observation of seismic anisotrophy in prestack seismic data

Click here to load reader

Post on 04-Apr-2018




0 download

Embed Size (px)


  • 7/30/2019 Observation of Seismic Anisotrophy in Prestack Seismic Data


    Observations of Seismic Anisotropy in Prestack Seismic Data David Gray, CGGVeritas


    A method for displaying prestack seismic data thathighlights azimuthal anisotropy is shown. Using thismethod to make observations of azimuthal anisotropy inprestack seismic data makes it clear that azimuthalanisotropy is observed in all types of seismic data and thatit is widespread from the shallowest to the deepest seismicreflectors. Therefore, we must account for anisotropy whenwe are processing and interpreting these data.


    Visualizing anisotropy in prestack seismic data has beendifficult. Cheadle et al (2001) describe a means of

    displaying prestack offset and azimuth data as a 3D cube.Henceforth, I will refer to this as a COCA (Common-Offset, Common-Azimuth) Cube. Todorovic-Marinic(unpubl. results) came up with a way to view these cubes in2D that allows a more intuitive understanding of theanisotropic variation in conjunction with the AVO variationin the data. This and more conventional methods are usedto demonstrate the large degree of azimuthal anisotropy inboth land and marine prestack P-wave seismic data.


    The COCA cube extends the common offset stack byadding the third dimension of azimuth (Cheadle et al,

    2001). In our implementation, a 3D cube is created whereeach bin represents a range of shot-receiver offsets andazimuths. For example, the first bin might represent allazimuths from 0-10 degrees and all offsets from 0-100 m.Then all traces that have these parameters are stacked andthe resulting trace placed in this bin. The process isrepeated for all desired combinations of offsets andazimuths. E.g. an offset binning of 0 - 4000 m by 100 mand an azimuth binning of 0 180 degrees by 10 degreeswould result in 40 offsets x 18 azimuths = 720 bins. Eachbinned offset is then represented as an inline and eachbinned azimuth as a cross-line in a 3D seismic volume,which for this example would have 40 inlines and 18 cross-lines, corresponding to offsets 50, 150, , 3950 m andazimuths 5, 15, , 175 degrees, respectively. The

    resulting 3D volume can be visualized using any 3Dvisualization software, as is shown in Figure 1. We havebeen using it for quality control of azimuthal processes,such as azimuthal NMO and azimuthal AVO, in much thesame way that the common offset stack is used for qualitycontrol of conventional NMO and AVO. Here the COCACube is used to demonstrate the ubiquitous nature of

    azimuthal anisotropy and the significant degree of thisanisotropy that can be observed in prestack seismic data.


    Using the COCA cube has allowed us to visualizeazimuthal anisotropy in a way we were unable to do before.This has demonstrated to us that azimuthal anisotropyoccurs throughout the seismic section from top to bottomand in all types of environments. Using the COCA Cubeand other tools that allow for visualization of anisotropy inprestack seismic data, I present prestack observations of azimuthal anisotropy gleaned from seismic gathers fromvarious seismic surveys throughout the world. Theseobservations come from land and marine seismic surveys

    (Wombell, 2006 and Hung et al, 2006) and from veryshallow to deep targets. The degree of azimuthal anisotropycan be huge. One such example, shown in Figure 2,demonstrates azimuthal velocity anisotropy from seismicshot over the Pinedale anticline. This velocity variationwith azimuth results in timing variations on the order of 20-30 ms. In these data, this is the dominant wavelength.Therefore, we can expect destructive interference if we tryto stack these long offsets. This response is from the top of the anticline, so dip is not a significant factor in theseobservations. Furthermore, any amplitude analysis,including AVO, is severely distorted by this phenomenon.Removal of the effects of azimuthal velocity (Figure 3)demonstrates that the resulting gathers are suitable forAVO analysis. Both AVO and azimuthal AVO can be seen

    on the event just below 2000 ms, as well as on many otherevents, after the azimuthal velocity effect has beenremoved. Figure 4 shows the effect of applying theazimuthal velocity correction before stacking the data. Theresulting stack shows a significant improvement in signalcontinuity throughout the section and especially close to thefaults. More examples of both azimuthal velocityanisotropy and azimuthal AVO will be presented. Fromthese observations, it can be seen that azimuthal anisotropyis a fundamental aspect of seismic data. Furthermore,these observations demonstrate that azimuthal anisotropy isubiquitous in conventional P wave seismic data: it appearsin land surveys and marine surveys and at depths rangingfrom the very near surface to greater than 5000 m.

    Cheadle et al (2001) make an observation that may explainthe observed shallow anisotropy. Refraction tomographyrun in different directions on the seismic data recorded atWeyburn Field in Canada shows different near-surfacevelocity profiles. This implies that azimuthal anisotropyoccurs in the near surface weathering layers. Furthermore,Cheadle et al show that these near-surface velocity profiles

  • 7/30/2019 Observation of Seismic Anisotrophy in Prestack Seismic Data


    Observations of seismic anisotropy in prestack seismic data

    change with time. Observations such as Cheadles andthose of Davis and Benson (2001) and Xia et al (2006)suggest that azimuthal anisotropy may vary with time aswell. These results suggest that differences in anisotropybetween seismic surveys may be a very sensitive 4Dindicator of reservoir stress field variations due toproduction. These observations mean that we can nolonger afford to process and interpret seismic data underthe nave assumption that the earth is isotropic.


    The COCA Cube is an excellent tool for examination andquality control of azimuthal anisotropy in seismic data andany processing done that might affect it. Just as thecommon offset stack serves as quality control for NMO andvisual confirmation of AVO anomalies, the COCA Cubealso serves as an important visual quality control of azimuthal velocity anisotropy and azimuthal amplitudeanisotropy attributes.

    The observations in this presentation suggest that seismicdata contains widespread azimuthal anisotropy and thatanisotropy is observable in all types of seismic data: landand marine, shallow and deep, conventional andmulticomponent. Therefore, at a minimum, it must beaccounted for in order to properly process the data.Optimally, we will use anisotropy to constrain possibleinterpretations of seismic data better than we can when weassume the earth is isotropic.


    The author would like to thank CGGVeritas for permissionto show these data and Dragana Todorovic-Marinic, LoreleiMatthews, Scott Cheadle, Richard Wombell, Barry Hung,Lei Wei and Wayne Nowry for their assistance in thiswork.

    Figure 1: A COCA (Common-Offset, Common-Azimuth) Cubedisplayed using 3D visualization .

  • 7/30/2019 Observation of Seismic Anisotrophy in Prestack Seismic Data


    Figure 2: A COCA Cube displayed in 2D with offsets increasing from right to left (Todorovic-Marinic, unpubl. results). Common-azimuth binsfrom 0 180 degrees are displayed within each common offset (i.e. between the yellow lines). Velocity anisotropy is observable as wobblesacross the reflections in the common azimuths corresponding to longer offsets on the left. Azimuthal AVO is observable in the amplitudechanges along the reflections in the common azimuths on the left.

    Figure 3: Evidence for azimuthal AVO becomes clearer in the amplitude variations within the common azimuths at longer offset reflections onthe left when the azimuthal velocity variations are removed.

  • 7/30/2019 Observation of Seismic Anisotrophy in Prestack Seismic Data


    Observations of seismic anisotropy in prestack seismic data

    Figure 4: The stack on the right, stacked after applying azimuthal velocities, shows much better event continuity and signal-to-noise ratio,especially around the fault (green oval), than in the stack on the left, which has been stacked with isotropic velocities.