You are on page 1of 4

Wide-azimuth processing for azimuthal anisotropy analysis

Cynthia Gomez, Erika Angerer*, CGG Technology London

variations, transform-based signal processing algorithms


Abstract need to be applied to data split or sorted into azimuth
Azimuth-friendly data processing is mandatory in a reliable sectors. A 3D FK cone filter preserves azimuthal
fracture characterisation workflow. Here, the effects of anisotropic variations, and for this the data are processed
standard processing methodologies in the presence of continuously. Moreover, surface-consistent processing
azimuthal anisotropy are investigated on synthetic and real steps are applied to the data set continuously since source
data examples. Common techniques for signal processing, and receiver related near-surface effects are azimuth
statics calculations, imaging and velocity analysis are independent.
adapted for wide-azimuth, wide-offset P-wave data. We
present a general processing sequence that preserves Statics
azimuthal anisotropy. Further, we compare anisotropy Residual static shifts have a significant impact on the
parameters derived from impedance inversion and reliability of amplitude-based anisotropy attributes. Figure
azimuthal AVO on a Middle East data example. 2a shows an overlay of the intensity or magnitude of the
inverted ellipses (in colour) with the noise-free intercept
Introduction trace (wiggle). Figure 2b is the overlay with the orientation
Observable effects of azimuthal anisotropy on seismic data of the maxima of the inverted ellipses as the colour
are of second order compared to the geological attribute. Without static problems the maxima of the
background. A careful preservation of azimuthal amplitude inverted intensity correspond to the maxima in the data and
and travel time variations is therefore crucial. One of the the orientation is consistent in time. Figures 2c and 2d
main questions in processing wide-azimuth data in the show the corresponding overlays in the presence of residual
presence of azimuthal anisotropy is when to process the static shifts that are up to 20% of the dominant wavelength.
entire volume continuously as a single data set and when to In Figure 2c, the intensity maxima occur at the zero-
split and process the data in azimuth-limited sectors. We crossings of the intercept trace. Moreover, in Figure 2d, the
devised a testing sequence using synthetic, anisotropic, inverted orientation shows cyclic 90° variations introduced
wide-azimuth data. Processing steps are applied to the data by random as well as systematic static shifts. Even small
with and without azimuth sectoring and the preservation of static shifts have a significant impact on the inverted
anisotropy is assessed. anisotropic attributes. However, static problems are readily
identified by the described effects in Figures 2c and 2d. In
Data processing sequence order to remove static problems consistently within the
Signal processing entire volume, the data need to be processed continuously.
Of specific interest are transform-based signal processing In the case of a limited-azimuth distribution (e.g. shot point
techniques like 2D and 3D FK, and FX noise suppression along the edge of the survey), there is a potential of
methods. Figure 1 shows the comparison of the resulting introducing an azimuthal bias to the surface-consistent
amplitudes before and after application of FX statics solution.
deconvolution to the synthetic data. The solid lines are the
exact amplitudes calculated using Rüger’s elliptic Imaging and velocities
approximations of the anisotropic reflection coefficient
Prestack imaging is performed on each azimuth sector
(1998). The magnitude of the azimuthal variation increases
separately keeping the parameterisation constant.
with increasing incidence angle. Figure 1a shows the best
Importantly, this approach corrects for the azimuthal effects
fitting ellipses to the amplitudes after adding noise introduced by lateral velocity variations (i.e. dipping
(S/N=0.5) and applying FX deconvolution in azimuth
reflectors). Dipping reflectors produce sinusoidal azimuthal
sectors. Figure 1b shows the resulting amplitudes after FX traveltime variations that interfere with azimuthal
deconvolution treating the data as a continuous volume and
anisotropy attributes extracted prior to imaging. Azimuthal
thereby mixing different azimuths. It is evident that the
residual moveout analysis is performed on azimuth-
azimuthal content of the data is preserved when the data are
sectored imaged data.
filtered in azimuth sectors. There is a general shift in the
ellipses that is due to the added noise. In Figure 1b
amplitudes are not correctly preserved, the azimuthal
Spectral balancing
For the extraction of amplitude-based attributes spectral
variation is smeared out and therefore anisotropy is not
preserved. Using the same approach we also tested 2D FK variations in offset and azimuth need to be compensated
for. These variations can be caused by moveout stretch,
filtering and τ-p transform. The results are similar to the
differential attenuation, and acquisition effects. A
ones shown in Figure 1. In order to preserve azimuthal
Azimuth-preserved processing

controlled approach to compensate these effects is to Azimuth-sectored Continuous processing


balance the power, amplitude or signal spectra of partial processing
azimuth and offset/incidence stacks. Transform-based noise 3D FK
attenuation: 2D FK,
Data examples of anisotropy attributes FX deconvolution Surface-consistent
Following the results of synthetic tests, we applied an processing
azimuth-friendly processing sequence to a Middle East data τ-p transform Statics
set. Here, we compare anisotropy attributes derived from Prestack imaging Trace-by-trace gains,
azimuthal AVO and azimuthal elastic impedance inversion. scalars, filters
Figure 3 shows the inverted anisotropy intensity and Spectral balancing
orientation of a crossline. The maxima of the intensity Table 1: Overview of azimuth-sectored and
coincide with peaks and troughs in the data and the
continuous processing techniques
orientations are vertically and laterally consistent
NOISY SYNTHETICS (signal to noise=0.5)WITH SPARN/NOISE FREE
confirming a reliable processing sequence.
80
Figure 4 shows the derived anisotropy intensity maps of
interval velocities of a 150 ms interval and (b) the intensity 70
a
derived from azimuthal AVO along the base of the interval. 60

In general, there is good agreement between the two maps 50


for zones of both high and low anisotropy. A zone of high
Amplitude

40
anisotropy, marked by red boxes occurs in both attributes, 30
thereby increasing the confidence in this anomaly. line 1
20
Figure 5 shows orientation-intensity maps along the same line
line
2
3
horizon as in Figure 4 derived from (a) azimuthal AVO and 10 line 4
line 5
(b) azimuthal impedance inversion as presented by Angerer 0 line 6
line 7
et al. (2003) using a layer-based, 3D algorithm. Again, both -10 line 8
maps show consistent trends. However, the impedance 0 20 40 60 80 100 120 140 160 180
inversion result shows improved lateral coherence and a Azimuth(as one gather)/NOISE FREE
NOISY SYNTHETICS WITH SPARN
reduced standard deviation of the inverted anisotropy 80
parameters. 70

60
b
50
Summary
Table 1 gives an overview as to whether to apply standard
Amplitude

40

processing techniques to wide-azimuth data with the data 30


treated as a single volume or separated in sectors. It 20
becomes evident that azimuth-sectoring is part of any
10
amplitude- and traveltime-preserving processing sequence.
This forms then the basis for the extraction of reliable 0

anisotropy parameters, which are then ready to be -10

integrated in a fractured reservoir characterisation 0 20 40 60 80 100 120 140 160 180


workflow. Azimuth

Figure 1: Amplitude preservation of FX


deconvolution on wide-azimuth data; solid lines:
References exact azimuthal amplitudes of noise-free synthetic
Rüger, A., 1998, Variation of P-wave reflectivity with test data; dashed lines: data with added random
offset and azimuth in anisotropic media, Geophysics, Vol. noise (S/N=0.5) and application of FX
63, No. 3, 935-947. deconvolution in (a) azimuth-sectored gathers and
Angerer, E., Lanfranchi, P., and Rogers, S., 2003, Fractured (b) in unsectored (mixed azimuth) gathers;
reservoir modeling from seismic to simulator – A reality?, incidence angles: 5° (red), 15° (green), 25° (blue),
TLE, 684-689. 35° (magenta).
Azimuth-preserved processing

a b c d

Figure 2: Overlay of inverted anisotropy parameters and intercept trace; (a) intensity (blue (low) – red (high))
and (b) orientation without static shifts; (c) intensity and (d) with static shifts.

0° 90° 180°

Figure 3: (a) Intensity (white (low) – red (high)) and (b) orientation attributes of a cross-
section. The red circle marks a spatially and vertically consistent high anisotropy zone. Input
data is shown as wiggle display.
Azimuth-preserved processing

a b

Figure 4: Intensity horizon displays; (a) derived from azimuthal interval velocity analysis and (b) derived from azimuthal
AVO along the base of the interval: colour scale: white (low) – blue (high). Red box denotes high anisotropy zones
occurring in both attributes.

a b

Figure 5: Orientation and intensity horizon displays; (a) derived from azimuthal AVO and (b) derived from azimuthal
elastic impedance inversion; background map: normalised fitting error (white (low) – blue (high)).

You might also like