Résumé de la publication

The shear-wave velocity is one of the parameters of interest for the geotechnical design of wind turbine foundations. Near-surface seismic parameters are generally inferred using surface wave analysis, i.e., effective Rayleigh dispersion curve inversion. However, in offshore context near southern Bretagne (France), where geologic structures are complex, the planar stratified media (1D) hypothesis can not be used, and recently developed full waveform inversion declinations for near-surface imaging are difficult to implement at the scale of civil engineering (number of shots, receivers, almost perfect knowledge of the source waveform and repeatability). In this context, this work aims to develop an alternative two-dimensional imaging approach based on particle swarm optimization and a sparse spatial discretization of the medium to obtain smooth parameter models. Exploiting the complete dispersion diagrams instead of dispersion curve only, the preliminary results show that swarm intelligence based algorithm allows to assess physical parameters (S-wave velocity) and lateral variation.