Joint wave-equation inversion of time-lapse seismic data

Conclusions

We have proposed two regularized least-squares inversion formulations for time-lapse seismic imaging. -hese formulations arise from the linearized least-square wave-equation inversion. By using a target-oriented approximation to the least-squares Hessian with appropriate spatial and temporal regularization, we have shown that image difference due to geometry dissimilarity and complex overburden can be attenuated. We show that we can directly invert for image differences (RJID) or multiple images (RJMI) using a concatenation of target-oriented Hessian operators and combinations of migrated images. From numerical tests using a synthetic 2D-subsalt model, we conclude that both the RJID and RJMI joint-inversion formulations give more accurate time-lapse images than either migration or separate inversion.