Source: Dissertation Abstracts International, Volume: 75-09(E), Section: B.

Adviser: Jeffrey J. Park.

Seismic anisotropy is often attributed to the development of lattice-preferred orientation (LPO) of olivine crystals in peridotite, induced by the dislocation creep component of mantle deformation (Karato et al., 2008; Ribe, 1992). Mantle-flow-induced seismic anisotropy is often modeled in the simple form of hexagonal symmetry, where the anisotropic volume is uniaxially fast or slow. This relationship between seismic anisotropy and mantle deformation allows for the mapping of mantle dynamics using measurements of seismic anisotropy.

Presently, methods of measuring seismic anisotropy in Earth's mantle include shear-wave splitting and surface-wave tomography. These methods are tuned to seismically fast axes laying in the horizontal or surface-tangent plane and are limited in discerning clipping seismic fast axes. This is a shortcoming. It is reasonable to suspect the presence of dipping seismic fast axes induced by mantle flow in several tectonic regimes such as subduction zones.

The slab rollback model of the subduction zone system has been argued to exhibit trench-parallel subslab anisotropy due to the lateral evacuation of the subslab mantle material (Hall et al., 2000; Russo and Silver, 1994). This model has been emboldened by the dominance of trench-parallel shear-wave-splitting measurements in the subslab mantle of global subduction zones. This model has significant geodynamic implications, requiring viscous decoupling between the subslab mantle and the sub-ducting slab. The Cascadian subduction zone is of particular scientific interest. While experiencing slab rollback (Zandt and Humphreys, 2008), trench-perpendicular shear-wave-splitting measurements are observed in the subslab mantle of Cascadia (Currie et al., 2004; Eakin et al., 2010; Long and Silver, 2008; 2009). This suggests either viscous coupling resulting in slab-entrained flow or the presence of an alternate relationship between finite strain in the mantle and seismic anisotropy. The ability to discern a clipping anisotropic axis would help gain insight into the mantle dynamics of regions such as Cascadia.

Lateral gradients of seismic anisotropy in Earth's upper mantle induce coupling among Earth's spheroidal and toroidal normal modes. This coupling can manifest as observable surface-wave polarization anomalies resulting from Love to Rayleigh wave conversions. These Love to Rayleigh conversions are known in the literature as Quasi-Love (QL) waves (Park and Yu, 1992) and are sensitive to both the strike and the dip of an anisotropic symmetry axis. In this dissertation I investigate the phenomenology of QL surface-wave scattering, including its sensitivity to the type and orientation of seismic anisotropy. I then apply my findings to observations of QL wave scattering in Cascada in order to further constrain subslab mantle anisotropy in the region.

First, I make initial observations and confirm the presence of QL scattering in Cascada and the western U.S. using data recorded on USArray. I then move on to develop an algorithm to model efficiently QL wave scattering in the presence of 3-dimensional anisotropic structure.

Using this forward-modeling algorithm, I investigate the dependence of QL wave scattering on the type and orientation of seismic Anisotropy. I find that P and S anisotropies exhibit independent effects on scattering. Scattering due to S anisotropy is stronger than that due to P anisotropy for all orientations and dominates in the observed scattering pattern. Both the phase and amplitude of the QL wave is dependent on the orientation (strike and dip) of the symmetry axis relative to the incident propagation azimuth of the source-receiver great-circle path. Due to this, the orientation of the anisotropic symmetry axis provides a distinct signature which is observable in the variation of QL wave scattering with wave-propagation azimuth.

Finally, using data recorded on USArray, I observe the variation in QL wave scattering with propagation azimuth. I then attempt to forward-model the observed behavior using the algorithm developed earlier. The best-fitting model suggests coherent trench-perpendicular mantle anisotropy with an eastward dip in the sublsab mantle of the Cascadian subduction zone. The resulting anisotropic model adds confidence to the entrained subslab mantle-flow model for Cascadia and further refutes the 3-D return-flow model associated with slab rollback.