Checkpointing-assisted reverse-forward simulation: An optimal recomputation method for FWI and RTM
Résumé
3D implementations of reverse time migration and full waveform inversion requires efficient schemes to access the incident field in order to apply the imaging-condition of RTM or build the gradient of FWI. Wavefield reconstruction by reverse propagation using final snapshot and saved boundaries appears quite efficient but unstable in attenuating media, while checkpointing strategy is a stable alternative at the expense of increased computational cost through repeated forward modeling. In this study, we propose a checkpointing-assisted reverse-forward simulation (CARFS) method in the context of viscoacoustic wave propagation with generalized Maxwell body. At each backward reconstruction step, the CARFS algorithm makes a smart decision between forward modeling using checkpoints and reverse propagation based on the minimum timestepping cost and an energy measure. Numerical experiments demonstrate that the CARFS method allows accurate wavefield reconstruction using less timesteppings than optimal checkpointing, even if seismic attenuation is very strong. The proposed CARFS method can be straightforwardly applied to general anisotropic viscous media, providing an accurate and efficient tool for practical RTM and FWI implementation.