Full text loading...
-
Extrapolated Full Waveform Inversion Via Model Extension
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 79th EAGE Conference and Exhibition 2017, Jun 2017, Volume 2017, p.1 - 5
Abstract
Extended images has been widely used in seismic data processing and migration velocity analysis because the extended images preserve the full kinematic information in seismic reflection data, regardless of the accuracy of the migration velocity. In this paper, we propose to expand the wavenumber band in the extended image space using a variable wavelet deconvolution, subject to a finite total variation (TV) norm constraint on the extended images. This explicit wavenumber extrapolation is nested within least-squares extended reverser time migration (LSERTM) iterations to ensure the extended, full wavenumber-spectrum images reproduce the kinematics and amplitudes of the bandlimited reflection data. We then synthesize the low-frequency reflection data using extended Born modeling with a low frequency source and the previously inverted all-wavenumber extended images. Numerical examples show that the extrapolated low frequency data reconstruct the relative amplitudes and phases of the low frequency data that we could have recorded in the field. Finally, we demonstrate on a synthetic Marmousi example that the extrapolated data can be used in full waveform inversion to fill in the frequency / wavenumber gap between the low wavenumber tomographic model and the high wavenumber reflectivity model.