in the lowermost mantle. In both Chapters 4 and 5, I present splitting measurements of SK(K)S phases that sample the lowermost mantle beneath Africa.

In Chapter 4, I present measurements of SKS and SKKS splitting at station DBIC in the Cote D'Ivoire. The splitting pattern is dominated by null measurements over a wide range of backazimuths, with non-null measurements found over a very limited backazimuthal range. Splitting at DBIC has previously been interpreted in terms of upper mantle anisotropy, but we argue that an apparently isotropic upper mantle can best explain this splitting pattern with a contribution from anisotropy in the lowermost mantle.

In Chapter 5, I present SKS and SKKS splitting measurements that likely reflects a contribution from lowermost mantle anisotropy beneath Africa. The vast majority of discrepant pairs sample the boundary of the African large low shear-wave velocity province (LLSVP), which dominates the lower mantle structure beneath this region. In general, I observe little or no splitting of phases that have passed through the LLSVP itself and significant splitting for phases that have sampled the boundary of the LLSVP. I infer that the D" region just outside the LLSVP boundary is strongly deformed, while its interior remains undeformed (or weakly deformed).

In Chapter 6, I examine the anisotropic structure of the mid-mantle (transition zone and uppermost lower mantle) beneath the Japan, Izu-Bonin, and South America subduction systems. In each region, I observe consistent splitting with delay times as large as 1 sec, indicating the presence of anisotropy at mid-mantle depths. Clear splitting of phases originating from depths as great as ~600km argues for a contribution from anisotropy in the uppermost lower mantle as well as the transition zone.

The goal of Chapter 7 is to evaluate predicted sub-slab splitting from 3D geodynamic models using a variety of different anisotropic fabric types. This builds on Chapter 3; in which only very simplified sub-slab dynamics and approximated LPO fabrics were used. Using realistic geodynamic models for Tonga and Central America, which respectively show 3D return flow and entrained flow dynamics, I predict splitting in the sub-slab mantle for several LPO fabrics. While splitting due to the hexagonal approximation olivine LPO fabric does a fair job matching the observations, it is not the best fitting fabric type in either modeled subduction zone. In Tonga, E-type fabric produced splitting that best matches observations, and C-type fabric is the best fit in Central America.