Shot-domain 4D time-lapse seismic velocity analysis using apparent image displacements

Hydrocarbon production modifies the stress conditions in the subsurface and changes the model parameters previously estimated from the prospect. The capability to remotely monitor the changes in the reservoir using seismic data has strategic importance since it allows us to infer fluid movement and evolution of stress conditions, which are key factors to enhance recovery and reduce uncertainty and risk during production. A model of the subsurface parameters is necessary to reconstruct the seismic waves traveling through the medium and thus correctly image reflectors in the subsurface. Geomechanical changes in the subsurface can be measured by changes of seismic images obtained from the recorded data of multiple time-lapse surveys. In this work, we estimate changes of subsurface model parameters using the apparent shifts between migrated images obtained by 4D time-lapse seismic surveys. We assume that the shift between the time-lapse images of the same reflectors is completely due to the perturbation of the model parameters, and we use the image from the first (baseline) survey as a reference to estimate this perturbation. The apparent shifts are measured using penalized local correlations in the image domain, and they are exploited using wavefield tomography with an objective function minimized using the adjoint-state method. Our time-lapse monitoring method is efficient due to the fact that inversion can be conducted for pairs of seismic experiments, which eliminates the need to construct costly gathers. Since relatively small amounts of data are needed, our method can be used to invert for model changes at short intervals, thus increasing the resolution of 4D monitoring.