This document discusses new techniques for reducing exploration risk in challenging subsurface environments like deep water and beneath salt or basalt. It describes how 3D seismic surveys have improved drilling success rates from 25% to 50% but success is still low in some areas like 10% in deep Gulf of Mexico wells. New methods of acquiring seismic data from multiple azimuths and with vertically aligned sources and receivers are providing higher quality images to further reduce risk by illuminating subsurface targets from more directions. Examples from various basins demonstrate how these improvements in seismic technology decrease the chance of drilling dry holes.

1. 26 Oilfield Review
Reducing Exploration Risk
Jose Camara Alfaro
Pemex
Tampico, Mexico
Chris Corcoran
Shell Exploration and Production
Houston, Texas, USA
Kevin Davies
Chevron
London, England
Francisco Gonzalez Pineda
Pemex
Reynosa, Mexico
Gary Hampson
Chevron
San Ramon, California, USA
David Hill
Gatwick, England
Mike Howard
BHP Billiton
Houston, Texas
Jerry Kapoor
Nick Moldoveanu
Houston, Texas
Ed Kragh
Cambridge, England
For help in preparation of this article, thanks to Robert Balaguer
and Mario Kieling, Rio de Janeiro; Andy Coutts, Patricia
Marçolla, Raúl Teran and David Wilson, Houston; Alberto
De Anda, Mexico City; Richard Harding, BHP Billiton,
Houston; and Richard Salter, Kuala Lumpur.
Q-Fin and Q-Marine are marks of Schlumberger.
With energy demand rising and production from mature fields on the decline, oil and
gas companies are expanding their exploration activities into increasingly challenging
areas—deep water, beneath salt and basalt, and in carbonate reservoirs. These
environments often yield murky seismic images, but innovative marine-seismic
technology can now provide high-quality results to reduce risk in these settings.

2. Spring 2007 27
Exploring for oil and gas is a risky business. By
increasing the likelihood of drilling success,
3D seismic surveys have probably done more
than any other modern technology to mitigate
this reality. In the 1970s and 1980s, before the
use of 3D surveys, the exploration drilling
success rate in the USA was approximately 25%.
Once E&P companies started using 3D surveys
widely, the success rate for exploration wells
increased to almost 50% by 2005, and for
development wells the rate reached 88%.1
While
advances in other technologies such as drilling,
LWD, visualization and real-time data delivery
have also contributed to these success rates,
explorationists credit 3D seismic methods with
having the greatest impact.
While three-dimensional seismic applications
have led to an improvement in overall explora-
tion drilling success, in some situations, the
success rate is still low. For example, in 2006,
of 119 Gulf of Mexico exploration wells drilled
in deep water—water deeper than 1,000 ft
[300 m]—only 11 hit pay.2
This 10% success rate
is typical of deepwater exploratory drilling in the
Gulf of Mexico in the last decade.3
With
deepwater wells costing up to US$ 100 million, it
is not surprising that oil and gas companies
are looking for ways to reduce the frequency of
dry holes.
Many companies consider the current level of
risk unacceptable. From 1996 to 2000, oil and gas
companies bought 3,000 leases in the Gulf of
Mexico. Of these, only 8% have been drilled.4
Because high rig demand will keep drilling costs
from declining in the near future, operators
await additional advancements in technology to
increase their likelihood of success.
The 3D seismic surveys that did so much to
improve drilling success rates on land and in
shallow water are not always adequate for
exploration in deep water and other problematic
areas, such as beneath hard seafloors, salt, basalt
and carbonate layers. Complex geology and
highly refractive layers cause ray bending that
leaves portions of the subsurface untouched by
seismic waves. Also, noise from near-surface
reflectors can mask the weak signals returning
from deep formations.
Deepwater subsalt prospects have been
particularly difficult to image properly. While
several fields have been discovered in the Gulf of
Mexico in the past decade using 3D seismic
technology—Atlantis, Mad Dog, Neptune, Puma,
Shenzi and Tahiti are a few that together hold
many billions of barrels of oil—only a small part
of the total resource is under development
(above). In some cases, the quality of seismic
data may have served exploration purposes, but
it may not be good enough to create accurate
models for reservoir development.
Recent improvements in seismic data
acquisition and analysis may be the answer to
achieving seismic images that are good enough to
reduce the risk of drilling wells in these difficult
areas. This article explains how new practices in
survey acquisition and data analysis are
improving the information gained from 3D
marine seismic surveys. We describe innovations
in seismic illumination made possible by probing
seismic targets from several angles, and discuss
new vertically aligned configurations of sources
and receivers that are increasing signal quality in
hard-to-image areas. Advances in imaging have
also helped reposition development wells in an
offshore carbonate reservoir and in a deepwater
heavy-oil field. Examples from the Gulf of
Mexico, West of Shetlands, and offshore Mexico
show how improvements in seismic technology
are reducing drilling risk.
Typical Marine Seismic Surveys
A typical 3D marine seismic survey is acquired by
a ship towing airgun seismic sources and
streamers, or cables instrumented with
receivers. The ship “sails” in a predetermined
direction above a subsurface target, the airguns
emit seismic energy, and the receivers record
signals that propagate from the sources down to
1. Crude Oil and Natural Gas Exploratory Wells, Selected
Years 1949–2005, http://www.eia.doe.gov/emeu/aer/
pdf/pages/sec4_13.pdf (accessed February 25, 2007).
Crude Oil and Natural Gas Exploratory and Development
Wells, Selected Years 1949–2005, http://www.eia.doe.
gov/emeu/aer/pdf/pages/sec4_11.pdf (accessed
February 25, 2007).
Farris A: “Guest Editorial: Emerging Technology and Its
Impact on the Industry,” JPT Online (October 2006),
http://www.spe.org/spe/jpt/jsp/jptmonthlysection/
0,2440,1104_11038_5742058_5742529,00.html (accessed
February 26, 2007).
2. Paganie D: “Deepwater GoM Key with 50 Bboe of Undis-
covered Reserves,” Offshore 67, no. 1 (January 2007): 46,
http://www.offshore-mag.com/display_article/283060/9/
ARCHI/none/none/Deepwater-GoM-key-with-50-Bboe-
of-undiscovered-reserves/ (accessed March 29, 2007).
3. “Deepwater Gulf of Mexico 2006: Interim Report of 2005
Highlights,” U.S. Department of the Interior, Minerals
Management Service, Gulf of Mexico OCS Region,
http://www.gomr.mms.gov/homepg/whatsnew/techann/
2006/2006-022.pdf (accessed February 27, 2007).
“Deepwater Gulf of Mexico 2005: Interim Report of 2004
Highlights,” U.S. Department of the Interior, Minerals
Management Service, Gulf of Mexico OCS Region,
http://www.gomr.mms.gov/homepg/whatsnew/techann/
2005/2005-023.html (accessed February 27, 2007).
4. Paganie, reference 2.
> Recent subsalt discoveries in the Gulf of Mexico, containing billions of
barrels of oil.
G U L F O F M E X I C O
New Orleans
Houston
S A L T
km
miles
150
150
0
0
Tahiti
Puma
Mad Dog
Shenzi
Atlantis
Neptune
G r e e n C a n y o n

3. subsurface reflectors and back. The process is
repeated at defined spatial increments until the
requisite number of seismic traces has been
recorded (below).
The current standards in seismic acquisition
have evolved from towing a single streamer in the
1970s and waiting 10 minutes between dynamite
shotpoints to the highly efficient operations of
modern surveys.5
Now, vessels tow eight to ten
streamers at a time, separated by 50 to 150 m
[160 to 490 ft]. Each streamer may be from 6 to
8 km [4 to 5 miles] long. The source consists of
an array of 12 to 18 airguns, and may be fired
every 10 to 20 seconds. These general values
apply for many 3D surveys, but the exact
acquisition parameters will vary depending on
the survey plan, which balances geophysical
objectives and economic constraints.6
The survey plan also specifies the depth at
which the sources and streamers should be towed
to minimize noise and maximize signal. Towing
streamers at a shallow depth—less than 8 m
[26 ft]—better preserves the high-frequency
content of the signal, but affects the low-
frequency content and also increases the noise
from waves and weather. Deeper towing better
retains the low-frequency content, and so
increases depth of penetration, but at the expense
of high-frequency content. A typical towing depth
for streamers is less than 10 m [33 ft].
Similar trade-offs must be considered when
determining the tow depth of the source. Airgun
source arrays typically are towed at a depth of
5 to 10 m [16 to 33 ft], depending on the required
frequency bandwidth.
Another aspect of typical marine seismic
acquisition is the collection of signals from a
group of receivers into a single recorded trace.
Standard streamers carry hundreds of hydro-
phone receivers spaced 0.5 to 1 m [1.6 to 3.3 ft]
apart, with groups of 12 to 24 hydrophones feeding
to a single recording channel. In principle,
summing the traces in a group before recording
enhances signal-to-noise ratio. But this process
can irreparably damage signal fidelity and reduce
the effectiveness of noise-reduction processing.
28 Oilfield Review
> Acquisition of a typical 3D marine seismic survey. A vessel towing sources and receiver cables sails above the target in a direction specified by the
survey plan (bottom). The length of each loop includes additional distance to allow the streamer to straighten after every turn. The target is subdivided into
“bins,” or small areas that are treated as reflection points for the purpose of processing the data. Most marine seismic projects acquire data at a range of
azimuths depending on the offset, or source-receiver distance. Azimuth (top) is the angle at the source array between the sail line and the direction to a
given receiver. For a few receivers near the head of the streamers, offsets are short and azimuths are large, but because the length of the streamers is so
much greater than the streamer spread, most of the seismic rays travel at small azimuth, or nearly parallel to the sail line.
AzimuthTail
network
Front
network
8,000 m
Hydrophone
Gyro
Compass
Float
Streamer
Source
Bin

4. Spring 2007 29
The two main advantages of this arrangement
of sources and receivers are efficiency in
acquiring data over most subsurface targets—
the marine seismic industry has built itself
around this method—and the ability to obtain
adequate images through standard processing.
The disadvantages become clear when surveying
in problematic areas or under suboptimal
circumstances. Images may be faint or
impossible to interpret. Noise generated in the
streamers from weather and sea conditions can
contaminate the seismic record at all levels. To
make matters worse, platforms and other
structures can obstruct the path of survey
vessels, creating gaps in coverage that must be
filled by additional shots and survey time. And
close alignment between source and receiver
arrays limits azimuthal coverage, meaning the
reservoir is illuminated from only one direction.7
The Q-Marine single-sensor marine seismic
system, introduced in 2000, has overcome several
of these limitations.8
Instead of recording signals
from grouped receivers, the Q-Marine system
records signals from individual receivers. This
improves spatial sampling of both the noise and
the desired wavefield, leading to several advan-
tages over conventional acquisition. Q-Marine
surveys have, on average, broadened bandwidth
by 40% compared with grouped-receiver tech-
nology, thus increasing the resolution of seismic
images. Streamer noise can be adequately
sampled, allowing signal-processing techniques
to suppress it without harming signal bandwidth.
This allows high-quality seismic data to be
acquired even in poor weather, thereby reducing
weather-related downtime.
Additional enhancements, such as repeat-
able, calibrated sources, and the ability to
position streamers and steer them with high
accuracy and repeatability, have led to seismic
images with greater resolution. However, in some
areas, further improvements are required.
New Directions in Seismic Surveys
Most marine seismic surveys—including
Q-Marine surveys—acquire data along a narrow
azimuth, and so illuminate the target from
essentially one direction. If all subsurface layers
were flat and smooth, narrow-azimuth surveys
would deliver adequate images. However, in
areas of E&P interest, subsurface targets and
their overburden are rarely flat and smooth. The
disadvantage of narrow-azimuth illumination in
these cases can be shown by analogy. Shining a
beam of light on an irregular landscape produces
shadows behind mountains and within canyons.
But, if the light shines from a different direction,
areas that were in shadow become illuminated.
Although they are called narrow-azimuth
surveys, most marine surveys acquire seismic
data at a range of azimuths depending on the
offset, or source-receiver distance. For a few
short offsets, the azimuth can be quite large, but
for most offsets, azimuths fall within a narrow
range—about 10° on either side of the sail line.
Because the short-offset, large-azimuth traces
are not acquired in great number, they do not
contribute much to the image computed using
standard processing. Thus, a target covered by
such a survey is essentially illuminated by rays
from a narrow range of azimuths, causing poor
signal-to-noise ratio and suboptimal seismic
resolution. Most narrow-azimuth survey designs
attempt to compensate for the lack of azimuthal
range by redundant sampling of the same
subsurface point, or bin.9
Increasing the fold, or
the number of traces per bin, is one way to
increase signal-to-noise ratio.
A particular type of noise that plagues all
seismic surveys is called a “multiple.” Multiples
are reverberations between interfaces with high
acoustic-impedance contrasts, such as between
the sea surface and sea bottom, or between the
Earth’s surface and the bottom of a layer of
unconsolidated rock (above). They appear as
later arrivals on a seismic section, and thus are
easy to confuse with deep reflections. Because
multiples have velocities that can be slower than,
the same as, or faster than the desired signal,
and can have frequency content higher than that
of the desired signal, they are difficult to
suppress through filtering and stacking.10
Significant efforts in surface-related multiple
elimination (SRME) have led to processing
techniques that improve data quality, but
modeling has shown that even greater
improvement in multiple attenuation can be
realized by increasing azimuthal coverage.11
5. “Why Norwegian Seismic Is Still Running in Anders
Farestveit’s Direction,” First Break 24 (December 2006):
11–14.
6. Vermeer GJO: 3–D Seismic Survey Design. Tulsa: Society
of Exploration Geophysicists, 2002.
7. Azimuth is the bearing from the source to the receiver.
8. Christie P, Nichols D, Özbek A, Curtis T, Larsen L,
Strudley A, Davis R and Svendsen M: “Raising the
Standards of Seismic Data Quality,” Oilfield Review 13,
no. 2 (Summer 2001): 16–31.
9. A bin is the area on the target surface, typically 25 m by
25 m [82 ft by 82 ft], to which seismic traces are assigned
according to their common midpoint (CMP).
10. Stacking is the summation of signals, and is performed to
enhance signal-to-noise ratio.
11. Regone C: “Using 3D Finite-Difference Modeling to
Design Wide Azimuth Surveys for Improved Subsalt
Imaging,” Expanded Abstracts, 75th SEG Annual
International Meeting and Exposition, New Orleans
(October 1–6, 2006): 2896–2900.
> Multiple reflections, or multiples. After leaving the source, seismic energy
can reflect several times before arriving at the receiver. Unlike the others,
the ghost multiple travels upward before reflecting. Energy from multiples
adds noise to seismic recordings.
Primary
reflection Ghost
Near-surface
multiples
Long-path
multiple

5. The azimuthal coverage of a survey can be
increased in several ways. One way is to repeat a
standard survey at one or more azimuths,
creating a multiazimuth survey (above).
Multiazimuth surveys acquired in this way
increase fold as well as azimuthal coverage. One
such survey has been acquired over the Nile
Delta in the Mediterranean Sea.12
Wide-Azimuth Surveys
Another way to increase azimuthal coverage in a
wider swath above the target is to displace the
seismic source from the sail line of the streamer
vessel, in a wide-azimuth survey (WAZ). Wide-
azimuth surveys require at least two source
vessels in addition to the streamer vessel, and
some may be acquired with multiple streamer
vessels to improve acquisition efficiency.
The improvement that WAZ acquisition brings
to subsalt illumination can be seen in a modeling
example, in which a flat, horizontal target beneath
a complex salt structure is probed by both narrow-
and wide-azimuth surveys (next page). In the
narrow-azimuth case, the number of hits, or times
that seismic energy reaches a point on the target,
is everywhere lower than in the wide-azimuth
case. In both cases, some portions of the target
are not hit at all, but there are fewer of these
misses in the wide-azimuth case. In addition to
this basic ray-tracing method, finite-difference
and wave-equation modeling are also used in
illumination studies.
In 2001, in the Norwegian North Sea, BP and
Petroleum Geo-Services (PGS) tested the wide-
azimuth concept including high fold and using
multiple vessels; earlier surveys had acquired
data in multiple directions.13
The results of the
2001 survey showed improved noise attenuation
with increased azimuthal coverage. Since then,
the method has grown in complexity and has
seen practical application. BP conducted the
first Gulf of Mexico WAZ survey with seismic
contractor Veritas over the Mad Dog field in 2004
and 2005.14
In 2006, Shell Exploration acquired a WAZ
towed-streamer survey with WesternGeco over a
deep objective below a complex salt structure in
a deepwater area of the Gulf of Mexico.15
The
prospect is located in 3,800 ft [1,160 m] of water,
and the discovery well was drilled to a total depth
of 29,414 ft [8,965 m].
The survey design attempted to satisfy several
seemingly conflicting objectives: full illumina-
tion, optimal noise suppression, easy processing
and low cost. Achieving these required a wide
range of azimuths to illuminate targets under the
salt, relatively uniform sampling, minimization of
inefficiencies—such as over-redundancy in
sampling—and consideration of practical
limitations, such as the number of vessels
available for the survey and the number of
streamer passes.
Four parameters are critical for WAZ survey
design. First is the maximum crossline offset,
which is the greatest separation between source
and receiver in the crossline direction—
30 Oilfield Review
> Traditional and new acquisition geometries (bottom) and azimuth-offset distribution plots (top). One way to plot azimuth-offset distribution is as a “rose
diagram.” The number of traces recorded at a particular azimuth-offset pair is plotted by color, with offset corresponding to the distance from the center
and azimuth corresponding to the angle from the top of the circle. Colors range from purple and dark blue for a low number of traces, to green, yellow and
red for a high number of traces. From left to right: Traditional marine surveys are acquired in one azimuth and produce data with a narrow azimuth-offset
distribution. Multiazimuth surveys are acquired in multiple directions, and have azimuth-offset distributions clustered along azimuths associated with the
survey sail lines. Wide-azimuth surveys are acquired in a single direction, but with additional source vessels, increasing the azimuth for many offsets.
Rich-azimuth surveys combine the concepts and advantages of multiazimuth and wide-azimuth surveys. (Adapted from Kapoor et al, reference 21.)
Azimuth0
30
60
90
120
150
180
210
240
270
300
330
0
30
60
90
120
150
180
210
240
270
300
330
0
30
60
90
120
150
180
210
240
270
300
330
0
30
60
90
120
150
180
210
240
270
300
330
Narrow-Azimuth Multiazimuth Wide-Azimuth Rich-Azimuth

6. Spring 2007 31
perpendicular to the inline, or streamer,
direction. Second is the source-line interval,
which is the distance between adjacent source
lines. Third and fourth are survey size and
number of vessels. The maximum crossline offset
and the source-line interval are determined from
synthetic modeling and migration of synthetic
data.16
Illumination of the target and successful
attenuation of multiples are the main criteria for
assessing the modeling results, which must also
include deterioration of the image caused by
gaps created when the streamers avoid surface
obstructions, and feathering, or streamer
deviation caused by strong currents.
After evaluating several survey geometries,
Shell selected a program comprising two dual-
source vessels and one streamer vessel towing
eight receiver cables spaced 150 m apart, for an
effective streamer swath of 1,200 m [3,937 ft].17
One source vessel sailed alongside the streamer
vessel, but offset 100 m [328 ft] from the outside
streamer, while the second source vessel sailed
12. Keggin J, Benson M, Rietveld W, Manning T, Cook P
and Page C: “Multi-Azimuth 3D Provides Robust
Improvements in Nile Delta Seismic Imaging,” First
Break 25 (March 2007): 47–53.
Michell S, Summers T, Shoshitaishvili E, Etgen J,
Regone C, Barley B, Keggin J, Benson M, Reitveld W
and Manning T: “Multi-Azimuth and Wide Azimuth
Towed Streamer Acquisition for Subsalt Imaging in the
Gulf of Mexico and Egypt,” paper OTC 18829, presented
at the Offshore Technology Conference, Houston,
April 30–May 3, 2007.
13. Widmaier M, Keggin J, Hegna S and Kios E: “The Use
of Multi-Azimuth Streamer Acquisition for Attenuation
of Diffracted Multiples,” Expanded Abstracts, 72nd SEG
Annual International Meeting and Exposition, Salt Lake
City, Utah, USA (October 6–10, 2002): 89–93.
Houllevigue H, Delesalle H and de Bazelaire E:
“Enhanced Composite 3D Cube Derived from Multi-
azimuth 3D Marine Acquisition,” Expanded Abstracts,
61st European Association of Geoscientists and
Engineers Conference and Exhibition, Helsinki,
Finland (June 7–11, 1999).
French WS: "Circular Seismic Acquisition System," US
Patent No. 4,486,863 (December 4, 1984).
14. Threadgold IM, Zembeck-England K, Aas PG, Fontana
PM, Hite D and Boone WE: “Implementing a Wide
Azimuth Towed Streamer Field Trial: The What, Why
and Mostly How of WATS in Southern Green Canyon,”
Expanded Abstracts, 76th SEG Annual International
Meeting and Exposition, New Orleans, (October 1–6,
2006): 2901–2904.
15. Corcoran C, Perkins C, Lee D, Cook R, Kapoor J and
Moldoveanu N: “Wide-Azimuth Streamer Acquisition for
Gulf of Mexico Subsalt Imaging,” paper OTC 19071,
presented at the Offshore Technology Conference,
Houston, April 30–May 3, 2007.
Corcoran C, Perkins C, Lee D, Cattermole P, Cook R and
Moldoveanu N: “A Wide-Azimuth Streamer Acquisition
Pilot Project in the Gulf of Mexico,” The Leading
Edge 26, no. 4 (April 2007): 460-468.
16. Corcoran C, Perkins C, Lee D, Cattermole P, Cook R and
Moldoveanu N: “Wide-Azimuth Streamer Acquisition for
Gulf of Mexico Subsalt Imaging,” Expanded Abstracts,
76th SEG Annual International Meeting and Exposition,
New Orleans (October 1–6, 2006): 2910–2914.
17. Corcoran et al, reference 16.
> Effect of narrow-azimuth and wide-azimuth acquisition on subsalt illumination. A complex but
realistic salt body (top), defined by its top (gold) and bottom (pink), overlies a target horizon (purple).
Salt overhangs are in green. Shotpoints for both surveys are in the black rectangle. The narrow-
azimuth acquisition hit map (middle) shows the number of traces that reach the target horizon. The
wide-azimuth hit map (bottom) shows more seismic traces reaching the target horizon, with fewer
areas of no illumination (white).
Shotpoints
Depth,km
0
8
16
0
10
20
30
40
0
20
40
60
80
100
X,km
Y,km
Numberofhits
0 625 1,250 1,875 2,500 3,125 3,750 4,375 5,000
Y,km
0 10 20 30 40 50 60 70 80 90 100
X,km
40
45
35
30
25
20
15
10
5
0
Y,km
0 10 20 30 40 50 60 70 80 90 100
X,km
40
45
35
30
25
20
15
10
5
0

7. behind the streamers, offset from the first source
vessel by 450 m [1,476 ft] (above). Then, while
the source vessels repeated the same source
lines, the streamer vessel performed additional
passes at 1,400-m and 2,600-m [4,593-ft and
8,530-ft] offsets from the original source line,
and repeated those offsets again on the other
side of the source line. Each source line was
acquired six times. After this coverage for one
source line was completed, the entire process
shifted 900 m [2,952 ft] to the next source line,
and was repeated. The azimuth-offset plot for
this wide-azimuth survey displays a much
broader distribution than for a narrow-azimuth
survey (next page, top).
Data from this wide-azimuth pilot survey
underwent initial processing, consisting of only
coherent-noise attenuation and common shot-
gather migration using a preexisting velocity
model.18
This dataset, with only basic processing
applied, was compared with more fully processed
WesternGeco narrow-azimuth data from a
multiclient survey covering the same area (next
page, bottom).19
The narrow-azimuth survey
processing included coherent-noise attenuation,
SRME multiple suppression and the same wave-
equation migration as the WAZ survey. Even
without additional multiple suppression, the
WAZ survey produced a clearer subsalt picture
than the narrow-azimuth survey, leading to more
confident assessment that the reflections were
coming from subsalt layers rather than multiples.
The improvement in clarity is most obvious on
the left side of the section, where sediments
truncate against the salt keel in the middle of the
section. The WAZ data convinced interpreters
that significant subsalt events continue updip
toward the salt keel, and that these events are
not multiples.20
Shell concluded that while not all areas
below the salt were illuminated, the WAZ survey
improved the image of subsalt sedimentary
structure in most places. The dominant multiples
were removed without specific processing.
Depopulation tests to see if adequate results
could be achieved with less data showed that in
32 Oilfield Review
18. Wave-equation migration uses a velocity model and
various solutions to the wave equation that describe
propagation of waves through rock. The objective is
to redistribute reflected seismic energy from its
assumed position at the source-receiver midpoint to
its true position.
19. Multiclient surveys are surveys acquired by seismic
contractors over unlicensed areas, then made available
to multiple clients. Partial funding for multiclient surveys
may be provided by one or more E&P companies.
20. A salt keel is what remains of the root of an intruding
salt body after the root has detached itself from the
parent salt layer.
> Acquisition of the Shell wide-azimuth survey. With two dual-source vessels and one streamer vessel towing eight receiver cables, six passes were
made at varying offsets for each source line (top). For the first pass, the forward source vessel sailed along the source line, while the streamer vessel
sailed with the nearest streamer 100 m away. The rear source vessel sailed 450 m from the source line. For the second pass, the source vessels repeated
the same source line, and the streamer vessel performed another pass at a 1,400-m offset from the original source line, and made a third pass at a 2,600-m
offset. The vessels then repeated the pattern again on the other side of the source line (Pass 4, Pass 5 and Pass 6). The combination of those six runs
constituted the acquisition of one source line. After acquisition of that source line, the vessels moved to the next source line, offset by 900 m (bottom left).
Pass 1
Pass 2
Pass 3
Pass 4
Pass 5
Pass 6
Source line 1
Pass 1
Pass 2
Pass 3
Pass 4
Pass 5
Pass 6
Source line 2
Source line 1
Pass 1 Pass 2 Pass 3
Pass 4 Pass 6Pass 5

8. Spring 2007 33
the deep subsalt areas, images acquired by
removing every other source line were almost as
good as the full dataset. These tests also
indicated that data acquired from sources in
front of the streamers provided better subsalt
imaging than data acquired from the sources
behind the vessel, where streamer feathering has
greater negative impact. This information could
help in the design of future WAZ surveys.
The two-phase wide-azimuth study was
designed to demonstrate the value of WAZ
technology in exploring complex subsalt settings.
Following the successful imaging results from
the first phase, the second phase was cancelled
and Shell Exploration moved to underwrite
acquisition of a large multiclient WAZ explora-
tion survey in the Gulf of Mexico.
> The azimuth-offset distribution (left) and survey parameters (right) for the Shell wide-azimuth survey. The survey acquired a
broader range of azimuths compared with a narrow-azimuth survey.
Numberoftraces,thousands 2,643
3,161
2,074
1,297
676
55
Offset,m
Offset, m
Azimuth
0
5,000
10,000
–5,000
–10,000
–10,000 –5,000 0 5,000 10,000
0
30
60
90
120
150
180
210
240
270
300
330
Shot interval
Shot sampling
Streamer separation
Streamer length
Number of streamers per vessel
Receiver interval
Source-line interval
Maximum inline offset
Maximum crossline offset
8
9,000 m
150 m
12.5 m
2
2
37.5 m
900 m
150 m x 450 m
9,300 m
4,150 m
6
0.44
Survey Acquisition Parameters
Number of sources on the front
source vessel
Number of sources on the back
source vessel
Number of passes
Aspect ratio
> Comparison between images acquired with narrow-azimuth (left) and wide-azimuth (right) methods. The narrow-azimuth data were fully processed
with noise-attenuation, multiple-suppression and migration steps. The wide-azimuth data were only migrated, and with the same velocity model as the
narrow-azimuth data. The wide-azimuth image shows clearer subsalt reflections, especially on the left side of the section. [Adapted from Corcoran et al,
reference 15 (April 2007).]
Salt keel
Bottom of saltBottom of salt

9. Using knowledge gained from modeling and
earlier WAZ surveys, WesternGeco has designed
and acquired two large WAZ surveys for
multiclient use (left). These surveys incorporated
design improvements such as crossline
reciprocity—sail lines in opposing directions—
and source vessels at the forward rather than aft
end of the streamers.21
They also use wider
crossline offsets for better signal-to-noise ratio,
multiple attenuation and illumination.
The first survey, completed in 2006, covered
222 Gulf of Mexico blocks, or 5,183 km2
[2,002 mi2], and deployed three single-source
vessels. Ten 7,000-m [22,967-ft] streamers were
used in a 1,200-m [3,936-ft] spread width, and
sail lines were run in reciprocal directions. The
second survey, completed in 2007, covered
252 blocks, or 5,895 km2 [2,278 mi2], and
deployed four single-source vessels, including
single sources on two streamer vessels
interleaving in two opposing directions.
The first multiclient WAZ survey provides
improved images compared with narrow-azimuth
surveys acquired over the same area. With
minimal processing, the new images show
enhanced clarity and illumination, especially in
areas beneath salt overhangs (below).
The third and fourth phases of the multiclient
Gulf of Mexico WAZ project will cover more than
10,000 km2 [4,000 mi2] starting in May, 2007. The
goal is to produce is a quick-look wave-equation
migrated volume through onboard processing,
allowing customers to meet lease sales deadlines.
34 Oilfield Review
> Two large multiclient wide-azimuth surveys acquired by WesternGeco in the Gulf of Mexico,
incorporating lessons learned from earlier surveys. The Phase 1 survey, completed in 2006, was
acquired with two source vessels and one streamer vessel (top). Phase 2, completed in 2007, used
two source vessels and two streamer vessels (bottom). In both Phase 1 and 2, each streamer vessel
also also carried a single source.
G U L F O F M E X I C O
New Orleans
Houston
Phase 1: July to December 2006
Phase 2: December 2006 to April 2007
Phase 2
Phase 1
Garden Banks
Green Canyon
Keathley Canyon
Walker Ridge
> Comparison between the first multiclient wide-azimuth survey and a narrow-azimuth survey. The narrow-azimuth survey (left) has had full processing
with multiple suppression, while the multiazimuth survey (right) has had only basic migration processing. The multiazimuth image shows better illumination
below salt overhangs and in deep sediments.
Wide-Azimuth, Basic Processing
Salt Salt
Narrow-Azimuth, Full Processing
Salt Salt

10. Spring 2007 35
Rich-Azimuth Surveys
Azimuthal coverage can be further enhanced by
combining the multiazimuth and wide-azimuth
concepts in what is called a rich-azimuth survey
(RAZ). BHP Billiton first implemented this type
of acquisition in 2006 over the Shenzi field in the
Gulf of Mexico, a 2002 discovery in water depths
up to 4,300 ft [1,300 m] approximately 120 miles
[190 km] from the coast of Louisiana (right).22
The field comprises Green Canyon Blocks 609,
610, 653 and 654. Development drilling will
assess recoverable reserves, which are estimated
at 350 to 400 million bbl [56 to 64 million m3].23
Coventurers in the project are Hess Corporation
and Repsol YPF.
The Shenzi discovery well was drilled to a
target identified by 3D seismic data, but the first
appraisal well encountered subsurface layers at
unexpected depths. New seismic data with
higher signal-to-noise ratio and resolution were
required to build a more reliable subsurface
model that would reduce the risk of future
drilling. A survey design and evaluation study
helped BHP Billiton and WesternGeco
geophysicists conclude that a rich-azimuth
survey was the best choice for the Shenzi
reservoir. Important factors in this decision were
the reservoir structural complexity, strong
currents, rig activity and platforms.
The RAZ survey, the world’s first, was
acquired with one streamer vessel equipped with
a source and two source vessels shooting along
three azimuths, 30°, 90° and 150° (below). The
survey plan allowed each shot location to be
repeated at least three times.
In another world’s first, the streamer vessel
continued recording during tight vessel turns.
Shooting and recording while turning improve
21. Kapoor S, O’Briain M, Desta D, Atakishiyev I and
Tomida M: “Subsalt Imaging—The RAZ/WAZ
Experience,” prepared for presentation at the 77th SEG
Annual International Meeting and Exposition,
San Antonio, Texas, September 23–28, 2007.
22. Howard MS and Moldoveanu N: “Marine Survey Design
for Rich-Azimuth Seismic Using Surface Streamers,”
Expanded Abstracts, 76th SEG Annual International
Meeting and Exposition, New Orleans (October 1–6,
2006): 2915–2919.
23. BHP Billiton to Develop Shenzi Oil and Gas Field in Gulf
of Mexico, http://www.bhpbilliton.com/bb/investorsMedia/
news/2006/bhpBillitonToDevelopShenziOilAndGasFieldIn
GulfOfMexico.jsp (accessed March 2, 2007).
> Location of the Shenzi field survey (left) and acquisition parameters for the
rich-azimuth survey (right).
Shenzi
field
New Orleans
km
miles
150
150
0
0
G U L F O F M E X I C O
Streamer depth
Streamer separation
Streamer length
Number of streamers
Source depth
Shot interval
Shot sampling
Vessels
10
7,000 m
120 m
12 m
10 m
38.5 m
115.5 m x 600 m
1 streamer vessel, 2
source vessels
Shenzi Survey Acquisition Parameters
> Survey plan for the world’s first rich-azimuth survey, acquired over the Shenzi field (left). The sail lines and platform obstructions (pink dots) are drawn
to illustrate the general concept and differ from the actual survey. The violet square represents source coverage from E-W lines, and the black hexagon
represents source coverage from three directions. The survey acquired data with a rich azimuth-offset distribution (right).
Numberoftraces,thousands
1,747
2,154
1,320
855
429
41
Azimuth-Offset Distribution
Offset,ft
Offset, ft
Azimuth
0
5,000
10,000
–5,000
–10,000
–10,000 –5,000 0 5,000 10,000
0
30
60
90
120
150
180
210
240
270
300
330
Shenzi Survey Sail Lines
Yosemite
Mighty
Joe Young
Mad Dog
Shenzi
Marco Polo
Atlantis
K2/Timon

11. operational efficiency and reduce nonproductive
time. Typically, vessels extend their sail lines and
turn well outside the boundary of the survey, so
that the streamers are straight and at the proper
separation by the time the vessel reenters the
boundary. This practice adds time to the survey
acquisition—an extra two hours per turn
(below). Usually, no data are acquired during the
turn, because the streamers do not track cleanly
around the turn and the receiver positions are
not accurately calculated. Also, towing through a
curve causes too much noise for useable data to
be acquired.
With the Q-Marine system, streamers can be
positioned and steered so that they maintain
spacing throughout the turn. Also, recording data
from individual hydrophones, instead of grouping
before recording, allows removal of noise from
turning, which resembles bad-weather noise.
These features promote acquisition of valuable
data during periods that would normally be
nonproductive. Acquiring additional data on the
edges of the Shenzi survey increased the area
that could be imaged effectively and improved
the images of the target near the boundary of the
survey area.
Comparison to a fully processed conventional
narrow-azimuth survey showed that the
Q-Marine rich-azimuth survey with basic
processing reduced noise artifacts and produced
clearer illumination of the base of the salt, and
subsalt reflections, than the fully processed
narrow-azimuth survey (next page, top).24
Reflections could even be identified within the
salt. However, the greatest improvement was in
reflections deep below the salt. BHP Billiton
geophysicists are now able to interpret
structures in zones that are better illuminated.
According to the survey plan, the RAZ survey
would have taken 105 days, compared with a
conventional survey that would have taken
72 days to cover the same area. As shot, the
Shenzi RAZ survey took just 88 days, plus 12 days
in mobilization, and delivered six times the data
of a conventional survey. The time savings arose
from several factors. Shooting in three directions
with the wide-azimuth acquisition made it
unnecessary to undershoot obstacles, and
minimized the time required to shoot infill lines.
The vessel crews experienced almost no weather
downtime, because the streamer cables were
towed deep, at 12 m [39 ft], and the Q-Marine
single-sensor recording adequately sampled any
weather-related noise. Following the success of
the first RAZ survey, BHP Billiton is preparing to
acquire another.
WesternGeco experience in WAZ and RAZ
surveys demonstrated several benefits of wide-
azimuth and rich-azimuth acquisition. Improved
attenuation of multiples and other coherent
noise provides increased signal-to-noise ratio of
subsurface events. Raypaths from different
azimuths result in better reservoir illumination,
and repeatability of shots in the same location
makes migration processing highly efficient.
Additionally, acquiring data while turning
increases efficiency, and is a practice that
WesternGeco expects to extend to conventional
3D Q-Marine surveys.
36 Oilfield Review
> Shooting and recording while turning, for improved operational efficiency. In traditional surveys
(top left), each sail line (dashed line) is extended to make sure the streamers are straight for all
shots. This additional line length at the beginning and end of each turn adds hours to the turning
time. The Q-Marine system allows streamers to stay in position throughout the turn (green line), so
data acquisition during turning is possible. The quality of data acquired while turning is comparable
to that acquired along sail lines (top right). Data acquired in the turns contribute to the overall image
of subsurface reflectors and fit smoothly with data acquired by straight streamers (bottom).
1,500m-radius
Finish shooting
Start shooting
Line Change Analysis
for Single-Vessel Option
Race Track View
Straight StraightTurning

12. Spring 2007 37
Towing-Depth Trade-Offs
A completely different approach to improving
seismic signal involves deploying sources and
streamers differently to more fully sample both
the noise and desired energy fields. As outlined
earlier, acquiring marine seismic data typically
involves towing sources and receiver cables at
constant depths specified by the survey design.
Selecting the depths calls for consideration of
several factors. One of these is the limiting effect
of the source and receiver depths on the
frequency content of the seismic energy that can
be recorded. Understanding how source and
receiver geometry affects signal-to-noise level
and frequency content requires examining
the way energy propagates from the towed
source to the receiver cable, and especially how
near-surface multiples add noise and reduce
signal (right).
Energy radiates from the seismic source in all
directions—not only downward but also upward.
The negative acoustic-impedance contrast at the
sea surface causes the upgoing wavefield to
reflect downward with a reversed polarity.25
Because of this polarity reversal, the wave
reflected at the sea surface, called the source
ghost, and the direct, downgoing, wave
destructively interfere to the extent that at
certain frequencies, the amplitudes cancel. The
result is a downward propagating wavefield that
is deficient in energy at certain frequencies. The
frequencies at which these deficiencies occur—
at which amplitude falls to zero—are called the
source-ghost notches, and are related to the
depth of the source. The greater the depth of the
source, the lower the first, nonzero source-ghost
notch frequency.
24. Howard M: “Marine Seismic Surveys with Enhanced
Azimuth Coverage: Lessons in Survey Design and
Acquisition,” The Leading Edge 26, no. 4 (April 2007):
480–493.
25. Polarity of a seismic signal refers to the direction of its
amplitude in a recorded seismic trace. Negative and
positive amplitudes have opposite polarity.
> An image from the Shenzi rich-azimuth survey, showing enhanced clarity of subsalt events relative to a more fully processed narrow-azimuth survey. The
narrow-azimuth survey (left) shows some indications of dipping layers beneath the salt, but the rich-azimuth image (right) illuminates subsalt layers clearly.
The survey configurations for each survey are adjacent to the seismic images. (Adapted from Howard, reference 24.)
Narrow-Azimuth, Full Processing Rich-Azimuth, Basic Processing
Bottom of salt Bottom of salt
> Ghost multiples, or reflections at the underside of the sea surface. Energy
traveling upward from the source reflects at the sea surface with a polarity
change, and interferes destructively with energy emanating downward
from the source. Similarly, reflected energy traveling upward from the
subsurface can reflect at the sea surface before being recorded at the
streamer cable as a receiver ghost. These reflections also interfere
destructively with upgoing reflected energy, causing recorded amplitude
to fall to zero at certain frequencies that depend on the depths of source
and receiver.
Seafloor
SourceSource
ghost
Receiver
ghost
Receiver cable

13. Similarly, energy returning from a deep
reflector travels upward toward the receiver
cable and also reflects off the sea surface,
interfering destructively with energy traveling
directly from the deep reflector to the receiver.
This interference causes cable-ghost notches
at several frequencies (middle right). The
combination of the source- and cable-ghost
notches leads to a complex amplitude spectrum
that produces gaps in the available seismic
signal. The maximum frequency used in
processing is usually limited at the frequency
corresponding to the first receiver notch, and the
minimum frequency is limited by the acquisition
filter, which is typically 3 Hz.
Because the objective is to acquire data of the
greatest bandwidth and highest frequency
possible, the source and streamer are usually
towed at shallow depth, pushing the first,
nonzero notch frequency higher. However,
shallow towing has disadvantages. Wind, waves
and shallow currents add noise to the acquired
signal, and the need to keep noise to a minimum
can restrict acquisition to seasons with good
weather. Also, shallow towing attenuates the
extremely low frequencies needed for inversion,
a processing technique that extracts acoustic
impedance and other rock properties from
seismic data.26
Another disadvantage to the lack of low
frequencies is the diminished probing power of
seismic waves. Because depth of penetration is
related to seismic wavelength, the lack of low
frequencies limits the depth to which seismic
energy propagates.
Increasing the towing depth allows data
acquisition in a quieter environment, extends
the acquisition season and extends the band-
width to lower frequencies. This allows imaging
at greater depth, but decreases the maximum
frequency of data that can be recorded, and so
decreases the overall image resolution.
Until recently, survey planners had to
sacrifice resolution and high frequencies for
depth of penetration and a quiet acquisition
environment, or vice versa. They were forced to
choose between towing shallow, deep, or at
some intermediate depth. However, a reemerging
acquisition technology called “over/under”
allows companies to enjoy the advantages of
both shallow and deep towing without
the drawbacks.27
Over and Under
The idea of towing one shallow streamer along
with one deep streamer—one over and one
under—has been around for more than
50 years.28
Practical application of the method
was attempted in the 1980s to reduce weather-
related noise and downtime.29
A key benefit of
the over/under method is the increased band-
width obtained from the ability to fill the nonzero
ghost notches through combination of the over
and under streamers, while still retaining the
low-frequency benefit of deep towed streamers.
Also, in theory, the vertical separation of two
linear arrays of receivers allows the upgoing
seismic wavefield to be separated from the
downgoing waves, facilitating the suppression of
surface-related multiples.30
38 Oilfield Review
> Modeling the effect of streamer depth on frequency content of the
recorded signal. Towing streamers at a shallow depth, 8 m (red dashed line),
produces signals up to 93 Hz before the first nonzero-frequency notch cuts
the signal amplitude to zero. Towing streamers deep, at 30 m (green),
produces signals up to 25 Hz before the first notch cuts signal amplitude to
zero. (Adapted from Moldoveanu et al, reference 27.)
30-m streamer depth
8-m streamer depth
Amplitude,dB
Frequency, Hz
93 Hz
1000 25 50 75 125
Depth,m
0
5
10
15
20
25
30
360 m
> Over/under streamer configuration for the Chevron 2D survey in
the Gulf of Mexico. An over/under pair of streamers (darker blue)
was towed at 18 and 25.2 m, along with position-monitoring
streamers at either side at 18-m depth, and an experimental-control
streamer at 7.2 m. Sources were deployed at 5 and 10 m.

14. Spring 2007 39
However, the success of this acquisition
geometry in removing the ghost notches from the
seismic data relies on keeping one streamer
directly below the other, in one vertical plane all
along the streamer length. Achieving this
configuration was not feasible with the towed-
streamer technology of the 1980s, and so the idea
was not put into practice for many years. In spite
of these practical limitations, scientists continued
to develop wavefield-separation techniques for
eventual application.31
Recently, acquisition technology caught up
with processing advances. The Q-Marine system
is capable of steering pairs of streamers in a
vertical plane with sufficient accuracy for the
over/under method to succeed. To demonstrate
the feasibility and promise of the method,
Chevron and WesternGeco performed a 2D
experiment over the Genesis field in the Gulf of
Mexico in 2004.32
The acquisition plan for the 2D line called for
streamers to be towed at 18 and 25.2 m [59 and
82.7 ft]. However, to accurately calculate
receiver positions, another two streamers were
deployed, one on each side of the over/under
streamers, at a depth of 18 m (previous page,
bottom). One additional streamer was towed at
7.2 m [23.6 ft], directly above the over/under
pair, to acquire conventional Q-Marine data as an
experimental control. All five streamers were
equipped with Q-Fin marine seismic streamer
steering devices and a complete acoustic
network to allow accurate positioning of the
central streamers. The recording bandwidth was
increased to record from 1.5 to 200 Hz to include
the anticipated low-frequency response.
Throughout acquisition, monitoring of the
vertical and horizontal separations of the
streamers showed that variation in the 7.2-m
vertical separation averaged less than 10 cm
[4 in.], and the horizontal, or crossline,
separation was always less than 6 m [20 ft]—
believed to be adequate for separation of the
upgoing and downgoing wavefields.
Comparison between the control dataset and
the upgoing wavefield from the over/under line
shows improved low-frequency response and
higher signal-to-noise ratio for the over/under
data (above right). The deep sediments in basins
bounded by faults and salt diapirs are much
clearer in the over/under line, as are the
near-vertical flanks of the salt and highly
dipping features throughout the section. The
bandwidth extension to lower frequencies allows
deeper penetration of seismic energy and better
imaging of deep reflections than seen in the
control dataset.
26. Low-frequency information is needed for inversion of
seismic data because typically, the seismic data
themselves provide only high-frequency information,
such as relative changes in reflectivity or acoustic
impedance at each layer boundary. Low-frequency
information, typically derived from checkshots or
integrated sonic logs, provides the baseline from which
the high-frequency data vary. By combining both low-
and high-frequency information, absolute changes in
rock properties may be calculated. Because seismic
inversion requires a combination of seismic and well
data, it usually cannot be performed reliably in the
absence of wells. However, if seismic data could be
acquired with low-frequency content intact, inversion
could be carried out in more areas.
27. Moldoveanu N, Combee L, Egan M, Hampson G, Sydora L
and Abriel W: “Over/Under Towed-Streamer Acquisition:
A Method to Extend Seismic Bandwidth to Both Higher
and Lower Frequencies,” The Leading Edge 26, no. 1
(January 2007): 41–42, 44–46, 48, 50–54, 56–58.
28. Haggerty PE; “Method and Apparatus for Canceling
Reverberations in Water Layers,” U.S. Patent
No. 2,757,356, July 21, 1956.
29. Brink M and Svendsen M: “Marine Seismic Exploration
Using Vertical Receiver Arrays: A Means for Reduction
of Weather Downtime,” Expanded Abstracts, 57th SEG
Annual International Meeting and Exposition,
New Orleans (October 11–15, 1987): 184–187.
30. Sonneland L, Berg LE, Eidsvig P, Haugen R, Fotland B
and Vestby J: “2-D Deghosting Using Vertical Receiver
Arrays,” Expanded Abstracts, 56th SEG Annual
International Meeting and Exposition, Houston
(November 2–6, 1986): 516–519.
31. Posthumus BJ: “Deghosting Using a Twin Streamer
Configuration,” Geophysical Prospecting 41, no. 3
(April 1993): 267–286.
32. Moldoveanu et al, reference 27.
> Improved low-frequency content and signal penetration from over/under data (bottom) compared
with a traditionally acquired control survey (top). The migrated data from the upgoing wavefield of
the over/under line show better imaging of steeply dipping interfaces and improved low-frequency
content in signals from deep sediment layers. (Adapted from Moldoveanu et al, reference 27.)
Time,s
1
2
3
4
5
6
Time,s
1
2
3
4
5
6

15. Spectral analysis of data acquired from
all three central streamers demonstrates that
the upgoing wavefield also has better high-
frequency response than data from any of the
streamers individually, and that the low-
frequency response is similar to that of the
deepest streamer (above).
Another innovation that was tested in this
over/under survey was the concept of over/under
sources.33
The airgun source arrays were towed at
5- and 10-m depths and their signals were
recorded by the streamer towed at 7.2-m depth.
After the over/under data were processed to
separate the upgoing and downgoing wavefields,
comparison between the migrated upgoing image
obtained from the over/under source combin-
ation and the migrated image from the upper
source alone showed an improvement in low-
frequency content and in the signal-to-noise
ratio at deep reflectors.
After these successful tests, Chevron decided
to apply over/under streamer and source
technology to a large project in the northeast
Atlantic, east of the Faroe Islands and west of the
Shetland Islands.34
Water depths in the Faroe-
Shetland Channel exceed 1,000 m [3,300 ft], and
weather can be harsh. Also known as the West of
Shetlands, this offshore UK region has seen
significant exploration success (above right).
However, some prospective areas await better
delineation with seismic imaging technology that
can see through basalt layers that obscure
underlying structure.
Basalt is a highly attenuative medium, and
although several geophysical methods have been
proposed to improve imaging through it, basalt
remains an obstacle to hydrocarbon exploration
in many areas.35
Its high seismic velocity bends
rays and prevents all but the lowest-frequency
seismic energy from penetrating.
In 2005, the vessel Western Pride completed
a regional 2D over/under survey in this area.36
The survey objective was to improve the
subbasalt image compared with what could be
produced from a single-source, single-streamer
acquisition configuration. The two source depths
were 12 and 20 m [39 and 66 ft], and the two
streamer depths were 20 and 30 m [66 and 98 ft].
The combination of the “over” source at 12 m and
the “over” streamer at 20 m was taken as the
equivalent of what would have been recorded in
conventional acquisition.
Compared with the image from the conven-
tional acquisition, the over/under image shows
far richer low-frequency information beneath
the basalt (next page, top). This allows
interpretation of subbasalt structure with
greater confidence. Moreover, the deep-
penetrating low frequencies are not generated at
the expense of high frequencies. A comparison of
shallow images shows that the over/under
method produces sufficiently high frequencies to
image shallower objectives.
The improvement in bandwidth associated
with over/under acquisition can be seen in a
comparison of amplitude spectra for both the
signal and the noise extracted from a shallow
and a deep window of the conventional and
over/under datasets (next page, bottom). In the
shallow window above the basalt, the over/under
data display a signal bandwidth from 2 to 60 Hz,
while the conventional-data bandwidth is 5 to
37 Hz, where the notch can be seen. In the deep
window, below the basalt, the over/under data
have a lower peak frequency than the conven-
tionally acquired data. In general, the over/under
data have greater bandwidth and greater
separation between signal and noise.
The over/under technique has also been
tested in 3D, with equally positive results.37
In
this case, a 67-mi2 [173-km2] survey was acquired
with four pairs of over/under streamers over a
subsalt play in the Gulf of Mexico. The 3D
experiment demonstrated that the 3- to 55-Hz
bandwidth achievable with conventional
acquisition could be increased to 2 to 63 Hz with
over/under acquisition. The resulting migrated
images showed that the seemingly small
extension in bandwidth produced significant
results in imaging reflections below and within
the salt.
40 Oilfield Review
> Comparison of amplitude spectra from the over/under data and the
shallow-streamer, reference dataset. Amplitude spectra from the upgoing
wavefield obtained from the over/under data (black) show a wider
bandwidth than the shallow-streamer, reference data (green), the data from
the “over” streamer (red) and the data from the “under” streamer (blue).
The upgoing wavefield also has a better high-frequency response than
any of the individual streamers, and has a low-frequency response similar
to that of the deepest streamer. (Adapted from Moldoveanu et al,
reference 27.)
Amplitude,dB
Frequency, Hz
200 40 60 80 100 120 140 160 180 200
25-m streamer depth
7-m streamer depth
18-m streamer depth
Over/under upgoing
wavefield
> The offshore UK region west of the Shetland
Islands, a deepwater hydrocarbon province with
water depths in excess of 1,000 m. Several
companies have had exploration success here,
finding oil in the Clair, Schiehallion and Foinaven
fields, among others (green and red). Many areas
of potential hydrocarbon-bearing formations lie
beneath basalt, which hinders deep penetration of
seismic energy.
km 100
miles 60
S C O T L A N D
Orkney
Islands
Shetland
Islands
Faroe
Islands
Clair
0
0
Faroe-Shetland
Channel
Foinaven
Schiehallion

16. Spring 2007 41
More recently, over/under surveys have been
acquired elsewhere for similar reasons. In the
Barents Sea, salt outcropping at the seafloor
creates a high acoustic-impedance contrast that
attenuates seismic energy. Offshore India, the
hard sea bottom and deep basalt have made
imaging difficult in the past. The over/under
method shows promise for imaging beneath and
within carbonates and in other salt-prone
provinces, such as offshore West Africa.
33. Moldoveanu N: “Source Array for Use in Marine
Seismic Exploration,” U.S. Patent No. 6,961,284,
November 1, 2005.
Moldoveanu N: “Vertical Source Array in Marine
Seismic Exploration,” Expanded Abstracts, 70th SEG
Annual International Meeting and Exposition, Calgary
(August 6–11, 2000): 53–56.
34. Hill D, Combee L and Bacon J: “Imaging Beneath Basalt
Using an Over/Under Towed-Streamer Configuration,”
World Oil 227, no. 5 (May 2006): 55–61.
Hill D, Combee L and Bacon J: “Over/Under Acquisition
and Data Processing: The Next Quantum Leap in Seismic
Technology?” First Break 24, no. 6 (June 2006): 81–95.
35. Kumar D, Bastia R and Guha D: “Prospect Hunting
Below Deccan Basalt: Imaging Challenges and
Solutions,” First Break 22, no. 7 (July 2004): 35–39.
36. Hill et al, reference 34 (May 2006).
37. Moldoveanu et al, reference 27.
> Over/under and conventional images from the Chevron West of Shetlands survey. This over/under survey towed pairs of both sources and streamers.
The image produced from the resulting upgoing wavefield (top right) shows significantly more low-frequency information below the basalt than the
conventional image (top left). The top of basalt is indicated by the yellow arrow. The generation of low frequencies does not decrease the high-frequency
signal content, as seen in a close-up view (in yellow boxes) of a shallow section above the basalt. The shallow section of the over/under image (bottom
right) shows a high-frequency content similar to that seen in the shallow section of the conventionally acquired image (bottom left). [Adapted from Hill
et al, reference 34 (June 2006).]
2.3
4.1
Two-waytraveltime,sTwo-waytraveltime,s
2.0
7.4
2.3
4.1
Two-waytraveltime,sTwo-waytraveltime,s
2.0
7.4
> Comparison of signal and noise amplitude spectra from over/under (top) and conventionally
acquired data (bottom) from West of Shetlands. Spectral analysis of shallow data in a window from
2.9 to 4.0 s demonstrates the wider bandwidth of the over/under data (top left) than the conventional
survey (bottom left). The over/under data exhibit a signal bandwidth from 2 to 60 Hz, while the
conventional data contain signal from 5 to 37 Hz. Comparing signal content deeper, below the basalt,
the over/under data (top right) show a peak at a lower frequency than the conventional data (bottom
right). The greater bandwidth and larger separation between signal (blue) and noise (black) levels
help produce higher resolution images with deeper penetration. [Adapted from Hill et al,
reference 34 (June 2006).]
0
–10
–20
–30
–40
Amplitude,dB
FullOver/UnderCombination
Shallow Analysis Window 2.9 to 4.0 Seconds
20 dB
Deep Analysis Window 4.7 to 5.7 Seconds
Frequency, Hz
0 10 20 30 40 50 60 70
Amplitude,dB
12-mSourceDepth,20-mCableDepth
20 dB
0
–10
–20
–30
–40
Frequency, Hz
0 10 20 30 40 50 60 70

17. Targeting Carbonate Porosity
Carbonates pose another challenge to 3D seismic
imaging. Like basalts, their high seismic velocities
bend rays, hiding both their own internal
structure and that of underlying formations.
After the 2003 discovery of the Lobina field
offshore northeastern Mexico, Pemex needed a
high-resolution 3D survey to assess and rank
potential drilling locations. The Lobina field is
adjacent to the Arenque field, a 1968 discovery
that also would benefit from enhanced reservoir
description. Low resolution caused by insuf-
ficient frequency content limited the usefulness
of a 1996 3D survey.
A new survey, using the Q-Marine system, was
designed to capture a wider bandwidth and
preserve true amplitudes for inversion of
reservoir properties.38
Obstructions and shallow
water depths, ranging from 30 to 80 m [100 to
260 ft], presented design challenges, but the
acquisition of the 320-km2 [124-mi2] survey was
completed in just two months.
Initial processing indicated that the new
survey doubled the maximum recorded
frequency compared with the 1996 survey,
recording up to 60 Hz compared with the 30-Hz
maximum recorded in 1996. The increased
frequency content significantly enhanced
interpreters’ ability to map key reservoir layers.
The objective was to identify high-porosity
zones within two carbonate layers, primarily a
Jurassic limestone, and secondarily, a shallower
Cretaceous carbonate target. The carbonate
layers themselves are high-velocity layers, but in
some zones, high porosity causes a marked
decrease in seismic velocity.
Inverting the stacked seismic data allowed
geophysicists to obtain a trace-by-trace quanti-
tative measure of acoustic impedance.39
After
calibration with acoustic impedances from sonic
and density logs from 40 wells in the survey area,
the seismic acoustic-impedance sections were
converted to porosity sections, using an acoustic-
impedance-to-porosity relationship from the wells.
The resulting porosity volume showed the
internal architecture of the best reservoir units
and allowed Pemex to optimize drilling
locations in the Lobina and Arenque fields
(left). Mapping maximum porosity between the
top and base of the reservoir let Pemex
calculate the volume of sweet spots. In this
example, at the location of proposed Arenque
Well B, the high porosity in the lower reservoir
made this a high-priority target.
Another way to prioritize potential drilling
locations is by comparing the product of porosity
and height for gross reservoir intervals (left).
Comparing the porosity-height product
computed at several wells with the porosity-
height product computed from the seismic data
at the well locations shows a good correlation.
These wells were not used in the inversion of the
seismic data for porosity, and so the comparison
is an excellent test of the qualitative power of
the inversion.
42 Oilfield Review
38. Salter R, Shelander D, Beller M, Flack B, Gillespie D,
Moldoveanu N, Gonzalez Pineda F and Camara Alfaro J:
“Using High-Resolution Seismic for Carbonate Reservoir
Description,” World Oil 227, no. 3 (March 2006): 57–66.
39. Salter R, Shelander D, Beller M, Flack B, Gillespie D,
Moldoveanu N, Pineda F and Camara J: “The Impact of
High-Resolution Seismic Data on Carbonate Reservoir
Description, Offshore Mexico,” presented at the
International Exposition and 75th Annual Meeting of
the SEG, Houston, November 6–11, 2005.
> Porosity section produced by borehole-calibrated inversion of Q-Marine
seismic data over the Arenque and Lobina fields, offshore Mexico. Color-
coding highlights high-porosity zones in two carbonate reservoirs, allowing
Pemex to optimize locations of development wells. These results indicated
that the proposed location of Arenque Well B would encounter high
porosities in the lower carbonate, raising the priority of this well. Other
wells are indicated by black lines. Marker-horizon names are labeled on
the left. (Adapted from Salter et al, reference 38.)
Well B
Dmi
Kti
Jpi
Jsa
Bas
Porosity0 0.32
> Correlation of the porosity-height product derived from inversion of seismic
data with that interpreted from measurements at well locations. Overall, the
porosity-height product from seismic measurements (purple) compares
closely with that calculated from logging measurements (blue). These results
increase confidence in the seismically derived porosity values. (Adapted from
Salter et al, reference 38.)
Porosityxheight,m
30
25
20
15
10
5
0
Arenque5Lobina101Arenque17
Lobina1M
acarela1Arenque4Lobina301stArenque50Arenque56Arenque58Arenque103Arenque101Arenque104
Logging measurements
Seismic measurements

18. Spring 2007 43
Using the porosity-height attribute and
maximum-porosity maps, Pemex reduced the
priority of two drilling locations and gave higher
priority to two others. The seismic results were
available in time for a rig traveling to one
location to be diverted to a preferred location.
The Arenque Well B, drilled with the seismic-
porosity results as a guide, produced oil and
tested at 2,000 bbl/d [318 m3/d]. The seismically
derived porosity results show excellent
correlation with the measured porosity in the
well (right).
Reducing risk in development drilling is an
application in which high-quality seismic data
play a vital role. Many fields that were discovered
with the help of exploration 2D or 3D surveys
have been further characterized by additional 3D
surveys designed to identify optimal development-
well locations. When 3D data are of superior
quality, they can be used to map properties
within reservoirs, increasing confidence in infill-
drilling plans.
A Rich Future
The basic 3D marine seismic survey that
revolutionized exploration in the 1990s has
advanced in many ways. New acquisition methods
that enhance azimuthal coverage, such as wide-
azimuth and rich-azimuth surveys, have provided
a step-change improvement in difficult subsalt
environments. These surveys deliver increased
signal-to-noise ratio and improved reservoir
illumination. The examples in this article
demonstrate the improvements possible with only
basic processing. Full processing of the wide-
azimuth data is in progress and is expected to
significantly improve the results by applying
multiple attenuation and an optimized velocity
model for imaging. Through further application
and experimentation, seismic practitioners expect
even more from WAZ and RAZ surveys, such as
extension to carbonate and subbasalt exploration.
Other potential benefits are improved velocity
models for imaging, especially where anisotropy is
involved, better characterization of fractured
reservoirs, geomechanical studies around planned
deepwater well locations and high-resolution
imaging of shallow drilling hazards.
The improvement in signal bandwidth and
penetration that comes with the over/under
method of towing vertically aligned streamers
and sources has helped illuminate deep
structures that previously were hidden. The
technique is a viable geophysical solution for
> Logging results from Arenque Well B, drilled to exploit the high-porosity
zone identified by inversion of seismic data. Color-coding indicates
seismically derived porosity in two carbonate reservoirs. Well B targeted
the high-porosity zone in the lower carbonate (orange and yellow), just
below the Csa marker (blue-green dot). The porosity calculated from well
logs is projected along the well trajectory, and shows good correlation with
the seismically derived porosity values in both reservoirs. During testing,
the well flowed oil at a rate of 2,000 bbl/d. (Adapted from Salter et al,
reference 38.)
Kti
Jpi
Jsa
Bas
N
Porosity
boosting resolution and low-frequency content in
areas where conventional seismic methods fail.
One future challenge for the over/under method
is to tow more streamers. In recent surveys,
vessels have towed four pairs of streamers.
Theoretically, a few WesternGeco vessels can tow
eight pairs, but this has not yet been attempted.
Researchers are investigating still other
arrangements of streamers for next-generation
over/under acquisition.
The ability to acquire high-fidelity seismic
data with single-sensor technology ensures that
companies have the optimal input data for
inversion routines designed to output reservoir
properties, such as acoustic impedance, porosity
and other rock and fluid characteristics. This
allows interpreters to look inside reservoirs and
determine where to place development wells
with greater confidence. Current inversion
methods require low-frequency information from
well logs or well seismic surveys. However, as
surface seismic acquisition methods—such as
the over/under technique—increase signal low-
frequency content, inversion may become
practical in areas far from existing wells.
As more companies reap the wealth of
recent advancements in marine seismic
technology, they will discover not only more oil
and gas, but also how to use seismic data to
reduce risk, replace reserves, lower finding costs,
decrease the number of development wells,
optimize well placement and accelerate
development programs. –LS