Novel anisotropic teleseismic body-wave tomography code AniTomo to illuminate heterogeneous anisotropic upper mantle: Part I - Theory and inversion tuning with realistic synthetic data

Abstract

Considering only isotropic wave propagation and neglecting anisotropy in teleseismic tomography studies is a simplification obviously incongruous with current understanding of the mantle-lithosphere plate dynamics. Therefore, we have developed a code for anisotropic- teleseismic tomography (AniTomo), which allows to invert relative traveltime residuals of teleseismic P waves simultaneously for coupled anisotropic-isotropic P-wave velocity models of the upper mantle. Due to a more complex anisotropic propagation of S waves, the AniTomo is applicable only to P-wave data. Weak hexagonal anisotropy together with isotropic velocity heterogeneities are interpreted as a cause of the observed P-wave traveltime residuals. Moreover, the axis of the hexagonal symmetry can be oriented freely in all directions, which represents a unique approach among recent approaches that usually incorporate only azimuthal or radial anisotropy into the body-wave tomography. Apart from outlining the theoretical background of AniTomo, we examine various aspects coming along with anisotropic tomography such as choice of a set of initial anisotropic models and setup of parameters controlling the inversion. Synthetic testing furthermore allows investigation of the well-known trade-off between effects of P-wave anisotropy and lateral variations of isotropic velocity. The target synthetic models are designed to schematically represent different heterogeneous anisotropic structures of the upper mantle. Considering realistic distributions of stations and events at teleseismic distances, a separation of seismic anisotropy and isotropic velocity heterogeneities is plausible and a stable output model can be achieved within a few iterations. Careful testing of the new code on synthetics, concentrating on its functionality, strength and weaknesses, is a necessary step before AniTomo is applied to real data sets.