Separating intrinsic and apparent anisotropy

Abstract

Seismic anisotropy plays a key role in studies of the Earth's rheology and deformation because of its relation to flow-induced lattice-preferred orientation (LPO) of intrinsically anisotropic minerals. In addition to LPO, small-scale heterogeneity produces apparent anisotropy that need not be related to deformation in the same way as intrinsic anisotropy. Quantitative interpretations of observed anisotropy therefore require the separation of its intrinsic and apparent components.We analyse the possibility to separate intrinsic and apparent anisotropy in media with hexagonal symmetry - typically used in surface wave tomography and SKS splitting studies. Our analysis is on the level of the wave equation, which makes it general and independent of specific data types or tomographic techniques.We find that observed anisotropy can be explained by isotropic heterogeneity when elastic parameters take specific combinations of values. In practice, the uncertainties of inferred anisotropy are large enough to ensure that such a combination is always within the error bars. It follows that commonly observed anisotropy can always be explained completely by a purely isotropic laminated medium unless all anisotropic parameters are known with unrealistic accuracy. Most importantly, minute changes in the poorly constrained P wave anisotropy and the parameter η can switch between the possible or impossible existence of an isotropic equivalent.Important implications of our study include: (1) Intrinsic anisotropy over tomographically resolved length scales is never strictly required when reasonable error bars for anisotropic parameters are taken into account. (2) Currently available seismic observables provide weak constraints on the relative contributions of intrinsic and apparent anisotropy. (3) Therefore, seismic observables alone are not sufficient to constrain the magnitude of mantle flow. (4) Quantitative interpretations of anisotropy in terms of mantle flow require combined seismic/geodynamic inversions, as well as the incorporation of additional data such as topography, gravity and scattered waves.