Central-difference time-lapse 4D seismic full waveform inversion

Full waveform inversion (FWI) can be applied to repeated seismic surveys to estimate time-lapse (4D) changes in the subsurface caused by natural or man-made processes. However, reliable estimates are difficult to obtain when the changes are small or the data are contaminated by noise. We propose a two-step “central difference” method by which the forward and reverse bootstrap estimates are averaged. A test example is shown on the SEAM time-lapse pilot project. Although it can recover more 4D changes than parallel FWI, the forward bootstrap FWI method is prone to artifacts preventing reliable 4D estimates. Interestingly, the reverse bootstrap implementation (i.e. monitor inversion before baseline inversion) can regenerate these forward bootstrap artifacts with a phase reversal; thus, the central-difference summation can attenuate such artifacts. The proposed method provides more accurate time-lapse velocity model estimates with reasonable computational costs.