Adjoint Tomography of South America based on 3D Spectral-Element Seismic Wave Simulations

• Ano de defesa: 2021
• Autor(a) principal: Ciardelli, Caio Henrique
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: adjoint method; B-splines; elementos espectrais; equação de onda; finite-frequency theory; full-waveform inversion; harmônicos esféricos.; inversão de forma de onda completa; método adjunto; numerical optimization; otimização numérica; spectral elements; spherical harmonics.; teoria da frequência finita; visualização; visualization; wave equation
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/14/14132/tde-11092021-134206/

Adjoint tomography, a full-waveform inversion technique based on 3D wave simulations, is now commonly used in earthquake seismology, drawing on advances in computational power and numerical methods. In this study, we use 3D spectral element continental-scale seismic wave simulations (Komatitsch and Tromp, 2002a,b) and 112 earthquakes recorded by 1311 seismic stations to construct an adjoint waveform tomography model of South America. The thesis begins with a review of the wave equation in elastodynamics followed by an introductory explanation of the spectral-element method (Schubert, 2003; Igel, 2017). We also revisit the inversion problem in geophysics and the adjoint-state method (Plessix, 2006) in an intuitive way. We proceed with a simplified explanation of the finite-frequency theory (Dahlen et al., 2000) and a review of previous tomographic studies in South America. To carry out our adjoint tomography, we detect and remove noisy & problematic data using our multi-stage algorithm before the time-window selection, reducing the likelihood of discarding useful data or assimilating bad-quality waveforms in inversions. Our misfit function is a complex-exponentiated instantaneous phase (Yuan et al., 2020), which optimizes the information extracted from each time series without the need for short-time windows. We performed 23 iterations, gradually increasing the frequency content of the data to prevent local minima from hampering the convergence. Our final model (SAAM23, South American Adjoint Model, iteration 23) shows a 50% decrease in the misfit. We further assessed the improvement by using cross-correlation measurements using 53 earthquakes that were not included in the adjoint inversion. In the long wavelengths, the model is compatible with previous studies, such as Van der Lee et al. (2001), Feng et al. (2007), Celli et al. (2020), and Lei et al. (2020). The Nazca Slab is well imaged and is shown to be continuous in the 300-500 km depth following the Peruvian flat-slab segment. Beneath northern South America, the slab crosses the mantle transition zone and plunges into the lower mantle. In the central and southern part of South America, the slab appear to flatten near the 650 km discontinuity, before plunging into the lower mantle. In the stable platform, both exposed cratons (Amazonian and São Francisco), as well as covered cratonic blocks (Paranapanema and Parnaíba, beneath the intracratonic (Paraná and Parnaíba Basins, respectively), show high velocities at lithospheric depths. The seismic lithosphere/asthenosphere boundary (LAB) was measured at the depth with the steepest negative velocity gradient. Good agreement was found between this seismic LAB and the values obtained by S-wave receiver functions. In the Amazonian Craton, both positive lithospheric S-wave velocity anomalies, as well as LAB depth, increase with the average age of the geochronological provinces. On the other hand, no high-velocity anomalies were found beneath the Rio de La Plata Craton. The thesis ends with the presentation of SphGLLTools, an open-source toolbox we designed to allow easy and practical visualization of tomographic models defined on spectral-element meshes using either direct interpolation or the flexible expansion using spherical harmonics while taking advantage of GMT6 (Wessel et al., 2019) to create high-quality images. We also lead the reader through a comprehensible yet intuitive explanation of the theory and concepts used by the routines.