GEOPHYSICS, VOL. 59, NO. 11 (NOVEMBER 1994); P. 1650-1665, 22 FIGS.
A 3-D seismic case history evaluating fluvially deposited
thin-bed reservoirs in a gas-producing property
Bob A. Hardage* Raymond A. Levey*, Virginia Pendleton‡,
James Simmons*, and Rick Edson*
ABSTRACT
We conducted a study at Stratton Field, a large Frio
gas-producing property in Kleberg and Nueces Coun-
ties in South Texas, to determine how to best integrate
geophysics, geology, and reservoir engineering tech-
nologies to detect thin-bed compartmented reservoirs
in a fluvially deposited reservoir system. This study
documents that narrow, meandering, channel-fill res-
ervoirs as thin as 10 ft (3 m) and as narrow as 200 ft
(61 m) can be detected with 3-D seismic imaging at
depths exceeding 6000 ft (1800 m) if the 3-D data are
carefully calibrated using vertical seismic profile
(VSP) control. Even though the 3-D seismic images
show considerable stratigraphic detail in the interwell
INTRODUCTION
The Bureau of Economic Geology at The University of
Texas at Austin has completed several research studies to
better understand the internal architecture of complex, het-
erogeneous oil and gas reservoir systems (Finley et al., 1992;
Levey et al., 1992b). One of these research efforts is the
Secondary Gas Recovery (SGR) project, which is an ongoing
study to determine if reservoir compartmentalization in
older producing properties creates gas accumulations that
have either not been contacted, or not been effectively
produced, by the perforated intervals in production wells
(Sippel and Levey, 1991). The primary focus of this SGR
project is to determine how the depositional process and
diagenesis, rather than structure, contribute to reservoir
compartmentalization. Consequently, the field studies are
done in producing intervals that have minimal faulting be-
cause faulting introduces a reservoir compartmentalization
that overprints and complicates any compartmentalization
Manuscript received by the Editor April 18, 1994; revised manuscript received June 15, 1994.
*Bureau of Economic Geology, The University of Texas at Austin, Austin, TX.
‡Formerly Bureau of Economic Geology, The University of Texas at Austin, University Station, Box X, Austin, TX; presently, Integrity
Geophysics, 1503 Palma Plaza, Austin, TX 78703-3434.
© 1994 Society of Exploration Geophysicists. All rights reserved.
1650
spaces and indicate where numerous thin-bed com-
partment boundaries could exist, the seismic images
cannot by themselves specify which stratigraphic fea-
tures are the flow barriers that create the reservoir
compartmentalization. However, when well produc-
tion histories, reservoir pressure histories, and pres-
sure interference tests are incorporated into the 3-D
seismic interpretation, a compartmentalized model of
the reservoir system can be constructed that allows
improved development drilling and reservoir manage-
ment to be implemented. This case history illustrates
how realistic, thin-bed, compartmented reservoir
models result when geologists, engineers, and geo-
physicists work together to develop a unified model of
a stratigraphically complex reservoir system.
effects inherited from the depositional system. The studies
are also done in older producing properties because such
fields usually have enough well-by-well production history
and pressure documentation to confirm whether or not
reservoir compartment boundaries are present (Jirik, 1990;
Kerr, 1990; Kerr and Jirik, 1990; Levey et al., 1992a).
This case history summarizes the results of a SGR field
study that analyzed how fluvial deposition affects gas reser-
voir compartmentalization. We performed this study in a
portion of Stratton Field in Kleberg and Nueces Counties of
South Texas. The stratigraphic interval we studied was the
Oligocene Frio Formation, a thick, fluvially deposited sand-
shale sequence that has been a prolific gas producer in
Stratton Field and in several other fields along the FR-4
depositional trend (Figure 1). The regional structure and
stratigraphy of the Frio system are well documented by
Nanz (1954), Galloway (1977, 1982), Han and Scott (1981),
and Galloway et al. (1982) and will not be repeated here.
 3-D Seismic Thin-bed Case History 1651

THE STUDY SITE
We confined our study to a 7.6-mi2 area (Figure 2) where
3-D seismic data were acquired in this SGR effort. A large
number of wells, some as old as 40 years, existed inside this
3-D grid, and the logs recorded in these wells allowed us to
make a reasonably thorough geologic analysis of the Frio
reservoirs. As shown in Figure 2, we acquired additional
data in several wells (the circled dots) to supplement the
historic well log, production, and reservoir pressure data
bases. These supplemental data consisted of modern well
logs, cores, and various pressure tests. Vertical seismic
profile (VSP) data were recorded in two closely spaced wells
inside the triangle shown near the center of the 3-D grid.
We stationed receiver groups at intervals of 110 ft (34 m).
Each array consisted of 12 inline geophones spanning a
distance of 110 ft (34 m) centered on the receiver flag. Source
flags were positioned at intervals of 220 ft (67 m). At each
VP, eight linear sweeps (l0-120 Hz) were generated and
summed using a 4-vibrator array symmetrically positioned
relative to the source flag. This geometry created a grid of
110 ft x 55 ft (34 m x 17 m) stacking bins in which a stacking
fold of 20 existed over most of the image area. Before
migration, a trace interpolation was done in the source line
direction to reduce the bin size to 55 ft x 55 ft (17 m x 17 m).
The final processed data had an effective bandwidth of 10 to
80 Hz in most of the Frio interval.

3-D SEISMIC DATA ACQUISITION

The 3-D seismic data were recorded across the study area
in four overlapping swaths (Figure 3). Each swath consisted
of six east-west receiver lines spaced 1320 ft (402 m) apart.
The source lines were oriented north-south, spaced 880 ft
(268 m) apart, and extended from receiver line 2 to receiver
line 5 of each swath. The recording began at the south end of
the study area (swath 1, Figure 3a) and rolled northward.
When all of the vibrating points (VPs) shown in swath 1 were
recorded, swath 2 was created by dropping the southern
three receiver lines and adding three receiver lines on the
north side of swath 1. This modification allowed the VPs to
continue northward as continuous source lines (Figure 3b).
In the southern part of the grid (swaths 1 and 2), the source
lines were straight and uniformly spaced because bulldozers
could be used to clear lanes through the mesquite-covered
property. In the northern portion (swaths 3 and 4), permit-
ting restrictions prohibited the use of bulldozers, and the
source lines followed irregular paths along existing roads and
lanes (Figures 3c and 3d).

FIG. 2. Generalized map of the study area. The solid dots

FIG. 1. Map of the prolific Frio FR-4 gas trend in south Texas
showing the location of Stratton Field.
show existing production wells, many drilled at 40-acre
spacings. The circled dots locate wells where additional
geologic and engineering control was acquired in the form of
modern logs, cores, or 2 ressure tests. The 3-D seismic area
is approximately 7.6-mi . VSP control data were recorded in
two closely spaced wells inside the triangle near the center of
the 3-D seismic grid.
1652 Hardage et al.

FIG. 3. The 3-D seismic data were recorded in four overlapping receiver apertures, referred to as swaths 1, 2, 3, and 4. Each
aperture consisted of six east-west receiver lines spaced 1320 ft (402 m) apart. In each swath, north-south source lines spaced
880 ft (268 m) apart, extended from receiver line 2 (circled) at the south to receiver line 5 (circled) at the north. Receiver groups
were spaced 110 ft (33 m) apart, and source points were spaced at intervals of 220 ft (67 m). Data recording began at the south
end of the area using swath 1 (panel a) and ended at the north end with swath 4 (panel d). To convert from one swath to the
next, the three southernmost receiver lines of the recording aperture were dropped, and three receiver lines were added to the
north side. The solid dots show only key wells, not all wells.
 3-D Seismic Thin-bed Case History
THIN-BED INTERPRETATION PROCEDURE
The emphasis in this case study was to demonstrate how
geologic and engineering data are essential in interpreting
depositionally generated reservoir compartment boundaries
in 3-D seismic images. The seismic interpretation at Stratton
Field was particularly challenging because most of the Frio
reservoirs were thin [< 15 ft (5 m)], and they were closely
stacked, in some areas separated only 10 ft-15 ft (3 m-5 m)
vertically. These conditions required precise calibration of
stratigraphic depth-versus-seismic traveltime to extract a
depositional strata1 surface from the 3-D data volume that
would reliably depict the area1 distribution of a particular
Frio thin-bed reservoir.
Thin-bed depth-versus-time calibration
We used VSP as the primary measurement to define where
a specific thin-bed reservoir was positioned in the 3-D
seismic data volume. The locations of the two VSP calibra-
tion wells we used are shown in Figure 2. The zero-offset
VSP data recorded in one of these wells were used to
establish the precise depth-versus-time control needed for
the thin-bed interpretation. These VSP data are shown in
Figure 4, where the zero-offset image is spliced into a
north-south vertical slice from the 3-D data volume passing
through the VSP well. Also shown in the figure is a graphic
representation of the stratigraphic column penetrated by the
VSP well. Only producing or potentially producing Frio
reservoirs are shown in this diagram, and not all of the
reservoirs are labeled by name. The top and base of each
reservoir are accurately positioned in terms of two-way VSP
traveltime, and since there is no difference in the VSP and
3-D time datum in this instance, the reservoirs are also
correctly positioned vertically inside the 3-D seismic data
volume at the VSP well.
Using these VSP traveltime control data, we knew exactly
where each thin-bed reservoir belonged in the 3-D seismic
reflection waveform at the VSP well. We then extended this
thin-bed calibration away from the VSP well and across the
entire 7.6mi2 area imaged by the 3-D data.
Defining chronostratigraphic depositional surfaces
The fundamental assumption we made in our seismic
interpretation was that seismic reflections follow chronos-
tratigraphic depositional surfaces (Vail and Mitchum, 1977).
This assumption means that if we map a continuous seismic
reflection event over the entire 7.6-mi2 area imaged by the
3-D seismic data, we define a geologic surface that corre-
sponds to a fixed, constant depositional time. In other
words, we define a depositional strata1 surface. We were
able to find two such areally continuous reflection events in
the Frio interval. These two surfaces are shown on the
east-west vertical section crossing the VSP well in Figure 5.
At the VSP control well, the apex of the peak associated
with the shallower strata1 surface (the orange surface in
Figure 5) corresponded to the thick C38 reservoir (Figure 4),
and the apex of the peak at the deeper strata1 surface (the
green surface in Figure 5) correlated with the F11 reservoir.
Thus we assumed that the seismic time surface following the
apexes of all of the peaks of the orange event defined the
1653
ancient topographic Frio surface at the time when the C38
reservoir sediments were deposited. Likewise, we assumed
that the seismic time surface following the apexes of the
peaks of the deeper green event defined the ancient deposi-
tional surface associated with the F11 reservoir.
Once the 3-D data volume was flattened relative to one of
these two reference strata1 surfaces, we further assumed that
any horizontal time slice in this flattened data volume also
followed an ancient Frio depositional surface, as long as the
seismic reflection character in the immediate neighborhood
of the time slice was time-conformable with the reflection
character in the immediate vicinity of the reference surface
used to flatten the data volume. In our opinion, the entire
Frio section inside the 7.6.mi2 3-D grid was seismically
conformable to one of the two seismic reference surfaces we
created.
The thin-bed interpretational philosophy we followed can
be summarized as follows:
1) Choose one or more areally continuous Frio seismic
reflections that can be used as reference surfaces to
flatten the 3-D seismic data volume. Specifically, we
picked two such reflection events: the green and orange
surfaces shown in Figure 5.
2) Use VSP data to define the stratigraphic depth corre-
sponding to the peak (or trough) time of each selected
reference strata1 surface. We found that the shallow
(orange) reflection peak time correlated with the C38
reservoir at our VSP control well, and the deeper
(green) reflection peak time correlated with the F11
reservoir.
3) Flatten the 3-D seismic data volume relative to one of
the reference strata1 surfaces. In the immediate neigh-
borhood of the reference surface, this procedure re-
stores the flat, no-dip, surface topography that likely
existed at the time the meandering, fluvial channel
environment deposited the Frio sediments associated
with that seismic reference reflection.
4) Make horizontal time slices through this flattened seis-
mic data volume only in time windows in which the
seismic reflections are conformable to the seismic
reflection associated with the flattened reference sur-
face. These time slices are then assumed to follow
individual, fixed, depositional surfaces because the
conformable reference reflection event was assumed to
be a strata1 surface (Vail and Mitchum, 1977).
5) Use the same VSP control that defined where the
reference seismic surface was positioned in strati-
graphic depth to define where each time slice is posi-
tioned in stratigraphic depth. If two thin-bed reservoirs
occur at times and relative to the reference
strata1 surface at the VSP control well, then the depo-
sitional surfaces for these two reservoirs are horizontal
time slices at times and in the flattened
seismic data volume.
Following this procedure, we found that in this specific
interpretation problem, where we had many closely spaced
(vertically) thin-beds, the VSP-defined position of a particu-
lar thin-bed reservoir was rarely at the apex of a reflection
peak or trough. Invariably, each thin-bed of interest was
1654 et al.

FIG. 5. An east-west vertical slice from the migrated 3-D data crossing the VSP control well, which is at
crossline coordinate 118. The green and orange surfaces follow the two continuous Frio reflection events that
were used as reference strata1 surfaces for flattening the 3-D data volume. The yellow surface is the deepest
Frio reservoir level. Immediately below this yellow surface is the severely faulted Vicksburg section.
 3-D Seismic Thin-bed Case History
positioned at some intermediate, often nondescript, phase
point in the reflection waveform at the VSP control well.
To create a seismic image that emphasized the internal
complex architecture of a given thin-bed reservoir system,
we time-shifted the migrated 3-D data volume first so the
proper predefined reference strata1 surface was flat. Then we
made a horizontal time slice through this flattened data
volume at the exact VSP-defined time for the targeted
thin-bed, regardless of where that time slice was positioned
in the reflection waveform at the VSP control well. We then
assumed that the seismic time surface contained in this
horizontal slice was the fixed depositional strata1 surface
FIG. 6. Seismic reflection amplitude behavior on a strata1
time surface that passes through the F39 reservoir at the
VSP calibration well.
FIG. 7. Magnified view of the reflection amplitude behavior on the F39 strata1 time surface in the vicinity of
four critical information wells.
1655
where that thin-bed unit was deposited. We also assumed
that any seismic anomalies seen on this surface would be
related directly to stratigraphic heterogeneities within the
targeted thin-bed, and to a lesser degree, would be related to
stratigraphic variations in thin-beds positioned immediately
above and below the target thin-bed.
We now show the results of this thin-bed interpretational
procedure at Stratton Field and support the interpretations
with geologic and engineering control.
F39 RESERVOIR
The F39 reservoir was the deepest Frio reservoir we
studied. The depositional surface for the F39 reservoir
(defined by the steps described above) is shown by the
yellow horizon in Figure 5. This surface is immediately
above the severely faulted Vicksburg section, and we antic-
ipated some faulting effects might extend into the lowest Frio
and create some nondepositional reservoir compartmental-
ization at the F39 level.
The reflection amplitude behavior on our interpreted F39
depositional surface is shown in Figure 6. We assumed the
red, linear north-south trends near the center of the image to
be residual effects from the deeper Vicksburg faults. A
magnified view of this F39 surface in the vicinity of four key
wells is displayed in Figure 7. These wells were critical to
our study because we were able to acquire F39 reservoir
pressure measurements in all four wells at the same time
(Figure 8), and the differences in these static pressures told
us that each well was in a different F39 compartment. We
then examined the 3-D seismic image and the available
geologic control to see if these data indicated where the
boundaries were that segregated the F39 reservoir into these
distinct compartments.
Figure 9 displays our interpretation of the available geo-
logic control. Our interpretation of the log curves inferred
that the F39 reservoir in each well was deposited in a
channel environment that showed some evidence of splay
deposition. These log data, by themselves, do not provide
1656 Hardage et al.
much information about where compartment boundaries
may exist. However, the seismic image does indicate some
possible compartment boundaries. For example, the most
likely cause of the compartment boundary that separates
well 197 from the other wells is the depositional variation
that created the red/blue (positive/negative) amplitude
changes that trend north-south between crossline coordi-
nates 130 and 140 (Figure 7). Similarly, a probable seismic
indication of the compartment boundary that segregates well
75 from the other wells is the positive-to-negative (red-to-
blue) amplitude variations trending north-south between
crossline coordinates 110 and 120. However, this same logic
of looking for inter-well reflection amplitude changes does
FIG. 8. Static pressures in the F39 reservoir measured during
January 1992 in the four wells shown in Figure 7. The
bottom-hole pressure in well 75 was 300 psi at abandon-
ment.
FIG. 9. Stratigraphic cross-section of the F39 reservoir showing the depositional environments interpreted from log shapes and
the initial bottom-hole pressure (BHP) observed in each well. The date when each well was drilled is shown above the BHP
value. The log depths are measured from KB and the curves are shifted to align on a stratigraphic datum.
not explain why there is a compartment boundary between
wells 175 and 202, which are only 200 ft (61 m) apart, since
there is no appreciable change in the reflection phase be-
tween these two wells (i.e., there is no color change in the
reflection amplitude plot between wells 175 and 202 in
Figure 7).
We relied on offset VSP imaging to verify that there was a
significant change in the F39 reservoir seismic reflection
character between wells 175 and 202. One of these VSP
images is shown in Figure 10. The classical VSP-CDP image
(right side of Figure 10) revealed a subtle change in the F39
reflection waveform in the vicinity of the 175 well, but the
most definitive indication of a compartment boundary was
found by examining the individual reflection traces before
they were binned. These traces and the VSP-CDP stacking
bin overlay are shown on the left side of Figure 10. When the
F39 reflection peaks are followed across the binning corri-
dors toward well 175 (which is located in the eighth stacking
corridor from the left edge), we find the peaks terminate in
the sixth corridor, at least 50 ft (15 m) short of well 175, and
they do not resume in a robust fashion until corridor 12,
some 100 ft (30 m) beyond well 175.
Thus, we believe the VSP data provide seismic evidence
of a compartment boundary between wells 175 and 202 and
forces us to reconsider why this boundary is not evident in
the 3-D image (Figure 7). We propose the following expla-
nation. The well positions in Figure 7 show only where the
wellheads are located, not where the wellbores penetrated
the F39 reservoir. We did not run deviation surveys to
determine the true bottom-hole locations of the wells. All
available information simply indicated the holes were verti-
cal, and we now believe the assumption of true vertical wells
can lead to interpretational difficulties when the depositional
stratigraphy varies as rapidly in the lateral direction as it
does in this example. For example, the F39 reservoir in these
wells is at a depth of approximately 6700 ft (2042 m), and at
 3-D Seismic Thin-bed Case History
this depth, a 1 degree deviation from vertical is a horizontal
movement of approximately 120 ft (37 m). If a wellbore is
within 3 or 4 degrees of vertical, most people define the well
to be vertical. An inspection of Figure 7 shows, however,
that the borehole for either well 175 or 202 has to deviate
eastward only 100 ft (30 m) or so to move that wellbore from
a red zone (positive reflection amplitude) to a blue zone
(negative reflection amplitude). Thus, the 3-D seismic image
may be revealing the compartment boundary between wells
175 and 202 if we knew exactly the inline/crossline seismic
coordinates where the wellbores intersected the F39 reser-
voir .
Our conclusion from analyzing the seismic, geologic, and
engineering data associated with the F39 reservoir is that we
can seismically detect F39 reservoir compartments, at least
in the vicinity of wells 75, 175, 197, and 202, but we must
interpret the seismic image with the assistance of reservoir
pressure data to infer which of the many stratigraphic
changes revealed in the seismic image are most likely to be
the compartment boundaries.
1657
F37 RESERVOIR
The F37 reservoir was approximately 20 ft (6 m) above the
F39 reservoir in our VSP calibration well, which is a
two-way traveltime difference of only 4 ms. Using the
previously described thin-bed interpretation procedure, we
created a time slice through our flattened 3-D data volume
4 ms above the F39 strata1 surface. This F37 surface is
displayed in Figure 11. Comparing this image with the
F39 surface (Figure 6), we saw red, linear north-south
channels in the central part of the F37 image similar to those
observed in the F39 image, implying that Vicksburg faulting
was still controlling sedimentation in this part of the field.
However, there was a significant difference in the southeast
quadrant of the F39 and F37 images. Specifically, meander
channel features occurred at the F37 level but were not
present at the deeper F39 surface. We focused considerable
attention on this F37 depositional topography and show an
enlarged plot of the meander features in Figure 12.

FIG. 10. Offset VSP imaging of the interwell space between wells 175 and 202. The F39 reservoir is imaged by the low-amplitude
black peaks immediately above the dashed line labeled F39. The VSP geometry allowed stacking bins only 25 ft (8 m) wide to
be used, so well 175 is positioned at the ninth trace in the VSP-CDP image, which is 200 ft (61 m) away from well 202, the
receiver well. In the VSP-CDP image, the F39 peaks exhibit some change in reflection character near the 175 well, but a more
definitive indication of an interwell stratigraphic discontinuity in the F39 reservoir is provided by the display on the left, which
shows each individual receiver trace before the traces are summed to create the VSP-CDP image. The VSP-CDP stacking
corridors are defined by the superimposed grid of lines sloping up to the right. In this prestack display, the F39 reflection peak
disappears in stacking corridor 6, about 50 ft (15 m) short of well 175, implying an interwell stratigraphic break of some type.
1658
FIG. 11. Seismic reflection amplitude behavior on a strata1
time surface that passes through the F37 reservoir at the
VSP calibration well. This surface is conformable with, and
only 4 ms above, the F39 strata1 surface (Figure 6).
Hardage et al.
We made a log-based stratigraphic cross-section of the
F37 reservoir across the meander features and continuing
southward beyond the seismic grid (Figure 13). The deposi-
tional environment (either channel or splay) at each well is
an interpretation based on log curve shape and was made
before the 3-D seismic data were recorded.
This initial geologically based interpretation of the F37
depositional environments indicates that the meander fea-
ture seen in the F37 seismic surface is indeed a depositional
channel. Specifically, the log interpretation (Figure 13) im-
plies the F37 reservoirs found in wells 189 and 185 were
deposited as channel fill, and the seismic image shows these
wells to be directly atop a meander feature. The initial
depositional interpretation for the extremely thin F37 reser-
voir in well 211 was that this wellbore could have penetrated
either splay or channel fill. The splay option is indicated in
Figure 13. Because the 211 wellhead is approximately 300 ft
(91 m) north of the meander feature (Figure 12), the log-
based interpretation of the F37 depositional environment at
the 211 well is also supported by seismic evidence, since the
bottomhole location could be either in channel fill or in a
splay.
We analyzed pressure histories recorded in several F37
reservoirs near these seismic meander features to determine
if reservoir compartmentalization existed. These pressure
histories, summarized in Figure 14, show there are at least
three, and perhaps four, individual F37 reservoir compart-
ments in this area of the field. We relied heavily on these
pressure data to guide the interpretation of the thin-bedded
F37 reservoirs.
A reservoir model that honors all three data bases-the
seismic, the geological, and the reservoir engineering-is
proposed in Figure 15. This model assumes that the F37

FIG. 12. Magnified view of the reflection amplitude behavior on the F37 strata1 time surface in the vicinity of
four critical information wells. An interpretation of the stacked thin-bed channel features revealed in this
image follows as Figure 15.
 3-D Seismic Thin-bed Case History 1659

reservoir in the southeast quadrant of the 3-D grid is com-

posed of three intermeshed channels, labeled A, B, and C,
and a grid overlay of seismic inline and crossline coordinates
is provided so these channels can be correlated with features
in the 3-D seismic image. The location of the F37 strati-
graphic cross-section (Figure 13) is shown, but this geologic
information defines channel locations along only a single 2-D
profile of the model. The important information again is the
reservoir pressure data, because without this engineering
data there would be no reason to conclude that a 3-channel
model would be appropriate. Thus, the model places well
129 in channel A and well 185 in channel B, which allows
these two wells to be in different F37 pressure regimes; i.e.,
in different compartments (channels). Wells 127 and 161 are
proposed to be in channel C, south of the 3-D seismic
coverage. Only one meander loop of this hypothesized
channel C extends into the 3-D seismic grid. The rapid F37
pressure decay observed in well 189 (Figure 14) implies that
this well is not in pressure communication with well 185, FIG. 14. F37 reservoir pressure histories observed in wells
even though both wells are in channel B. For this reason, we near the seismically imaged meander features. These pres-
show a stratigraphic variation in channel B where there may sure decline curves indicate that these wells are positioned in
be an intrachannel compartment boundary. at least three different F37 compartments, labeled as chan-
nels A, B, and C (same labeling notation used in Figures 13
We wish to emphasize that the reservoir model in and 15). There may be an additional, intrachannel, F37
Figure 15 is hypothetical and may not yet be the correct compartment boundary in channel system B that segregates
picture of the compartmentalized nature of these F37 reser- well 189 (the bold Xs) from well 185 (the open circles).
voirs. However, we do know that the F37 reservoir in this
portion of Stratton Field is segregated into distinct compart-
ments, that this compartmentalization must be caused by the

FIG. 13. Stratigraphic cross-section of the F37 reservoir showing the depositional environments interpreted from log shapes and
the initial bottom-hole pressure (BHP) observed in each well. All log curves are depth-shifted to a marker datum. The channels
labeled A, B, and C refer to meander features shown in map view in Figure 15. The northern three wells are inside the 3-D
seismic grid; the southern three wells are not.
1660 Hardage et al.
fluvial deposition because the seismic data show no evidence
of faulting, and that the proposed reservoir model honors all
existing data that provide any information about the F37
reservoir system.
We concluded that the seismic image in Figure 12 revealed
not just one meander channel system but at least three
intermeshed thin-bed channels, and that again by using
common existing reservoir engineering data (i.e., pressure
histories), we were able to use 3-D seismic images to define
where compartment boundaries most likely existed in the
interwell spaces.
D11 RESERVOIR
Our third example of a compartmented fluvial reservoir iS
the D00 system, which is one of the very thin reservoir
immediately above the D35 reservoir shown in Figure 4
Again, using our VSP calibration technique to define the
appropriate flattened time slice that equated to thE
D11 stratigraphic level, we generated the reflection image ir
Figure 16.
The information in Figure 17 summarizes the D11 geologic
and engineering data available in this portion of the 3-C

FIG. 15. Proposed model for the F37 reservoir system imaged in Figure 12. This model honors the
stratigraphic cross-section shown in Figure 13 and the pressure history shown in Figure 14. The channels are
arbitrarily drawn as A being the deepest and C the shallowest. An intrachannel barrier is proposed near
crossline, inline coordinates (125, 35) to explain the different pressure behaviors in wells 185 and 189 (see
Figure 14). South of the 3-D seismic coverage area, the boundaries of channels A, B, and C are drawn as
dashed lines to show that the channel shapes and positions are highly speculative.
 3-D Seismic Thin-bed Case History 1661

FIG. 16. Seismic reflection amplitude behavior on a strata1 time surface that passes through the D11 reservoir
at the VSP calibration well.

FIG. 17. Stratigraphic cross-section of D11 reservoirs showing the depositional environments interpreted from log shapes and
the initial D11 reservoir pressures observed in two critical wells (77 and 150). These pressure data show that well 150, drilled
in 1983, penetrated a D11 compartment with virgin reservoir pressure. This compartment evidently has no effective
communication with the D11 compartment where well 77 was drilled because well 77 was abandoned in 1972 because of
depleted pressure. Consequently, an intermeshed two-channel reservoir system is proposed for the D11 reservoir level in this
portion of Stratton Field.
1662
FIG. 18. Proposed model for the D11 reservoir system
imaged in Figure 16. This model honors the geologic and
engineering data summarized in Figure 17 and mimics the
meander pattern shown in Figure 16.
Hardage et al.
seismic grid. The D11 reservoirs in these four analysis wells
were all interpreted from log shape analysis to be channel
fill; some log character indicated splay deposition in the
younger channel system. The pressure information was
again the key information that told us how many reservoir
compartments should be seismically imaged. As noted in
Figure 17, the oldest production well (77) was drilled in 1956
and encountered a reservoir pressure of 1740 psi. This well
was abandoned in 1972 when the pressure declined to
uneconomic levels. Wells 135, 150, and 156 were drilled in
the 1980s, and all three wells encountered virgin D11 reser-
voir pressure, implying these wells were in a single D11
compartment that was depositionally segregated from well
77.
We thus interpreted our D11 seismic image with the
knowledge that we were dealing with at least two inter-
meshed or juxtaposed thin-bed channel systems. The reser-
voir model we inferred from Figure 16 is shown in Figure 18.
This model honored the geologic information by placing each
of the four wells in a seismically defined D11 feature that
appeared to be a channel and honored the pressure data by
placing wells 135, 150, and 156 in the same channel complex
and well 77 in a separate channel system.
We concluded again that fluvial deposition can create
compartmented reservoirs without the stratigraphic section
being faulted, that individual reservoir compartments can be
seismically imaged when geologic and reservoir engineering
data are used to guide the seismic interpretation process, and
that detailed VSP data or their equivalent should be used to
precisely calibrate the location of thin-bedded reservoirs in
the seismic reflection response. This VSP calibration is
critical because when many thin-beds are stacked close
together, a particular thin-bed seismic response can be

FIG. 19. Stratigraphic cross section of the F21 reservoirs showing the depositional environments interpreted from log shape
analysis and the initial reservoir pressures (BHP).
 3-D Seismic Thin-bed Case History 1663

completely missed if the time slice is mispositioned by only
3 or 4 ms.
F21 RESERVOIRS
For our fourth and last example of seismic thin-bed
interpretation of a compartmented reservoir, we show our
3-D seismic interpretation of four closely stacked reservoirs:
FIG. 20. F21 reservoir pressure (normalized by depth) versus
cumulative gas production for completions A, B, C defined
in Figure 19. Because the data points follow a single straight
line trend, all three completions are assumed to share the
same reservoir compartment.
F20, F21, F23, and F25. These four reservoirs are extremely
thin and overlap each other in a vertical section only 50 ft
(15 m) thick (approximately). For convenience, the reser-
voirs in this 50 ft (15 m) interval are collectively referred to
as the F21 reservoir system.
We focused our attention in the area of wells 130, 164, 176,
195, and 201, where we had the geologic and engineering
control shown in Figures 19 and 20. This control showed that
we were dealing once again with a splay/channel-fill deposi-
tional regime (Figure 19) and that wells 176, 195, and 201
(completions A, B, and C) produced from the same F21
reservoir compartment (Figure 20). Transferring this infor-
mation to the seismic image in Figure 21 showed that we had
as much depositional heterogeneity in the interwell spaces at
this F21 reservoir level as we had observed at the F39
(Figure 7), F37 (Figure 12), and D11 (Figure 16) levels. Yet
the reservoir pressure behavior implied the F21 reservoir
system had no interwell compartment boundary whereas the
other three reservoir levels (F39, F37, D11) did.
Our proposed model for the F21 reservoir system imaged
in Figure 21 is shown in Figure 22. Although the meander
channels in this model overlay and intersect each other much
as do the channels shown in the models proposed in
Figures 15 and 18, pressure data forced us to conclude that
all of the intermeshed F21 channels between wells 176, 195,
and 201 (Figure 21) comprise a single reservoir compart-
ment, whereas, the intermeshed channels in Figures 15 and
18 create distinct reservoir compartments.
We were particularly impressed that thin-bed stratigraphy
as subtle as what existed in the F21 reservoirs could be
revealed in 3-D seismic images. Note that some of the F21

FIG. 21. Seismic reflection behavior on a strata1 time surface that passes through the F21 reservoir at the VSP
calibration well. Note the stratigraphic resolution in this 3-D image. The intermeshed channels are as narrow
as 200 ft (61 m), and log control indicates the channel fills are approximately 10 ft (3 m) thick (Figure 19). This
subtle stratigraphy is imaged even though the channels are at a depth of 6600 ft (2012 m).
1664 Hardage et al.

FIG. 22. Proposed model for the F21 reservoir system imaged in Figure 21. This model honors the available
geologic control (Figure 19) and engineering control (Figure 20), and duplicates the meander pattern revealed
in Figure 2 1.

channels are only 200 ft (61 m) wide (Figure 2 1) and 10 ft
(3 m) thick (Figure 19). Yet, by using VSP control to define
exactly how to time slice the 3-D data volume, we could find
these small-scale depositional features even though they
were at a depth of 6600 ft (2012 m) (Figure 19).
In this F21 reservoir study, we concluded that 3-D seismic
data, when properly calibrated using VSP control, can reveal
thin-bed depositional heterogeneities that are not flow bar-
riers even though they appear to be as complex as the
heterogeneity where known compartment boundaries exist.
This conclusion further emphasized the importance of incor-
porating pressure data into the seismic interpretation, for
without this engineering control we never knew if the
stratigraphic variation revealed by the seismic imaging was,
or was not, a barrier to lateral fluid flow.
CONCLUSION
We show four examples of seismic thin-bed interpretation
in a fluvially deposited gas reservoir and support these
interpretations with geologic and reservoir engineering data.
In three of these examples, the 3-D seismic data reveal
stratigraphic variations where reservoir pressure informa-
tion implies that a compartment boundary should exist. In
the fourth example, we show an equivalent degree of depo-
sitional heterogeneity as in the first three cases, yet reservoir
pressure behavior indicates that none of these stratigraphic
variations creates a compartment boundary. These examples
illustrate that although fluvial deposition creates numerous
compartment boundaries, determining which seismically im-
aged stratigraphic changes are compartment boundaries re-
quires that geologic and reservoir engineering data (particu-
larly reservoir pressure data) be incorporated into the
seismic interpretation.
We found it was particularly important to have an accurate
and reliable way to translate thin-bed stratigraphy (known in
depth) into precisely defined seismic time windows. We are
convinced that VSP data, when properly recorded and
processed, are the best information to establish the detailed
depth-versus-time calibration required to seismically distin-
guish closely spaced thin-beds. We used this VSP calibration
procedure to seismically distinguish thin-beds that were
vertically separated by as little as 4 ms [i.e., 20 ft (6 m)].
 3-D Seismic Thin-bed Case History
ACKNOWLEDGMENTS
Funding for this research was provided by the Gas Re-
search Institute under Contract No. 5088-212-1718 and the
U.S. Department of Energy under Contract No. DE-FG21-
88MC25031. Publication has been authorized by the Direc-
tor, Bureau of Economic Geology.
REFERENCES
Finley, R. J., Levey, R. A., Hardage, B. A., and Sippel, M. A.,
1992, Evidence and strategies for infield reserve growth of natural
gas reservoirs, in Thompson, H. A., Ed., Preprints of the 1992
International Gas Research Conference: Gas Research Institute,
American Gas Association, U.S. Dept. of Energy, 13-22.
Galloway, W. E., 1977, Catahoula Formation of the Texas Coastal
Plain-depositional systems, composition, structural develop-
ment, groundwater flow history, and uranium distribution: The
University of Texas at Austin, Bureau of Economic Geology
Report of Investigations No. 87.
- 1982, Depositional architecture of Cenozoic Gulf Coastal
Plain fluvial systems, in Ethridge, F. G., and Flores, R. M., Eds.,
Recent and ancient nonmarine depositional environments: Mod-
els for exploration: SEPM Spec. Publ. No. 31, 127-155.
Galloway, W. E., Hobday, D. K., and Magara, K., 1982, Frio
Formation of the Texas Gulf Coast basin-depositiona1 systems,
structural framework, and hydrocarbon origin, migration, distri-
bution, and exploration potential: The University of Texas at
Austin, Bureau of Economic Geology Report of Investigations
No. 122.
Han, J. H., and Scott, A. J., 198 1, Relationship of syndepositional
structures and deltation, Vicksburg Formation (Oligocene), South
1665
Texas: SEPM, Gulf Coast section, 2d Annual Research Confer-
ence, Program and Abstracts, 33-40.
Jirik, L. A., 1990, Reservoir heterogeneity in middle Frio fluvial
sandstones: Case studies in Seeligson field, Jim Wells County,
Texas: Gulf Coast Association ‘of Geological Societies Transac-
tions, 40, 335-351.
Kerr, D. R., 1990, Reservoir heterogeneity in the middle Frio
Formation: Case studies in Stratton and Agua Dulce fields,
Nueces County, Texas: Gulf Coast Association of Geological
Societies Trans., 40, 363-372.
Kerr, D. R., and Jirik, L. A., 1990, Fluvial architecture and
reservoir compartmentalization in the Oligocene middle Frio
Formation, South Texas: Gulf Coast Association of Geological
Societies Transactions, 40, 373-380.
Levey, R. A., Langford, R. P., Hardage, B. A., Grigsby, J. D.,
Ambrose, W. A., Finley, R. J., and Guevara, E. H., 1992b,
Incremental gas reservoir development in fluvial-deltaic plays of
the Gulf Coast: Houston Geological Society, 34, no. 8, 30-31 and
42-44.
Levey, R. A., Sippel, M. A., Finley, R. J., and Langford, R. P.,
1992a, Stratigraphic compartmentalization within gas reservoirs:
Examples from fluvial-deltaic reservoirs of the Texas Gulf Coast:
Gulf Coast Association of Geological Societies Trans., 42, 227-
235.
Nanz, R. H., 1954, Genesis of Oligocene sandstone reservoir,
Seeligson Field, Jim Wells and Kleberg Counties, Texas: AAPG
Bull. 38, no. 1, 96-117.
Sippel, M. A., and Levey, R. A., 1991, Gas reserve growth analysis
of fluvial-deltaic reservoirs in the Frio and Vicksburg Formations
located in the Stratton Field, onshore Texas Gulf Coast Basin:
SPE paper no. SPE 22919, 337-344.
Vail, P. R., and Mitchum, R. M., Jr., 1977, Seismic stratigraphy and
global changes of sea level, part l-overview: AAPG Memoir 26,
Seismic stratigraphy-applications to hydrocarbon exploration,
51-52.