Mantle structure and anisotropy from the particle motion and slowness of compressional body waves

I present two new methods to obtain information on Earth structure from P wave polarization in its various flavours. The two observables related to polarization used here are particle motion direction measured at three-component seismic stations, and slowness direction determined from seismic array data. Both show deviations from the source-receiver great circle suggesting lateral structure or azimuthal anisotropy.; The first part of this dissertation is a global study of the particle motion direction of teleseismic P arrivals, which exhibit anomalies of up to ∼15°, an order of magnitude larger than those predicted by current high-resolution isotropic mantle models. The polarization data also show no significant variance reduction in joint inversions with other data for isotropic structure. However, they can be explained with realistic amounts of anisotropy in the uppermost mantle and crust. The global station set gives fast directions of anisotropy that agree with those derived from P_n in most regions, and fast directions of SKS in many regions.; The second part expands further on the comparison between fast directions measured by the P polarization method versus results from shear wave splitting and azimuthal travel time variation of P and P_n. Of particular interest are reports that compressional and shear phases may see contradictory fast directions of anisotropy. I calculate results for all phases using a series of homogeneous and layered models of anisotropy with progressively more complex symmetry and orientation. While true P-S discrepancies do not occur, several assumptions in previous studies can be identified that may have led to an apparent contradiction. The depth sensitivity decreases progressively from P delays to SKS splitting, P polarization, and P_n.; Last, I investigate the second P-wave observable, the direction of the wave vector or slowness. Data from the Anza array, situated in a region with active tectonics in southern California, reveal large and strongly frequency dependent azimuthal anomalies which are antisymmetric about the array, whereas much weaker effects are seen at the Norsar (Norway) and Gräfenberg (Germany) arrays in stable regions. A probable cause for the effect at Anza is lateral velocity contrasts that change sign with depth.