Seismic data acquisition research

1. During Vermeer’s time as a seismic processor in NAM (1976-1980), the quality of the 2D seismic data strongly improved due to technological advances, in particular the availability of more recording channels and a corresponding reduction of spatial sampling intervals. However, there was not a thorough understanding of the properties of the 2D seismic lines. To fill this gap Vermeer wrote an extensive report that focused on the spatial properties of the prestack wavefield of the 2D line. The main title of the report was “The four domains” which were the four spatial domains of common shot panel, common receiver panel, common midpoint panel and common offset panel (panel = gather). Symmetric sampling of shots and receivers was introduced as a direct consequence of the reciprocity theorem. Much attention was also given to patterns (= field arrays) as a means to compensate for not using the ideal sampling technique: sampling at the basic sampling interval. The report was published as an internal Shell report in 1982.
2. Permission to publish “The four domains” in the open literature was originally not granted by Shell management and came only after Anstey had published his “Whatever happened to ground roll?” in The Leading Edge in 1986. Finally, in 1990, the SEG published Vermeer’s book “Seismic wavefield sampling”. From the Introduction: “Also, as a result of the discussion that followed Anstey’s article, my understanding of the subject matter matured considerably and many refinements and discussions of articles published by various authors in the 1980s that are included were absent in the Shell report.” The sampling paradox (proper sampling of shot and receiver gathers does not lead to proper sampling in the common offset gather nor in the common midpoint gather) was resolved with reference to the N-dimensional sampling theorem of Petersen and Middleton (1962). Some readers appeared to be surprised to find much new information in the book that had not been published earlier in journal papers. The book is still available as an eBook at SEG’s Bookmart.
3. In 1991 Vermeer joined Justus Rozemond (the later inventor of the slip-sweep technique) in a Shell research project on 3D seismic data acquisition. Types of geometry were identified as line geometries (parallel, orthogonal, zigzag, slanted) and areal geometry. All line geometries can be considered as a collection of 3D basic subsets consisting of a combination of a shot line and a receiver line. Sampling should be dense along the acquisition lines and is permitted to be sparse between the lines. The 2D line was found to be a special 3D subset of parallel geometry with a collocated shot line and receiver line. In areal geometry (nowadays often called node or nodal geometry) sparse receivers are covered by dense shots (sparse and dense in x and y) and produce a 3D common receiver gather as the basic 3D subset. The basic subsets of orthogonal geometry (the cross-spread) and of areal geometry are limited by maximum useful offset; therefore they should extend as far in the crossline as in the inline direction. To test this research a 3D microspread (cross-spread with 2-m sampling intervals) was acquired in The Netherlands in the spring of 1992 and a first wide-azimuth test survey was acquired in Nigeria in December 1992. Vermeer published an SEG Expanded Abstract “3-D symmetric sampling” in 1994, followed by a Geophysics paper with the same title in 1998. These papers (and a 1997 OTC paper) also covered the explanation of spatial discontinuities for the long offsets in marine streamer acquisition.
4. In a neat little paper (Brühl et al., 1996) the Fresnel zone for a broadband wavelet was properly defined. Using this definition it is clear that for imaging more than the Fresnel zone is required to get a proper image; the Fresnel zone only includes constructive interference whereas for a proper image also destructive interference is required. This larger zone is called the “zone of influence”.
5. In another 1996 paper Vermeer shook the industry not only with the title of his SEG paper (DMO in cross-spread: the failure of existing software to handle amplitudes correctly) but also with the contents. Not a single processing contractor produced correct stable amplitudes when applying DMO in the cross-spread. Not surprisingly, various contractors presented the results of their much improved DMO technique at the 1997 SEG Conference.
6. In 1997 and later the subject of resolution was addressed in papers all titled “Factors affecting spatial resolution”, first in an SEG 3D sampling workshop and culminating in a 1999 Geophysics paper. In this paper the width of the spatial wavelet is measured in the horizontal direction, this width being representative of the minimum horizontally resolvable distance. For 3D spatial resolution the paper analyzes various minimal data sets (singlefold data sets suitable for migration, Padhi and Holley, 1997) such as cross-spread, 3D shot, and common-offset gather. The paper also looks at the effect of sampling and explains why sampling does not have much effect on simple resolution analysis. Also, for resolution analysis it is important to compare samplings for the same aperture (= number of samples N times sampling interval). Von Seggern (1994) used (N-1) * sampling interval and this led to erroneous results. The paper also argues that bin fractionation or flexi-binning does not improve resolution as the sampling intervals of the cross-spread are not altered by those techniques.
7. In 1999 Vermeer addressed the properties of converted waves as recorded in the various 3D subsets of different acquisition geometries. These properties turned out to be quite different, especially because of the asymmetric raypaths even in a horizontal geology. Recognition of those properties leads to recommendations on how to design surveys aimed at the acquisition and exploitation of converted waves.
8. Prestack imaging of crossed-array geometries. Until the late 1990s, velocity analysis for prestack migration was carried out by migration of small ranges of absolute offset. This was fine for narrow-azimuth acquisition, but not for wide-azimuth orthogonal geometry data. Vermeer and Grimbergen (1998) illustrated this struggle for a better solution using cross-spreads. Later that year Vermeer presented the use of offset-vector tiles (OVTs, then still called offset/azimuth slots) in Exploration Geophysics. This was still presented as just an idea, but the idea was worked out in much more detail at the Mintrop seminar in 2000. In the meantime Peter Cary presented his ideas on common-offset-vector gathers (his name for OVT gathers) at the 1999 SEG conference and in a 1999 Crewes report.
9. Early 2001 Vermeer collected results of his work (including as yet unpublished material) on 3D seismic data acquisition in a PhD-thesis defended successfully at Delft University of Technology. A commercial version of this thesis was published by the SEG in 2002.
10. In thesis and SEG book the virtues of using different types of gathers (in particular cross-spreads and OVT gathers) in processing were clearly described. And indeed, using the well-sampled subsets of 3D acquisition geometries such as cross-spreads in noise removal was gradually being picked up and reported upon. Yet, using OVT gathers in processing was not taken up yet by the industry. To try and get the message across, Vermeer presented “Processing orthogonal geometry – what is missing?” at the 2005 SEG Conference. It was only in 2008 that Calvert et al. (GXT) published papers at the SEG Conference and in First Break showing the benefit of OVT processing. Next year, Lecerf et al. (CGG) demonstrated the benefit of OVT gathers for azimuthal moveout correction. Today, the whole industry uses OVT gathers in processing as a standard tool for velocity modeling, azimuthal anisotropy analysis and other applications.
11. Important insight. OVT gathers are the closest one can get to common offset-vector gathers, which are minimal data sets. Minimal data sets being singlefold gathers may be used in prestack migration, even though their S/N may be pretty small. So, OVT gathers are used as the best alternative to minimal data sets and may produce M images of the subsurface for an M-fold regular geometry. The geometry must be regular to allow construction of M singlefold OVT gathers.
12. There is now general consensus in the industry that wide-azimuth acquisition is required for the best illumination of the subsurface, especially in complex areas and also in marine data acquisition. All of the existing wide-azimuth towed streamer configurations are full of compromises to keep cost within acceptable bounds. In 2009 Vermeer proposed a potential alternative with less compromises. This method uses one or more separate source vessels to shoot zigzag lines across a streamer spread being towed at sufficiently deep depth. The idea of zigzag lines has not been tried yet, but a source vessel above the streamers was introduced in 2016 with CGG’s TopSeis.
13. In 2012 SEG published the Second Edition of “3D seismic survey design”. This version of the book captures new insights and new technology as gathered over the years since publication of the first edition. A very special contribution (Appendix B) was written by Gary Hampson on aperture for sampled wavefields. It proves the validity of the formula Length = N * sample interval (see point 6 above). As OVT gathers are not suitable for imaging of converted waves, the book includes an idea on how to get rid of the edge effects of all cross-spreads in orthogonal geometry. Somebody may try it. Another idea is the use of areal geometry in the desert with a sparse grid of receivers each consisting of an array of individually recorded geophones and a dense grid of vibrator points.
14. In “Seismic wavefield sampling” a formula is given relating NMO stretch to offset, traveltime and Vrms, including the derivative of Vrms with respect to traveltime. Theoretically, the stretch factor S = 1 / cos i, where i is the angle of incidence at the reflection point. Using these formulas Vermeer showed at the 2013 SEG Conference that survey design, especially for AVO requirements, could be improved by including a better relationship between maximum offset and maximum acceptable stretch.
15. An improved version of the Survey Design Wizard was installed at this website in 2016. This new version includes the improved formula for the stretch factor and also more flexibility in the choice of station interval. This software is freely usable on the website or also as a download.