Paper submitted to SEG 2nd Advances in Geophysical Solutions for Complex Geological Settings Workshop, by Nizar Chemingui, Raafat Abdul Alim, David Cavalin, Chaoshun Hu and Khaled Abdelaziz (TGS)
Summary
Seismic inversion is fundamentally a multi-scale and multi-parameter problem. The subsurface may be viewed as a smoothly varying background model that controls wave propagation, onto which are superimposed shorter-wavelength perturbations associated with impedance contrasts and seismic scattering. In that sense, a reliable long-wavelength background model is essential for wavefield propagation, and structural positioning, while the estimation of higher-wavenumber perturbations is equally critical for recovering impedance variations, density changes, and the reservoir-scale signatures. Our previous work showed that separating these scales within the inversion is key to minimizing cross-talk between parameters and to enabling joint updates of velocity and reflectivity within a single wave-equation-based Full Waveform Inversion (FWI). In our formulation reflectivity is expressed through an impedance-gradient that describes the backscattered energy and carries the high-wavenumber information linked to geological boundaries and reservoir heterogeneity. By using inverse-scattering-based kernels and a dedicated scaleseparation strategy, the inversion can update the macro-model and the perturbation model in a more stable and geologically meaningful manner.