Seismological forward and inverse modelling for upper mantle seismic anisotropy studies

Seismic anisotropy is the dependence of seismic wave velocity on the propagation direction and it is mainly generated by strain-induced lattice preferred orientation (LPO) of intrinsically anisotropic minerals. Despite previous studies have demonstrated that neglecting anisotropy introduces notable imaging artifacts, most tomographic methods rely on the assumption of isotropy, interpreting fast and slow velocity anomalies as related to seismically isotropic sources (e.g., temperature anomalies, presence of a liquid phase, etc). In this Thesis we carried out numerical simulations aiming at improving strain-induced fabric estimates and predicting realistic elastic properties in 2-D and 3-D synthetic domains. We generated synthetic datasets with forward waveform modelling and explored different inverse methodologies (e.g., P- and S-wave travel time tomography, automatic partitioned waveform inversion of surface waves) both with real and synthetic data. Among the results, we present ani-NEWTON21, the first 3D anisotropic teleseismic P-wave tomography revealing upper mantle structures and dynamics beneath the Central Mediterranean. By performing synthetic seismic data inversions we tested how ray density, data quality and regularization (i.e., damping and smoothing factors) influence the tomographic image. Finally, from the comparison of purely isotropic and anisotropic tests, we observed that the first-order effect of including anisotropy in the inversion is to reduce the magnitude of isotropic anomalies, more significantly for low-velocity zones relative to high-velocity zones. The research activities described in this Thesis altogether provide important insights for predicting and isolating seismic anisotropy, and for obtaining more reliable and physically consistent imaging of the Earth’s internal structure.