Systematic Measurements of Splitting Parameters of P-to-S Converted Phases from Mantle Discontinuities

Due to the near-vertical incidence of XKS (SKS, SKKS, and PKS etc.) phases, the techniques that measure XKS splitting parameters have an excellent horizontal resolution, but have a very poor vertical resolution. In principle, the anisotropy responsible for the observed splitting could exist anywhere from the CMB to the surface. Most previous studies assume that the anisotropy is mostly located in the upper mantle in the subcrustal lithosphere, the asthenosphere, or both. Some studies, however, found anisotropy in the crust, mantle transition zone, or lower mantle. The depth of anisotropy can be constrained by splitting of P-to-S converted phases (PdS) from velocity discontinuities at different depths in the mantle. We have developed a semi-automatic procedure to measure PdS splitting parameters by stacking hundreds of source-normalized seismograms at a station. The procedure include the following steps: 1). Computing radial and transverse receiver functions (RFs) by deconvolution for all the available seismograms with strong P-arrivals; 2). RFs from a narrow (10-30 degrees, depending on the event population) back-azimuthal range are stacked based on the predicted moveout time of PdS at a range of candidate depths; 3). The results of the stacking (amplitude versus depth) are converted into amplitude versus time; 4). The stacked radial and transverse components, together with the mean backazimuth of the events participated in the stacking, are used to obtain optimal splitting parameters at various depths. By comparing the results with those from XKS, the contributions of the major layers (e.g., lower mantle, mantle transition, and upper mantle) can be separated.