Sie sind auf Seite 1von 8

Fundamentals of 3-D seismic

volume visualization
GERALD D. KIDD, Paradigm Geophysical Corporation, Houston, Texas

3-D volume visualization is a a) b)


method of seismic interpretation in
which the seismic reflectivity of the
subsurface is directly evaluated in
three-dimensional space by applying
various levels of transparency to the
data. The technology and philosophy
of 3-D volume visualization differ dra-
matically from conventional inter-
pretation and include new strategies
and methodology. This article will
present some basic introductory infor-
mation which should help geophysi-
cists adapt to this new technique.
It will probably be difficult for
some interpreters, particularly those
with considerable experience and
expertise in traditional methods, to
master this new, much different inter-
pretation technique. However, I sug-
gest they make the effort because this
method has several advantages. In
particular, integration of several kinds Figure 1. (a) Seismic sample to voxel relationship. (b) ZS opacity editor and
of data allows examination of many its relationship to seismic.
key features prior to significant digi- of potential fluid contacts, can be per-
tal interpretation. This, in turn, leads formed quickly by looking into the
to higher quality interpretations in a volume. In contrast to surface visual-
shorter time frame. ization (where surfaces constitute the
3-D model), the 3-D data themselves
Background. The two basic types of constitute the model.
visualization are map-based (or sur- In voxel-based volume visualiza-
face visualization) and volume-based tion (Figure 1a), each data sample is
or (volume visualization). Surface converted into a voxel (a three-dimen-
visualization results from mapping sional pixel with dimensions approx-
individual horizons and faults, and imating the bin spacing and sample
then reinterpreting them collectively interval). Each voxel has a value that
in 3-D space as a 3-D model. This corresponds to the parent 3-D vol-
requires two interpretation steps and ume, an RGB (red, green, blue) color
the data represent only a portion of Figure 2. Workflow.
value and an opacity variable that
the 3-D volume. Surface visualization allows the degree of transparency to rected if necessary. Poor data scaling
is the logical end product of conven- be regulated. Therefore, each seismic (e.g., that severely underutilizes the
tional interpretation. Volume visuali- trace is converted to a column of vox- datas full dynamic range) or exces-
zation is based on an entirely different els. sive clipping greatly reduce the effec-
attribute of the datatransparency The data are scaled to eight-bit tiveness of volume visualization.
and represents a major paradigm shift and displayed as a histogram repre- Next, data are prepared for opaque 2-
in 3-D seismic data interpretation. senting the distribution of the voxel D slice overviews by setting defaults,
Volume visualization assumes that values (Figure 1b). Data parameters parameters, and colors so that the data
the reflectivity of the subsurface is an such as phase, frequency, and seismic represent settings similar to those in
in situ 3-D model of the subsurface signatures should be reviewed prior conventional interpretation. Then,
which, by its nature, consists of inte- to detailed visualization. This is crit- scanning is performed along all three
grated structural, stratigraphic, and ical for guiding the application of primary axes for both 3-D regional
amplitude features in 3-D space. Both transparency and in designing visu- and prospect-specific overviews.
3-D regional and prospect-specific alization strategies for specific inter- After an objective is identified, a
evaluations, including identification pretation problems. focusing strategytime windowed,
New technologies introduce new detection, or horizon-keyedisolates
Editors note: The material in this article was workflows. Figure 2 is a generalized it as preparation for application of
presented at the 1999 Offshore Technology volume visualization workflow. Start- opacity. This is a critical step in vol-
Conference, 3-6 May 1999. A longer version ing at the top, survey parameters and ume visualization because it removes
of the article is in the OTC Proceedings. data scaling are inspected and cor- extraneous data above and below the
702 THE LEADING EDGE JUNE 1999 JUNE 1999 THE LEADING EDGE 703
objective interval. In low-dipping val because they only include data
strata, time windowed visualization within the zone of interest (Figure 4).
is very effective for evaluating geol- This is ideal for evaluation of struc-
ogy (Figure 3). Detection is based on tured depositional units. A single hori-
the physical connectivity of voxels zon may be used to create a
within a user-defined amplitude bulk-shifted interval or any variation
range. The results can be immediately from two horizons may be used to
evaluated for internal amplitude vari- customize sequence visualizations
ations or to generate surfaces. This is (Figure 5).
particularly useful in evaluating spe- After focusing, opacity is system-
cific stratigraphic features, mapping atically applied, utilizing the zone
individual fault blocks, or creating system (ZS), to achieve the highest
Figure 3. Volume visualization of regional surfaces. Horizon-keyed quality volume visualization.
a Lower-Frio bright amplitude focusing strategies are the most accu-
using a time-windowed strategy. rate method for evaluating an inter- Application. ZS, an iterative and sys-
tematic process, is based on the most
important tool of volume visualiza-
a) tionthe opacity editor. ZS involves
dividing the histogram into six verti-
cal zones, three positive zones corre-
sponding to peak amplitudes and
three negative zones corresponding
to trough amplitudes (Figure 1b).
Zone one positive (Z1+) includes the
highest positive amplitudes, zone two
(Z2+) the intermediate amplitudes,
and zone three (Z3+) the lowest pos-
itive amplitudes. A three-color scale
b) (in this case red, yellow and white) is
designed to correspond to each posi-
tive zone. The correlation between a
zoned colored-coded amplitude
within the semitransparent visualiza-
tion to its corresponding position
within the opacity editor serves as the
link which directs where and how
Figure 4. (a) Horizon-keyed visualization is used to encapsulate a complex the opacity curve should be adjusted
stratigraphic zone. (b) All events within the zone will contribute to the to enhance that specific geologic fea-
interpretation. ture. This process is iterated until the
geologic features are optimally
enhanced in all zones (Figure 3).
a) The number of data points within
each zone (Figure 1b) is generally
inversely proportional to amplitude
strength. Therefore, the magnitude of
the opacity adjustments in each zone
should decrease away from maxi-
mum amplitudes. In other words,
since the relative percentage of vox-
els in Z3 is very large, opacity adjust-
ments within Z3 should be relatively
small, because small Z3 adjustments
have a large impact on the visualiza-
b) tion. On the other hand, relatively
large opacity adjustments in Z1 have
less impact on the image due to the
smaller number of voxels.
The key to distinguishing features
with clarity is the geoscientists abil-
ity to enhance the contrast between
two features. The primary compo-
nents in feature enhancement are the
base color scale, their color variations
at different opacity levels, and the
background color.
Figure 5. (a) Various horizon and (b) sequence-keyed intervals for focusing The general process of managing
on dipping stratigraphic units. colors within conventional interpre-

704 THE LEADING EDGE JUNE 1999 JUNE 1999 THE LEADING EDGE 0000
tation systems is simplethe colors single base color, red for example, can ally would group the 256 base color
generally maintain their opaqueness be expressed in 256 variations, due to opacity variations into several
throughout the interpretation. How- the eight-bit opacity editor. Each opac- shades. The background color
ever, in volume visualization, the ity level produces a different variation becomes critical since it plays a key
management of colors is complex. A of red. In practice, the interpreter usu- role in foreground-to-background
contrast. A black background is best
for most visualization because black
provides the greatest contrast against
a wide range of colors, especially dur-
ing the enhancement of subtle Z3
amplitudes at low opacity levels. Not
all colors behave in the same way with
respect to background color and opac-
ity level. Therefore, colors can be clas-
sified into two groupscolors that
contrast well at high-opacity levels
and colors that contrast well at low-
opacity levels. This is critical, espe-
cially when working with subtle Z3
features, where Z3 amplitudes can be
Figure 6. Interpretation objectives between 1448 and 1520 ms. further subdivided into subzones,
allowing additional contrast for very
subtle features. Therefore, in Z3 and
a) b) sub-Z3 visualizations, where opacity
levels are usually less than 50%, only
colors that remain bright at low opac-
ity should be used. Colors that work
well in low opacity (with respect to a
black background) are white, yellow,
green, and light blue.

A stratigraphic visualization. The


objective is to define the depositional
d) environment and delineate prospec-
c)
tive stratigraphic facies in the upper
portion of a 400-ms channelized sec-
tion (Figure 6) with variable ampli-
tudes and discontinuous nonparallel
events.
After focusing on the interval, a
first-pass opacity curve sorts the fea-
tures into the three positive zones
(Figures 7a-b). Then, a visual or ampli-
tude inventory of each zone is made
Figure 7. (a) Initial opacity curve with (b) key amplitudes classified by according to color, and identifiable
zone. (c) Adjusted opacity curve increases the visibility of Z1 and Z2 geologic features are chosen for
amplitudes and decreases the visibility of Z3 amplitudes. (d) Resultant enhancement. As a feature is
visualization. enhanced, ideas are generated to assist

a) b)

Figure 8. Detailed visualization of


complex channel fan systems using Figure 9. Volume visualization (a) and interpretation (b) of a lowstand
time-windowed focusing and ZS. channel fan system.

0000 THE LEADING EDGE JUNE 1999 JUNE 1999 THE LEADING EDGE 705
identification of other events, which
a) are then classified by zone and
enhanced. In this case, the opacity of
Z1 and Z2 amplitudes are increased
and opacity of Z3 decreased (Figures
7c-d). The process should be iterated
until the geologic features in all zones
have been visually optimized (Figure
b) 8).
Figures 9a-b show visualized
stratigraphic details and facies inter-
pretation of the northern lowstand
channel-fan system. Sand is inter-
preted to bypass the incised valley and
wide channelized zone, and have been
deposited basinward as an amalga-
c) mated unit. Updip, in the wide chan-
nelized zone, the southernmost
channel shows a south-trending cre-
vasse splay with two delta-like progra-
dational lobes. The amalgamated
sands and crevasse splay facies are
bright Z1 amplitudes. During back-
filling of the incised valley, fine grain
Figure 10. Correlation between volume visualization and 2-D cross-lines. sediments are interpreted to have been
(a) Proximal contourite (A) and crevasse splay (B). (b) Distal contourite (C). diverted and transported southward
(c) Braided fan (D). by transverse paleocurrents and
deposited as contourites. The con-
tourites are classified as Z2 amplitudes.
The very low amplitudes, outboard of
the contourites and north of the chan-
nelized zone, are classified as Z3 and
interpreted as sediment-starved areas.
Figures 10a-c illustrate the corre-
lation between three 2-D dip sections
and the time-windowed visualization
(Figure 8). In Figure 10a, cross-line 70
intersects the proximal portion of the
contourite (A) and the crevasse splay
(B), both of which exhibit southward
depositional directions. In Figure 10b,
cross-line 110 intersects the distal por-
Figure 11. 2-D lines showing reflection free channels. tions of the contourite (C), which are

a)

b)

c)

Figure 12. Volume visualization results after apply- Figure 13. Cross-referencing between (a) 2-D seismic
ing ZS to Figure 11. events, (b) digitization, and (c) volume visualization of a
point bar and cut bank.
706 THE LEADING EDGE JUNE 1999 JUNE 1999 THE LEADING EDGE 0000
a)

b) Figure 15. Visualized results of


strategies in Figure 14.
related to the northern incised valley.
Cross-line 210 (Figure 10c) intersects a
northwest-trending lowstand braided-
fan facies (D), sourced from the south-
c) ern incised valley. Multiple low-
sinuosity braided channels and lens-
shaped bars characterize the braided
fan facies. Without volume visualiza-
tion, in-line 210 could be misinter-
preted as massive contorted slump
Figure 14. (a) Wavy and discontinuous variable amplitude interpretation deposits. In this case, the initial depo-
objective. (b) Horizon-keyed focusing strategy. (c) Zonal classification for sitional model, based strictly on 2-D
application of visualization. lines, was significantly simpler than
that shown in the visualization.

Visualization-guided interpretation.
The information within volume visu-
alization is primarily a visual display
that cannot be physically grasped like
a paper map. If a standard hard copy
is made, the 3-D aspects are lost due
to projection onto a 2-D surface. To
grasp the seemingly nebulous visu-
alization, conventional interpretation
processes, mainly digitizing and map-
ping, are performed in conjunction
with the visualization. Normally, the
digital interpretation is independent
of the visualization and resultant sur-
faces are later inserted into the semi-
transparent volume to complete the
integration.
Figure 11 shows two 2-D sections
Figure 16. Zonal classification of a faulted reflector within a time- through a highly channelized interval.
windowed interval. The objectives are to map the reflec-
tion-free channels (arrows), under-
a) b) stand the complexity of the
depositional environment, and iden-
tify prospective facies. Due to the low
(Z3) reflectivity of the channel-fill
and geologic complexity, automatic
horizon and mapping tools are not
effective. As a result, conventional
interpretation would be time-con-
suming, difficult, and uncertain.
However, a technique called visuali-
zation-guided interpretation can
improve the situation.
Figure 17. (a) Zonal classification of a
faulted horizon with respect to the opac-
ity editor and color scale. (b) Opacity
adjustment, rendering peak amplitude
and fault reflectivity transparent.
0000 THE LEADING EDGE JUNE 1999 JUNE 1999 THE LEADING EDGE 707
Focusing on the objective interval view (Figure 13c), a coupled cursor is interpretation are continued until a
(1692-1720 ms) and application of ZS used to correlate features between dis- sufficient grid is made and interpo-
generates an interpretation with two plays. This correlation makes the 2-D lated into a map.
primary sinuous channels and asso- interpretation and digitizing process
ciated channel facies (Figure 12). The easier, because an event on the 2-D Horizon-keyed or sequence visuali-
channels are represented by very low line can be directly correlated with its zation. Interval objectives with sig-
amplitudes (Z3) and are visually map expression with respect to the nificant dip require surfaces to isolate
defined by enhancing the contrasting surrounding geology as revealed the data. Thick complex stratigraphic
data that confine them. within the visualization. For exam- units cannot be fully represented by
By utilizing multiple windows to ple, in the interpretation of in-line 201, a surface; however, they are usually
display 2-D lines as cross-sections the abrupt amplitude termination associated with a nearby, mappable,
(Figures 13a-b) and using the semi- (Figure 13a, at cursor location 2) cor- sequence boundary. Therefore,
transparent visualization as the geo- responds to the cut bank in the visu- sequence boundaries are often used to
logic depositional model in map alization (Figure 13c). Digitizing and isolate stratigraphic packages for
visualization. Since the sequence
boundary itself is not the economic
objective and its relief is generally
smooth, its interpretation should be
minimized.
Figure 14a shows data that would
be a difficult interpretation challenge
for conventional techniques. The zone
of interest is characterized by discon-
tinuous, wavy, and shingled reflec-
tors. Due to the high frequency of
discontinuities and cycle splitting,
auto detection tools are of little use.
Therefore, detailed and time-con-
suming line-by-line interpretation
would be performed.
However, horizon-keyed visuali-
zation can make the interpretation
quick and simple. In Figure 14b, an
underlying high-amplitude continu-
ous reflector was easily interpreted
and bulk-shifted upward twice (20 ms
Figure 18. Structural visualization using ZS, rendering the seismic reflec- and 48 ms) to encompass the zone of
tivity of the fault transparent. interest (Figure 14c). Large arcuate
dune-like features are revealed
(Figure 15) and interpreted as deep-
water contourites. Small down-to-
basin fault complexes appearing as
lineaments in the center of the visu-
alization are perpendicular to paleos-
lope. Similar techniques can evaluate
seismic constructive and destructive
interference patterns at unconformi-
ties and onlap and downlap relation-
ships at sequence boundaries.

Structural visualization. The primary


benefit of structural volume visuali-
zation is to quickly show subtle and
complex fault patterns. In
conventional fault mapping, the inter-
preter chooses what faults to map and
not to map, and how each fault is con-
nected or not connected. As a result,
the quality and thoroughness of the
task is highly variable. The goal of
structural volume visualization is to
reveal all the faults in the seismic
record.
Figure 19. Workflow for visualizing high-reflectivity amplitudes. (a) At the prospect level, subtle fault
Scanning and identifying the objective. (b) Applying opacity and (c) isolat- definition becomes critical. Detailed
ing the objective to determine (d) amplitude extent and structural relief. (e) surfaces are usually available and
Internal amplitude variations and (f) depositional indicators. utilized in horizon-keyed visualiza-

708 THE LEADING EDGE JUNE 1999 JUNE 1999 THE LEADING EDGE 0000
tions to define fault patterns and
extents. Prospect-level structural and
stratigraphic detail can be achieved
during regional reconnaissance visu-
alizations. Due to the wide variabil-
ity of fault orientation and dip, data
should be seen from many viewpoints
to determine optimal perspectives for
faults and fault blocks.
Faults can be delineated by using
their seismic signatures in conjunc-
tion with opacity. The mechanics
behind the application of ZS to visu-
alization of fault patterns depend on
several factors. In reflectivity vol-
umes, faults are normally picked at
abrupt amplitude terminations with
associated offset. The overall lateral
amplitude change at the fault is usu-
ally represented by a decrease in
amplitude, often to Z3 amplitude lev-
els (Figure 16). By lowering the opac-
ity of Z3 amplitudes to transparent
or semitransparent levels, the fault
will appear as physical gaps within
the horizon (Figure 17). To increase Figure 20. Application of ZS on a detected amplitude. (a) Amplitude extent
the visibility of the fault gaps, a dark and relief, (b) internal amplitude variations, and (c) enhanced depositional
background projects through the indications.
brightly colored surface. Similar tech-
niques can visualize fracture zones tify and evaluate an amplitude. In whats in the 3-D in one working
and carbonate reefs. Figure 19b, Z1 amplitudes are day.
The fault patterns of a high- revealed and a particular amplitude Because of the speed and advan-
amplitude (Z1) low-relief reflector selected. Subvolume detection iso- tages of volume visualization, it is
(Figure 16) are visualized by trim- lates the amplitude (Figure 19c) recommended that geoscientists
ming the volume along the time axis where extent, structural relief, high- review as many drilled prospects as
and applying opacity with particular amplitude sweet spots, and possible possible and become familiar with
attention to the weak Z3 amplitudes. stratigraphic information are deter- the visualization characteristics and
The volume visualization of the mined. This method is useful when parameters required to see what was
objective interval, an 84-ms slab, high reflectivity is associated with encountered. The goal is to build a
shows fault extents and interrela- hydrocarbons and internal amplitude visualization knowledge, skill, and
tionships (Figure 18). Since the visu- variations showing stratigraphic and expertise base as quickly as possible.
alization is a volume of data, 3-D structural details are critical. Figure However, application of conven-
perspectives of the interval provide 20 illustrates the opacity curves used tional interpretation philosophies
additional insight. to reveal structural, amplitude, and and workflows to volume visualiza-
stratigraphic details within the objec- tion can lead to pitfalls and substan-
High-reflectivity visualization. tive. dard visualizations. A thorough
Visualization of high reflectivity (Z1) understanding of visualization and
amplitudes is easy with high-cut Tips and pitfalls. During the initial the results of its various processes
opacity filters. Because Z1 ampli- stages of volume visualization, many must be kept in mind during all
tudes constitute a small portion (a significant conclusions may be stages of visualization. The most
few percent in many cases) of a 3-D accomplished in a few hours. Early common pitfalls are misinterpreta-
data volume, the number of compu- consensus on key structural, strati- tions of objects in 3-D space due to
tations is greatly reduced and the graphic, and prospect issues can be the visual alignment of unrelated
visualization can be extremely fast. achieved, especially in an immersive events, false truncation of events due
To visually integrate high ampli- environment. Leads can be identi- to the physical limits of 3-D volume,
tudes with surrounding geology, fied and prioritized and key hori- and misuse of opacity. One of the
opacity curves should be designed zons selected for mapping. During best quality control checks is simply
to include Z2 and Z3 amplitudes. Z1 the digitizing process, volume visu- checking interpretations and ideas
visualization difficulties can arise alization usually assists in the task with key 2-D lines. Pitfalls related to
when high-amplitude reflectors are and may resolve interpretation prob- the opacity editor, decimation, flat
included and/or numerous ampli- lems. Because a large amount of events, seismic phase, improper
tudes confuse the view. In these cases, information is revealed in a short dynamic range, and amplitude eval-
horizon- or sequence-keyed volume period of time, it is recommended uations need detailed discussion and
visualization, subvolume detections, that the work be documented as it such articles are planned for the near
or specialized amplitude extraction proceeds. Experienced volume visu- future. LE
methods should be used. alization geoscientists can easily have
Figure 19 is a workflow to iden- a fairly detailed presentation of (Continued on p. 712)

0000 THE LEADING EDGE JUNE 1999 JUNE 1999 THE LEADING EDGE 709
Acknowledgements: Thanks to James L. Allen
of Mayne & Mertz for valuable suggestions and
editing, to thank Kimberly Madsen-Cline of
Paradigm Geophysical for creative contribu-
tions toward image enhancements and visual-
ization techniques, and to Amy Jacobsen for
tireless efforts in preparing text and figures for
publication. I also thank Veritas DGC and the
Bureau of Economic Geology for the use of
their 3-D data. I am very thankful for the ener-
getic support and guidance of Oz Yilmaz, also
of Paradigm Geophysical.

Corresponding author: G. Kidd,


gkidd@paradigmgeo.com

Gerald D. Kidd attended Wright State


University where he earned a bachelors degree
in geology (1978) and a masters in geophysics
(1980). He is currently a 3-D visualization
geophysicist at Paradigm Geophysical
Corporation. He has more than 17 years of
interpretation experience, including six years
in the application of volume visualization on
numerous 3-D data sets in both exploration and
production environments around the world.

Das könnte Ihnen auch gefallen