Full waveform inversion (FWI) of limited-offset seismic reflection data achieves broadband imaging only in the shallow subsurface sampled by the diving waves. In order to image deeper targets, Reflection WI (RWI) exploits wavepaths generated by reconstructed reflectors to constrain the smooth part of the velocity field. Joint FWI (JFWI) complements the RWI kernel at shallow depths by inverting simultaneously for diving and reflected waves. The JFWI solution, used as starting model, provides FWI with the low wavenumbers necessary to converge to a broadband velocity. This approach is not immune to the inherent non-linearity of L2-norm waveform inversion with respect to the wavefield kinematics, and benefits from the use of misfit functions robust to cycle-skipping. Here we attempt to reconstruct a complex subsurface from limited-offset streamer synthetic reflection data, using a 1D starting model. The strategy successfully combines JFWI with a graph-space Optimal Transport (GSOT) objective function, whereas L2-based JFWI suffers from severe cycle-skipping. Furthermore, the results attest the benefits of jointly inverting for diving and reflected waves in JFWI.
Presentation Date: Monday, October 12, 2020
Session Start Time: 1:50 PM
Presentation Time: 3:30 PM
Location: Poster Station 3
Presentation Type: Poster
