The vast majority of teleseismic XKS (including SKS, SKKS and PKS) shear wave splitting studies interpret the observed splitting parameters (fast orientation and splitting time) based on the assumption of a spatially invariant anisotropy structure in the vicinity of a recording station. For such anisotropy structures the observed splitting parameters are either independent of the arriving azimuth of the seismic ray paths if the medium traversed by the ray paths can be represented by a single layer of anisotropy with a horizontal axis of symmetry (i.e. simple anisotropy), or demonstrate a periodic variation with respect to the arriving azimuth for a more complicated structure of anisotropy (e.g. multiple layers with a horizontal axis of symmetry, or a single layer with a dipping axis). When a recording station is located near the boundary of two or more regions with different anisotropy characteristics, the observed splitting parameters are dependent on the location of the ray piercing points. Such a piercing-point dependence is clearly observed using a total of 360 pairs of XKS splitting parameters at three stations situated near the northeastern edge of the Sichuan Basin in central China. For a given station, the fast orientations differ as much as 90°, and the azimuthal variation of the fast orientations lacks a 90° or 180° periodicity which is expected for double-layered or dipping axis anisotropy. The observed splitting parameters from the three stations are spatially most consistent when they are projected at a depth of ~250 km, and can be explained by shear strain associated with the absolute plate motion and mantle flow deflected by the cone-shaped lithospheric root of the Sichuan Basin.