Data-oriented strategy and Vp/Vs model-constraint for simultaneous Vp and Vs reconstruction in 3D viscoelastic FWI: Application to the SEAM II Foothill dataset

Abstract

Full waveform inversion (FWI) of onshore targets is 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 timedomain spectral-element-based approach for elastic wavefield simulation in foothill areas. The challenges of the elastic multiparameter FWI in complex land areas are highlighted through the inversion of the pseudo-2D dip-line survey of the SEAM Phase II Foothill dataset. As the data is dominated by surface waves, it is mainly sensitive to the S-wave velocity. We then propose a two-steps data-windowing hierarchy to simultaneously invert for Pand S-wave speeds, focusing on early body waves before considering the whole data. By doing so, we aim at exploiting the maximum amount of information in the observed data and getting a reliable model parameters estimation, both in the near-surface and in deeper part. The model constraint that we introduce on the ratio of compressional and shear velocities also plays an important role to mitigate the ill-posedness of the inversion process.