Ultra-long offset node acquisitions are beneficial for full waveform inversion (FWI) because the wide variety of wave types recorded by these geometries are suitable for broadband velocity model building. However, long propagation distances induced by long offsets exacerbate cycle skipping and the sparsity of the acquisition can inject wraparound artifacts into the reconstructed velocity model.
In this presentation, we propose a FWI technology that is suitable for sparse long-offset OBN acquisition. Cycle skipping is mitigated through anextended search space in the framework of augmented Lagrangian and alternating direction method of multipliers (ADMM). Also, the robustness of the method for sparse acquisition is achieved with phase retrieval and sparsity-promoting regularization.