Data-Windowing Hierarchy in Multi-Parameter Elastic FWI: 3D Synthetic Foothills Case Study
Résumé
Full waveform inversion (FWI) in onshore targets remains very challenging due to the complex free-surface-related effects and 3D geometry representation. In such areas, the seismic wavefield is dominated by highly energetic and dispersive surface waves, converted waves and back-scattering energy. We use a time-domain spectral-element-based approach for elastic wavefield simulation in complex foothill area. The structurally-based nonstationary and anisotropic Bessel smoothing filter is considered for gradient preconditioning to stabilize the inversion, and further constraint the model parameters estimation. The challenges of the elastic multi-parameter FWI in complex land areas are highlighted through a 3D subset of the SEAM Phase II Foothills benchmark. As the data is dominated by surface waves, it is mainly sensitive to the S-wave velocity. We then propose a two-step data-windowing hierarchy to simultaneously invert for P- and S-wave speeds, focusing on early body waves before considering the whole data. By doing so, we exploit the maximum amount of information in the observed data and get a reliable model parameters estimation. Surface waves can be treated as an additional source of information, to provide better constraints on both P- and S-wave velocities estimation at near-surface. Appropriate preconditioning design helps to enhance the model parameters estimation.