Waveform modelling of 3-D seismic anisotropy in the Earth's mantle

Considerable progress in seismic tomography in recent years has resulted in the agreement of large scale features in isotropic shear velocity models. Despite this, the mapping of more complex parameters such as radial anisotropy, is far from consistent across models owing to various factors such as the data sets, parametrisations, forward and inversion modelling schemes adopted by different research groups. Despite these challenges, this powerful probe provides crucial information to understand mantle flow in our dynamic planet. This study tests the robustness of upper mantle features in 3-D isotropic and radially anisotropic global mantle models such as S40RTS (Ritsema et al., 2011), S362WMANI+M (Moulik & Ekström, 2014), SAVANI (Auer et al., 2014) and SGLOBE-rani (Chang et al., 2015) and begins to interpret them in terms of geodynamical processes. We use the spectral element method (Komatitsch and Tromp, 1999) to compute full waveforms for over 300 paths for earthquakes with Mw between 6.4-7 and focal depths < 50km, and with stations within an epicentral distance of 40o - 140o. We include crustal thickness perturbations of up to +/-10km from CRUST2.0 in SGLOBE-rani's forward simulations, which enable us to investigate crustal effects, especially for surface waves propagating across the oceans. Synthetic waveforms are compared with real seismograms, with a special focus on fundamental mode surface waves in the period range T 25-100s. Our initial results show that SGLOBE-rani fits observed waveforms well and better than some other models. Moreover, we assess the adjustments in radial anisotropy needed to further improve the waveform fits in the Pacific. We find that overall SGLOBE- rani's anisotropic structure conforms well with the data, with errors in radial anisotropy in the order of ±10% for a large percentage of paths. In addition, preliminary results show greatly improved phase misfits when crustal thickness perturbations are taken into account in the modelling. Similar to previous studies, we observed that while the isotropic shear wave speed models portray a lithosphere-asthenosphere boundary deepening with age of the oceanic lithosphere, the relationship is less clear for radial anisotropy. Thus, we perform synthetic tests to investigate the depth resolution of radial anisotropy beneath the Pacific.