SEISMIC DETECTION OF SUBTLE STRATIGRAPHIC TRAP

Robert T. Ryder
U.S. Geological Survey
Reston, Va.

Myung W. Lee
Warren F. Agena
U.S. Geological Survey
Denver

Robert C. Anderson
Berrong Enterprises Ltd.
Golden, Colo.

This is the second part of a two part article on seismic detection of a subtle stratigraphic trap at Patrick Draw field, Sweetwater County, Wyo. The first part was a discussion of acquisition and processing of the seismic profile and identification of reflections (OGJ, Dec. 17, p. 54).

STRATIGRAPHIC TRAP DETECTION

The detailed geologic cross section illustrated in Fig. 5 shows the structural and depositional setting of subsurface strata in the vicinity of the Patrick Draw/Table Rock field seismic profile.

The 20 mile long (32 km) cross section is based on 20 wells that begin in either the Lower Tertiary Fort Union or the Wasatch formations and extend into rocks as old as the Upper Cretaceous Ericson sandstone (Fig. 2). Formation boundaries identified in the geologic cross section are consistent with those identified in stratigraphic sections1 2 20 that have one or more wells in common with the authors' section.

Fig. 5 and the stratigraphic sections by Weimer1 and McCubbin and Brady2 show that the principal reservoirs in Patrick Draw field consist of two overlapping shallow marine sandstone units at the top of the Almond formation.

Weimer informally defined the reservoir as the upper Almond 5 (UA-5) sandstone, whereas McCubbin and Brady informally defined the reservoir as the Patrick Draw UA-5 sandstone.

SANDSTONE BARS

Weimer and McCubbin and Brady interpreted the reservoir sandstones as being nearshore bar deposits. McCubbin and Brady subdivided the Patrick Draw UA-5 sandstone into a "western bar" and a slightly younger "eastern bar," a practice that the authors follow in this paper. The western and eastern bars are as much as 35 ft (11 m) thick and pinch out westward into paludal and lagoonal shale and siltstone of the Almond and form the top, basal, and lateral seals for the stratigraphic trap accumulation.

The eastern bar pinches out eastward into the Lewis shale, at which point the top of the Almond shifts down-section to a 100-225 ft thick (30-69 m) marine sandstone interval informally identified by McCubbin and Brady as the Table Rock sandstone.

This relatively abrupt 75-100 ft (23-30 m) thickness change in the Lewis shale at the expense of the uppermost Almond formation defines a net east to west transgression of the Almond-Lewis shoreline across the Rock Springs uplift; in addition, similar stratigraphic relations between the Almond and the Lewis in the outcrop and nearby subsurface are described by Weimer, McCubbin and Brady, and Roehler.21

Judging from a digital seismic model study of the geologic cross section between wells 2 and 14 shown in Fig. 4, Anderson and Ryder22 concluded that the westward pinchout of the Patrick Draw sandstone could not be detected with the range of frequency values normally recorded in field seismic investigations. Average velocity (V) and density (D) values used in the model for the Patrick Draw sandstone were 11,500 fps (3.5 km/sec) and 2.35 g/cc.

In the model the Patrick Draw sandstone was treated as a continuous unit to its pinchout edge between wells 3 and 4. An 8-70 hz symmetrical band-pass filter simulated the source wavelet.

The 8-70 hz seismic model of Anderson and Ryder produced a strong peak reflection near the top of the Almond that maintained about the same shape and amplitude over the field and away from the field (Fig. 6). They reasoned that the 5 ms time difference between the small negative impulse response accompanying the top of the Patrick Draw sandstone and the much greater positive impulse response accompanying the base of the Patrick Draw sandstone was too small to be resolved with an 8-70 hz wavelet.

Therefore, at low frequencies, the high-amplitude peak reflection generated at the base of the sandstone masks the weak trough reflection generated at the top of the sandstone and merges westward without a noticeable amplitude change and with the high-amplitude peak reflection generated from the top of the nonreservoir part of the Almond formation.

A detailed seismic model study of the Patrick Draw stratigraphic trap by Ryder et al.23 gave somewhat more encouraging results. This model, based on wells 2-16 on the geologic cross section, differed from the Anderson and Ryder model in that the Patrick Draw sandstone was subdivided into the western and eastern bars, and a 10-20 ft thick (3-6 m) oyster-bearing shale 2 was added to the top of the Almond above the western bar (Fig. 7).

The relatively high acoustic impedance of the oyster-bearing shale (V=12,50014,900 fps; D = 2.55-2.68 g/cc) greatly increased the negative impulse response of the top of the western bar compared to the Anderson and Ryder model and added a high positive impulse response to the top of the Almond above the western bar.

These changes created high-amplitude peak reflections at the top of the Almond formation in the vicinity of the western bar and high-amplitude trough reflections at the top of the western bar (Fig. 8). Consistent with the Anderson and Ryder model are the low-amplitude trough reflections marking the top of the Almond in the vicinity of the eastern bar and the high-amplitude peak reflections marking the base of the western and the eastern bars.

The Ryder et al. model shows that an 80 hz Ricker wavelet, which corresponds with the frequency content of a 10-140 hz band-pass filter, can resolve the updip and downdip pinchout of the western bar and, to a lesser extent, the updip pinchout of the eastern bar.

REVEALING ANOMALY

Abrupt termination of high-amplitude peak reflections marking the base of the western bar between wells 3 and 4 is the most revealing wave-form anomaly in terms of identifying the Patrick Draw stratigraphic trap.

The gaps in the reflecting horizon marking the base of the western bar near well 6 result from decreases in the thickness of the western bar.

When the peak frequency of the Ricker wavelet in the Ryder et al. model is reduced to 40 hz (Fig. 9), the updip pinchout edge of the Patrick Draw sandstone (western bar) is nearly as difficult to detect as it is in the model by Anderson and Ryder.

The only hint of the western bar pinchout on the 40 hz model by Ryder et al. is the amplitude reduction of the peak reflections at the top of the Almond between wells 3 and 4 and the 10 ms time shift between these peak reflections and ones representing the base of the Patrick Draw sandstone. The updip pinchout edge of the eastern bar in the 40 hz model is even more subtle than that of the western bar, which appears only as a seismically quiet zone juxtaposed against a discontinuous group of high-amplitude peak reflections from near the top of the Almond formation near well 10. To compare the Patrick Draw seismic model with seismic line no. 4, the impulse response model of the stratigraphic section shown in Fig. 7 was convolved with a 20 hz Ricker wavelet.

The 20 hz model was limited to three one-dimensional synthetic seismograms located at wells 3, 8, and 16 and to five layers representing a shale unit in the Lewis shale, an unnamed silty shale in the lowermost Lewis shale, the trapping facies of the Almond formation, the Patrick Draw sandstone (western and eastern bars combined), and the upper part of the Ericson sandstone (Fig. 10). Velocity, density, and thickness values assigned to the five layers were derived from wells 3, 8, and 16 and, except for slight velocity and density modifications near the Almond-Ericson contact, are consistent with values assigned to the same five layers in the two-dimensional models.

Although geologically simpler and containing a lower range of frequency values than the 40 and 80 hz models, the 20 hz one-dimensional model shows the same high-amplitude peak reflections at the base of the Patrick Draw sandstone and at the top of the unnamed silty shale of the Lewis shale (Fig. 10). However, in the vicinity of well 3, peak reflections that identify the top of the nonreservoir Almond formation and the unnamed silty shale in the 40 and 80 hz models have merged into a single peak reflection in the 20 hz model. The resultant composite peak waveform has a greater amplitude than its component peaks and a 2530 ms faster arrival time than the peak waveform defining the base of the western bar.

OTHER INDICATIONS

In contrast to the 10 ms time shift between the same horizons on the 40 and 80 hz models (Figs. 8, 9), this 25-30 ms time shift between the composite peak reflections near the top of the Almond and the peak reflections near the base of the western bar may be detected on seismic line 5.

Moreover, the amplitude decay of peak reflections near the top of the unnamed silty shale (eastward across the Patrick Draw sandstone as indicated in the 20 hz model) may be an equally detectable seismic anomaly.

On the seismic profile in the vicinity of the Texaco 15 gas well, peak reflections that mark the approximate top of the Almond formation compose a more or less continuous horizon between the Table Rock anticline and SP 110 (Fig. 4).

This high-amplitude Almond event near the Texaco 15 gas well is the result of the high acoustic impedance contrast between the base of the Lewis shale and the underlying 50-75 ft thick (15-23 m) Table Rock sandstone (Fig. 3). The Almond seismic event near the Table Rock anticline can be followed uninterrupted to near SP 220, where it splits into two reflecting horizons (Fig. 4).

Judging from the seismic model studies, the upper reflection horizon involved in the split probably marks the top of the unnamed silty shale of the Lewis shale (Fig. 11). West of about SP 165, the amplitude of the silty shale reflector becomes consistently greater probably as a result of constructive interference with peak reflections from the oyster-bearing shale at the top of the Almond, as shown in the 20 hz model.

The eastward pinchout of the oyster-bearing shale between wells 11 and 12 supports this interpretation (Figs. 4, 7, 11). Termination of the silty shale reflection on line No. 5 east of well 16 may result from a marked decrease in the velocity contrast between the silty shale and the overlying shale as suggested by the data shown in Fig. 7.

Additional comparisons between model studies and seismic line No. 5 indicate that the lower peak reflection horizon involved in the split near SP 220 and the preceding high-amplitude trough reflections probably mark the base and top, respectively, of the Patrick Draw sandstone (Figs. 4, 11). The high-amplitude peaks that compose the basal Patrick Draw sandstone reflection horizon from SP 130 to SP 170 define the approximate limits of the western bar (Figs. 4, 11).

The updip limit of the western bar, between wells 4 and 5, as defined on the seismic profile, is approximately 0.50.75-mile (0.8-1.2 km) short of the actual updip limit of the bar between wells 3 and 4, as defined on the geologic sections. This slight discrepancy may result from the westernmost perimeter of the western bar being too thin to be resolved using 10-48 hz seismic data as was the 10 ft thick (3 m) Patrick Draw sandstone near well 6 (Figs. 7, 11). Considering the relatively low frequency content of the seismic data, the authors believe that the Patrick Draw sandstone is surprisingly well defined.

DISCUSSION, CONCLUSIONS

The most significant result of the investigation is the seismic detection of the updip pinchout of the principal sandstone reservoir in Patrick Draw oil and gas field.

This pinchout is manifested on the seismic profile by the westward termination of a continuous horizon of high-amplitude peak reflections believed to represent the approximate base of the 20-35 ft thick (6-11 m) reservoir sandstone.

That the horizon of high-amplitude reflections terminates as much as 0.75 miles (1.2 km) east of the actual sandstone pinchout is of little consequence considering the 10-48 hz frequency content of the seismic data and the 10-20 ft (3-6 m) thickness of the sandstone reservoir near its pinchout edge. The successful attempt to detect the stratigraphic trap at Patrick Draw, where the sandstone reservoir is near the one-quarter wavelength lower detection limit proposed by Widess,24 provides encouragement for additional scientifically and (or) economically motivated studies of this kind.

Critical to the detection of the Patrick Draw stratigraphic trap is the identification of key stratigraphic units on the seismic profile by using a synthetic seismogram prepared from sonic log data from the Texaco 15 Table Rock gas well. Without the tie between the stratigraphic and seismic data provided by the synthetic seismogram, identification of the interval containing the Patrick Draw stratigraphic trap would have been nearly impossible.

The seismic model studies by Anderson and Ryder and Ryder et al. that preceded the study of the seismic line also proved to be very beneficial. Although the seismic models did not match the seismic data perfectly, they did produce an anomaly that could be pursued on the field seismic records.

It is important to recognize that the frequency content and the phase of the wavelet are key factors to be considered in detecting stratigraphic anomalies. Because the phase of the wavelet is not known accurately in most cases, it is beneficial to apply a constant phase shift or phase correction 25 after the deconvolution of seismic data to enhance the subtle features of waveform anomalies.

Also, careful analysis of the interference pattern of adjacent stratigraphic units is important in evaluating the influence of the horizon of interest on the overall seismic response, particularly in investigating anomalies associated with thin beds whose thickness is very small compared to the input wavelength.

A list of the references cited may be obtained from the authors.

ACKNOWLEDGMENTS

The authors are greatly indebted to the managements of Forest Oil Corp. and Champlin Petroleum Co. (now Union Pacific Resources Co.) for donating seismic line No. 5 to the U.S. Geological Survey and for granting permission to publish the reprocessed version of the line.

Without the companies' generosity and interest, the authors' investigation would not have been possible.

Louis Willhoyt, formerly a geophysicist with Forest, was an able intermediary in discussions among Forest, Champlin, and the USGS.

The authors also thank past and present members of the seismic-stratigraphic program of the USGS for their support and assistance during the numerous stages of the investigation. Particularly worthy of mention are Alfred H. Balch, former program chief who provided encouragement and advice through the early stages of the study, and John J. Miller, who painstakingly reprocessed several early versions of seismic line No. 5.