We compare the results from the application of a standard implementation of elastic full-waveform inversion (EFWI) with those from an amplitude versus angle (AVA) inversion to predict elastic models. The tests are carried out on synthetic seismograms from the Marmousi-2 model, considering the pressure component data only, thus simulating towed-streamer seismic marine data. EFWI is performed by applying a time-domain EFWI with a steepest-descent optimization. After a basic processing sequence, which includes pre-stack time-migration, the seismic data have been AVA-inverted by applying a Bayesian linear inversion. The same low-frequency velocity field, obtained by smoothing the true Marmousi-2 model, has been used as the starting model for EFWI and as the low frequency trend for AVA inversion. Both inversion strategies provide accurate Vp estimates, but the AVA inversion yields more correct predictions of Vs in the gas-reservoir. It seems that the error minimization in EFWI is dominated by the acoustic information (particularly diving waves) and the elastic information present in the data is not enough to drive the inversion towards the correct estimation of Vs when they are not directly correlated with Vp. Thus, more advanced EFWI implementations and/or inversion strategies are needed when dealing with single-component pressure seismograms.

Comparing Elastic Full-Waveform and AVA Inversions: Preliminary Results on Synthetic Data

Tognarelli Andrea;Aleardi Mattia;Mazzotti Alfredo
2018-01-01

Abstract

We compare the results from the application of a standard implementation of elastic full-waveform inversion (EFWI) with those from an amplitude versus angle (AVA) inversion to predict elastic models. The tests are carried out on synthetic seismograms from the Marmousi-2 model, considering the pressure component data only, thus simulating towed-streamer seismic marine data. EFWI is performed by applying a time-domain EFWI with a steepest-descent optimization. After a basic processing sequence, which includes pre-stack time-migration, the seismic data have been AVA-inverted by applying a Bayesian linear inversion. The same low-frequency velocity field, obtained by smoothing the true Marmousi-2 model, has been used as the starting model for EFWI and as the low frequency trend for AVA inversion. Both inversion strategies provide accurate Vp estimates, but the AVA inversion yields more correct predictions of Vs in the gas-reservoir. It seems that the error minimization in EFWI is dominated by the acoustic information (particularly diving waves) and the elastic information present in the data is not enough to drive the inversion towards the correct estimation of Vs when they are not directly correlated with Vp. Thus, more advanced EFWI implementations and/or inversion strategies are needed when dealing with single-component pressure seismograms.
2018
9789462822542
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11568/924688
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact