Elastic FWI for large impedance contrasts

Recent advances in full-waveform inversion (FWI) algorithms have allowed it to work stably and effectively in different geological settings, especially in areas with geobodies of large impedance contrasts such as salt, despite its presently taken acoustic assumption. This has resulted in a leap in salt velocity model building over the conventional interpretation-driven approach, providing significantly improved salt models and a step-change in subsalt imaging. Furthermore, it produces a directly interpretable FWI Image, the normal derivative of the FWI velocities, that is now widely accepted as another step- change for subsalt imaging. However, elastic effects at the large impedance contrasts cause smearing of the salt boundary and a considerable salt halo in acoustic FWI results. To mitigate this issue, we have incorporated an elastic modeling engine into our acoustic Time-lag FWI algorithm (A-TLFWI) that has proven effective in mitigating the negative impact of large differences between synthetic and real data at sharp contrasts. Elastic TLFWI (E-TLFWI) can model the energy at large impedance contrasts with better amplitudes and phases. This intrinsically reduces data mismatch between synthetic and real data and thus further improves the convergence. Using an ocean bottom node (OBN) data set in the Gulf of Mexico (GoM), we demonstrate that E-TLFWI can significantly reduce the salt halo that typically occurs in A- TLFWI. Although it does not change the model kinematics much compared to A-TLFWI, as evident in RTM images, it provides FWI Images of more balanced amplitudes, improved focusing, and higher signal-to-noise ratio (S/N).