P-wave polarization data may provide constraints on subsurface structure and anisotropy different from those of traveltimes and shear-wave splitting and are thus very important: Anisotropy in the near-receiver region gives rise to characteristic deviations of polarization from the wavefront normal direction. While this effect sometimes represents a nuisance, in our application to seismological array data it is used to constrain anisotropy in the upper few kilometers on the receiver-side. Polarization data are, however, influenced by a number of effects arising from the near-receiver region, such as heterogeneity and/or anisotropy as well as effects from conversions at low-velocity layers and the free surface. Before interpreting polarization data, they must be examined for signatures of undesired local effects, in particular from conversions.
P-wave polarization deviations axe in fact quite widespread (see references in Bokelmann, 1995a), but reported only occasionally. Such data will very likely play a larger role in the future, e.g. in inversions for subsurface anisotropy. This paper gives an overview of necessary considerations for attributing observed effects to distant heterogeneity /anisotropy or to local anisotropy and to distinguish from alternative effects. In particular, scattering from near-receiver heterogeneity may render polarization observations from single three-component stations useless: Spatial averaging is in most cases necessary to produce average polarizations, which are then very valuable.