--> ABSTRACT: Quantifying the Benefit from Pressure Gradient Data in Wavefield Reconstruction for Time-Lapse Seismic, by Eggenberger, Kurt; Christie, Phil; Muyzert, Everhard; Vassallo, Massimiliano; Özbek, Ali ; #90135 (2011)
[First Hit]

Datapages, Inc.Print this page

Quantifying the Benefit from Pressure Gradient Data in Wavefield Reconstruction for Time-Lapse Seismic

Eggenberger, Kurt 2; Christie, Phil 1; Muyzert, Everhard 1; Vassallo, Massimiliano 2; Özbek, Ali 1
(1)Geophysics, Schlumberger Cambridge Research, Cambridge, United Kingdom. (2) Data Acquisition and Conditioning, WesternGeco, London, United Kingdom.

The repeatability of time-lapse seismic data depends critically upon the repeatability of the acquisition geometry. Despite the availability of high precision navigation systems and steerable streamers, matching seismic survey geometries can require spatial over-sampling and/or wavefield interpolation. Because of sparse crossline sampling, this interpolation is usually aliased. Marine streamers exist which acquire acoustic pressure and vertical particle velocity data. It can be shown that the particle velocity vector is intimately related to the pressure gradient vector. If we could acquire the horizontal gradients of the acoustic pressure wavefield in towed marine seismic surveys, this would greatly alleviate the spatial sampling issue by at least doubling the crossline Nyquist wavenumber.

This paper explores the benefit of using inline gradients computed from field data in a simulated crossline reconstruction of seismic wavefields. Finely-sampled pressure and gradient data, from a common sail line acquired in two time-lapse seismic vintages, are taken from a seismic processing flow after surface multiple attenuation. The data are decimated without anti-Previous HitaliasTop wavenumber filtering and then reconstructed using a variety of algorithms both with and without the gradient measurement. The reconstructed traces are compared with the undecimated data to assess the interpolation errors.


Comparisons are made between the reconstructed traces and the undecimated traces both within the same time-lapse vintage and across the two time-lapse vintages: the former examines only the reconstruction error while the latter estimates the total time-lapse non-repeatability. The errors from reconstructing the data using the gradient information up to twice the pressure-only Nyquist wavenumber are well within acceptable time-lapse bounds and much less than the general survey-to-survey non-repeatability. The use of a multi-channel interpolation by matching pursuit algorithm, in particular, offers good promise for higher order de-aliasing beyond twice-Nyquist wavenumber.

 

AAPG Search and Discovery Article #90135©2011 AAPG International Conference and Exhibition, Milan, Italy, 23-26 October 2011.