Receiver extension strategy using a time-dependent relocalization approach for time-domain FWI
Résumé
Main objectives In this study, we propose a receiver-extension strategy for full waveform inversion, using time-dependent receiver position relocalization. This allows to fit multiple arrivals, when the model estimate is poor at the early iterations of FWI, mitigating cycle-skipping.
New aspects covered We obtain an optimal receiver position correction as a function of time, fitting multiple arrivals using a pseudo-global optimization strategy at each FWI iteration, which puts the observed and the calculated data in phase and the correct model update is obtained from the amplitude discrepancy.
New aspects covered We obtain an optimal receiver position correction as a function of time, fitting multiple arrivals using a pseudo-global optimization strategy at each FWI iteration, which puts the observed and the calculated data in phase and the correct model update is obtained from the amplitude discrepancy.
Origine : Fichiers éditeurs autorisés sur une archive ouverte