PhD Thesis (2023-2026)

Title:  Stochastic inversion of FWI images for reducing structural uncertainties

Supervisors: Guillaume Caumon (RING Team), Paul Cupillard (RING Team)

Despite advancements in seismic imaging technologies and interpretation methodologies, resolving fine-scale geological features and quantifying the associated uncertainty remains challenging.
In particular, full waveform inversion (FWI) methods are widely used to obtain from seismic data a quantitative image of the medium elastic properties which is often interpreted in terms of geological structures.
However, because this technique is based on frequency band-limited seismic data, subsurface property variations smaller than a fraction of the shortest propagated seismic wavelength are ‘seen’ as smooth heterogeneities (Capdeville et al., 2010). As a consequence, FWI models are not suited for detection of discontinuities in the elastic properties at the small-scale. 

In this research project, we address the problem of quantifying structural uncertainty through the downscaling of homogenized FWI images. In the proposed approach, we apply the homogenization operator in the context of the elastic FWI (HFWI) to obtain the corresponding (macro-scale) effective medium (Capdeville and Métivier, 2018). As a second step, we carry out the downscaling inversion (Bodin et al., 2015, Hedjazian et al., 2021): assuming the HFWI solution represents the effective elastic properties of a true earth model, we look for fine-scale models compatible with these effective properties and some a priori knowledge.