Imaging of phase changes and fluid movement in and between reservoirs at Teal South

Teal South is a small oil field in Eugene Island Block 354 in the Gulf of Mexico. It is located in shallow water (85 m) and has produced oil and gas from several reservoirs, ranging in depth from 4000 to 8000 ft. The reservoir of interest in this and many previous studies (starting with Ebrom et al., 1998a) is located in the “4500-ft sand,” a Tertiary sandstone named for its approximate depth. Only one reservoir among many small reservoirs at this approximate depth range was under production at the time of the time-lapse seismic data acquisition. At the time of discovery, the reservoir was saturated with light oil just above bubble point pressure, with no free gas. Initial production from the reservoir was under liquid-expansion drive, causing a steep drop in pressure. As gas came out of solution, pressure support was provided primarily by solution-gas drive.

The 4500-ft sand is poorly consolidated, highly porous, and overpressured, and as such should be very sensitive to pressure and saturation changes, as discussed by Pennington et al. (2001) and Islam and Pennington (2015). This sensitivity to fluid saturation and stress changes, together with its anticipated rapid depletion, made Teal South highly suitable for time-lapse studies. In 1996, Teal South was selected as a test site to investigate the efficiency of a novel 4C/4D permanent reservoir monitoring system (Ebrom et al., 1998a). Shortly prior to commencement of production in late 1996, a “legacy” seismic data set had already been obtained using a conventional streamer system. Under the time-lapse project (initially conducted by Texaco, and later through a consortium led by the Energy Research Clearing House [ERCH]), two sets of time-lapse data were acquired using ocean-bottom cables (OBCs). The first time-lapse survey took place in July-August 1997 after approximately eight months of production (phase I) and the second in April 1999 after approximately 30 months of production (phase II).

The Teal South project provided a total of three sets of seismic data, recorded at three different times, representing different reservoir conditions. The ocean-bottom time-lapse data (phases I and II) were acquired and processed with every effort to match the two surveys. The details of survey geometry and processing steps are described in previous studies (Ebrom et al., 1998a; Rodriguez-Suarez et al., 2000; Druzhinin and MacBeth, 2001). The legacy data, however, were acquired using streamer technology and were processed independently of the time-lapse data. Although prestack data for the two time-lapse OBC surveys were made available to us, only the poststack data were available for the legacy survey. As a result, we restricted our analysis to comparisons of the final, stacked data sets from all three surveys.

Many earlier studies have investigated the evolution of the reservoir at Teal South. For example, Ebrom et al. (1998a, 1998b) compare the legacy streamer data with phase I OBC data, demonstrating that the release of gas from solution could be observed in the producing reservoir. Christie et al. (2002) perform reservoir flow simulations, providing a history match for production data that was used by other studies, including this one. Pennington et al. (2001) qualitatively explain time-lapse changes in Teal South using inversion of the legacy stacked data set, amplitude-variation-with-offset (AVO) analysis of the two time-lapse data sets, and rock-physics models. Ezawi et al. (2012) use volume visualization of attributes using stacked data from the two later surveys to identify details of fluid movement. This study extends the analyses of Pennington et al. (2001) and Ezawi et al. (2012) to investigate the time-lapse effects of all three surveys using poststack data to better evaluate predictions of fluid migration.

Pennington et al. (2001) discover that the pressure drop caused by production from the reservoir is communicated across the fault to at least one “little neighbor” reservoir not under production. They conclude that the two reservoirs are in pressure communication through the water sand along a route starting downdip of the producing reservoir. They propose that regional pressure drops caused the fluid in the neighboring reservoir to drop below the bubble point, causing gas to come out of solution. They further predict that the gas expansion in the neighbor reservoir increased the hydrocarbon volume, pushing oil down, ultimately to escape from the spill point into the adjacent sandstone. In this study, we use volume visualization techniques to examine Pennington et al.’s (2001) prediction in some detail.

Ezawi et al. (2012) demonstrate that a specific seismic attribute, squared instantaneous amplitude, evaluated on time slices of stacked data can be used to help identify leakage of the oil and gas from neighboring reservoirs. They confirm the initial prediction by Pennington et al. (2001): The oil and gas from a neighboring reservoir is leaking into the adjacent sandstone from where it may be lost forever if not trapped by a secondary feature. They also identify water influx at the downdip end of the producing reservoir. Squared instantaneous amplitude was used to maximize the visual effect of the amplitude changes while minimizing the visual clutter from lower amplitude “noise” in the time-lapse data.

The light-oil reservoirs in the 4500-ft sand at Teal South demonstrate typical “brightspot” behavior, as shown on the tracked horizon presented in Figure 1. Other reservoirs were not under production at the time of the surveys. Along this horizon, the strongest negative amplitudes correspond to those reservoirs, and the largest such feature is the reservoir under production, referred to in this study as reservoir A. The only well producing from this horizon intersects reservoir A at the location shown. The various reservoirs are separated by bounding faults, but they are shown by Pennington et al. (2001) to be hydraulically connected downdip by continuous sand bodies. The semblance attribute is shown in the figure to highlight the faults separating these reservoirs. The primary little neighbor in that paper is referred to here as reservoir C, and another neighboring reservoir is referred to here and by Ezawi et al. (2012) as reservoirB.

The studies by Pennington et al. (2001) and Ezawi et al. (2012) are concerned only with changes in the seismic responses between the OBC surveys of phases I and II. In this work, we extend their work to include the legacy streamer data. We refer to the time period between the legacy survey and phase I as the early production period and the time period between phases I and II as the later production period. The data are analyzed in terms of production-induced changes in the seismic amplitude response, visualized through simple attributes, and production-induced compaction indicated by seismic traveltime shifts in the overburden exactly above the producing reservoir.

The data are first prepared through crossequalization to minimize survey- and processing-related differences. We observe the remaining significant time-shift differences between the data sets that are in the area of the producing reservoir, suggesting compaction of that reservoir as fluid was withdrawn. We also compare amplitudes throughout the 4500-ft sand, using the squared instantaneous amplitude as an attribute that visually maximizes the appearance of fluid-related changes while minimizing other, noisier, differences. An outline of these steps is provided in the following sections.

The legacy streamer data and the OBC data sets were acquired and processed with different survey geometries and different processing schemes; because of this, there are many differences between the seismic data sets that are unrelated to the production process. To suppress the differences caused by different acquisition and processing, the first step in this work is to crossequalize the legacy data with the time-lapse data. Because the prestack data were not available for the legacy (streamer) data set, this study makes use only of the poststack migrated data for all data sets. The crossequalization has two primary goals: All data sets should be well aligned in terms of two-way traveltimes, and the amplitudes should be properly scaled to preserve the production-related changes in amplitude. A flowchart illustrating the sequence of key steps used in crossequalization and time-lapse analysis is added in Figure A-1 in Appendix  A, and the details are presented in a dissertation (Islam, 2014), but a summary is presented here.

The two OBC data sets are well aligned, but the legacy data set is not matched with them, partially because of an additional path in the water column. We align the legacy data set with phase I data in two-way traveltime, using a strong continuous reference reflector that is particularly flat, concentrating on areas distant from reservoir A to avoid any impact of production on the seismic response. Figure 2 shows a 3D view of an intersecting pair of lines, one from phase I and the other from the legacy streamer survey before and after alignment. A static shift of $−77ms$ applied to the OBC time-lapse data temporally aligns the surveys with the legacy data. (We used the legacy survey as the “base” because it covered a larger area and because it represents the conditions prior to production.) A small lateral shift equal to four seismic lines in the crossline direction is also required. The source of the lateral misalignment is neither known nor determined in this study, but it could be a combination of different acquisition geometry, processing steps, different seismic velocities used in processing (e.g., migration), and changes in anisotropy because of production, or it could be a simple survey entry error.

After having corrected for mistie by visual inspection, the data sets are then further aligned using a commercial software package to determine residual statics. The legacy data set is used as the reference, and a maximum shift of 10 ms between surveys is allowed. Before they are applied to the time-lapse OBC data, the residual static shifts are smoothed using a low-pass filter derived from an average of several (typically 13) traces around any trace position.

Amplitudes of the two OBC data sets are scaled equivalently, but the legacy data set is scaled differently. After correcting for the timing mistie between legacy and time-lapse data sets, we place them on a similar amplitude scale using a series of analyses to determine the optimal relationship. These analyses are detailed in Islam (2014) and include a comparison of amplitudes based on energy, considered by Rickett and Lumley (1998) to be most appropriate. After investigating a series of approaches using a subvolume to avoid production-related changes, we settled on a single scale factor to apply to the legacy streamer data, bringing it to the same amplitude range as the OBC data.

All data sets were filtered with a low-pass filter at 50 Hz, with 2 dB/octave slope to match the limited high-frequency content of the legacy data set. No low-cut filter was deemed necessary because the low-frequency content of all three data sets was so similar.

Because phase I and phase II data have already been processed to match with each other, the time shifts and amplitude scaling computed by crossequalization of legacy and phase I data are applied to phase II data as well.

To analyze time-lapse changes at Teal South, we use squared instantaneous amplitude as the primary attribute for visualization. Instantaneous amplitude is a measure of the reflection strength at every sample point, estimated by complex trace analysis. It provides a good estimate of the overall size of a signal, regardless of zero-crossings within a complicated signal (Anstey, referenced by Barnes, 2007). Taner et al. (1979) use the Bracewell (1965) description of the analytic signal together with Anstey’s idea of reflection strength to propose a method of reflection strength computation by using the complex number description of a trace. Instantaneous amplitude can be thought of as the envelope of the seismic trace, described by a slowly varying function of time that connects the peaks of seismic trace. It tends to emphasize the bright reflections and to reduce effects due to thin beds and a finite seismic wavelet. In this study, we use the square of the instantaneous amplitude to better visually emphasize the large amplitude changes and to reduce the visual clutter from low-amplitude changes, which may be related to noise and lack of repeatability. The same goal could have been achieved through the use of a nonlinear color scale.

We create difference volumes between consecutive data sets. The early production period is represented by (phase I minus legacy) and the later production period by (phase II minus phase I). During the early production period, the pore-fluid pressure was significantly reduced and gas first came out of solution, whereas during the later production period, additional gas came out of solution and the pressure continued to decline. The interplay of effects on seismic response caused by an increase in free gas and by an increase in net pressure acting on the rock frame can be complicated, and this can result in significant AVO effects (see Pennington et al. [2001] and Islam and Pennington [2015] for details). In this study, we are restricted by legacy data availability to poststack results, and we recognize that the fluid evolution and movement can be tracked by the amplitude attribute we choose to use. The 4500-ft sand exhibits a typical class III AVO response and appears as a brightspot on stacked section. For a class III sand, release of free gas and migration of light oil (with or without free gas) into areas previously occupied by water are both identified by an increase in the squared instantaneous amplitude, and water encroachment is identified by a decrease in the squared instantaneous amplitude. We use this attribute in an effort to visually emphasize the strong changes in amplitude while minimizing the impact of lower amplitude noise.

In addition to the difference in reflection strength, we use semblance to delineate faults. Semblance is a measure of the coherence or similarity between neighboring traces, and it ranges from zero for no similarity to one for identical neighboring traces. Faults are associated with low values of semblance, usually shown by dark colors in the visualizations presented here.

The final data volumes represent seismic differences in terms of squared instantaneous amplitude for the early production period and for the later production period. Specifically, we compute the squared instantaneous amplitude from each aligned seismic volume and then take the difference of that value between consecutive surveys. In addition to amplitude changes, we use the spatially filtered time shifts that would best allow the data to be aligned as an interpretation attribute. Time shifts are related to other formation changes, including compaction. The observations based on these data volumes are presented in the following sections.

There are several reservoirs within the 4500-ft sand at Teal South, as shown in Figure 1. Here, we discuss those changes observed in reservoirs A (the only reservoir under production at the time of the surveys), B, and C (the primary little neighbors referenced by Pennington et al. [2001] and Ezawi et al. [2012], respectively), as well as one “tiny” reservoir lying between reservoirs B and C.

Reservoir A is the only producing reservoir in the 4500-ft sand and was the initial target for the ERCH Consortium time-lapse studies. Figure 3 shows the 3D structure of reservoir A, and Figure 4 shows time slices at 1480 ms from the early production difference volume and from the late-production difference volume. Reservoir A exhibits time-lapse changes on both difference slices: In the early production period, the impedance of the reservoir decreases, which appears as brightening throughout the reservoir. In the later period, most of the reservoir brightens even more, but there is some dimming at its downdip end suggesting an increase in impedance. The brightening of the reservoir demonstrates continuous fluid expansion and release of solution gas throughout the time periods covered, whereas the dimming observed in the later period is probably caused by water encroachment.

Closer examination of the downdip dimming in the later production period suggests that the dimming takes place at the top of the downdip end of the reservoir (Figure 5a); this is surprising because water should be expected to enter along the lower border of the downdip end. The seismic traces here seem to be slightly misaligned (see Islam [2014] for details), an effect that may be related to reservoir compaction or to seawater velocity variation caused by spatial changes in salinity or changes in weather conditions between the two surveys. The effects of such variations in seawater velocity are not often corrected with routine processing of the data (Bertrand and MacBeth, 2003). After applying a local time shift of 1 ms to the phase II data, a new difference cube is created, a section of which is displayed in Figure 5b. This adjusted difference cube shows that dimming, and presumably water encroachment, now appears at the bottom of the downdip end of the reservoir.

Reservoir C is a small reservoir located almost 1450 ft to the northwest of reservoir A and is almost 28 ms shallower than reservoir A. The two reservoirs are separated by a north–south-trending normal fault. Pennington et al. (2001) had identified reservoir C as the primary little neighbor exhibiting time-lapse changes due to production from reservoir A as a result of “regional blowdown.” They propose that pressure is communicated through the water sand extending downdip of reservoir A where they observe continuity of the water sand on the legacy data through acoustic trace inversion.

The squared instantaneous amplitude difference data for reservoir C are presented in Figure 6 on time slices for both production periods. These suggest that fluid exsolution (liberation of free gas) is apparent throughout the reservoir in the early time period, but that time-lapse changes in the deeper section (1456 ms) are only apparent for the early time period. Shallower time slices show some brightening, and the shallowest ones (1444 ms is shown) indicate that brightening occurs through both time periods at that depth. The complicated interplay in the stacked seismic response could be due to an increase in free gas saturation, the migration of free gas to a gas cap, and/or the frame-stiffening effect. These all likely play a role in the different responses observed in the main reservoir body and in the likely gas cap, but we lack sufficient constraints to fully account for these details (see Islam, 2014; Islam and Pennington, 2015). In any case, it is clear that the hydrocarbons in reservoir C are responding to changes induced by production in reservoir A that is 1400 ft away and 28 ms deeper.

Reservoir B is another small reservoir located almost 850 ft northeast of reservoir C and 1450 ft northwest of reservoir A (measured between the centers of the reservoirs, not the edges). Reservoirs B and C are separated from reservoir A by the same fault, and another small fault, almost at a right angle to the main fault, separates reservoirs B and C. Ezawi et al. (2012) report that reservoir B is leaking based on direct observations of fluid migration similar to those reported below.

Figure 7 presents time slices at 1460 ms, showing continuous brightening of reservoir B after the start of production. Because Ezawi et al. (2012) only evaluate the later production period, they do not recognize that reservoir B appears to have been originally water saturated: It does not appear as a brightspot on the legacy data set. The evolution of this reservoir after first production in reservoir A appears to include an invasion by hydrocarbons (either light oil or oil and gas), followed by expansion of gas or gas exsolution, and then a release of gas out to the neighboring formation, as detailed below.

Figure 8 shows the 3D structure of the body of reservoir B, based on the largest changes in the seismic response, which occurred in the later production period. In this figure, we color the geobody by amplitudes of seismic data from the (a) legacy survey, (b) phase I survey, and (c) phase II survey. This progression of images shows that the body of reservoir B was filled with water with or without minor oil before production (evident from the very weak reflections on the legacy data). Reservoir B could have contained residual hydrocarbons in the reservoir at a saturation too weak to cause a brightspot. Because reservoir B is downdip from reservoir C, it is very likely that reservoir B served as a primary migration path to reservoir C leaving residual hydrocarbons in reservoir B. As production started from reservoir A, oil migrated into reservoir B if it was not present initially (apparently from reservoir C, as discussed in the following paragraph), and/or gas likely came out of solution. Ultimately, this gas exsolution increased the hydrocarbon volume until some of it escaped from reservoir B through its spill point.

Figure 9 shows that reservoir B is connected with reservoir C at its spill point by following a random seismic line that connects the downdip end of C with B. Oil escaping from reservoir C had a path to reservoir B, where it was apparently trapped, displacing the original water. With continued production, the oil in reservoir B likely also expanded and released solution gas, whereas reservoir C may have contributed yet more oil and gas into reservoir B. Because of the limited volume of reservoir B, it could not store all of the hydrocarbons, and oil and gas appear to eventually leak from reservoir B into the neighboring sandstones. Figure 10 summarizes a possible path for hydrocarbons leaking from reservoir C, first into (water-saturated) reservoir B and eventually through its spill point to the neighboring formation. Even though no reservoir simulations have been performed using neighboring reservoirs B and C, this scenario seems plausible, but we recognize that it is not unique.

There is one additional tiny reservoir, surrounded by faults on all sides, neglected in the various studies to date because of its small size. This reservoir responds oddly to pressure changes in reservoir A. In the early production period, it brightens, but later, it dims. The location and structural complexity of this reservoir suggest that any effects of pressure change in nearby reservoirs will be associated with effects in this reservoir. Figure 11 shows a single image containing two time slices (1448 and 1480 ms) displayed together using transparency to allow other reservoirs to be visible. The tiny reservoir appears as a local high on seismic sections, surrounded by faults as evidenced by low semblance.

This reservoir appears to provide an element of complication into pressure changes and their distribution with time. Random lines (examples are shown in Figure 12) suggest that the most likely paths for pressure communication to the neighboring reservoirs pass through the tiny reservoir, first through B and then through C. It is difficult, however, to propose a simple scenario that involves dimming of the tiny reservoir during the later production period without proposing that water somehow replaced the hydrocarbon originally in place (the presence of hydrocarbons can be identified by the bright spot on the baseline survey), perhaps through dynamic considerations.

Production from a reservoir, especially an unconsolidated overpressured reservoir, induces time-lapse changes not only in seismic amplitude but also in the arrival times due to changes in thickness and velocities in the reservoir as well as in the overlying and underlying formations (Barkved and Kristiansen, 2005; Hawkins et al., 2007). These changes in arrival times are referred to as time-lapse time shifts (Hatchell and Bourne, 2005).

For compacting reservoirs, time-lapse time shifts within the producing reservoir and outside the producing reservoir are usually of opposing signs. A decrease in pore pressure implies an increase in effective pressure that may result in a decrease in reservoir thickness (also called reservoir compaction) and an increase in seismic velocity. (Any associated decrease in porosity is probably far too small to detect directly.) Each of these effects will lead to a decrease in the traveltime through the reservoir. In some cases, the “arching” support of the overburden can result in extensional stresses in the surrounding rock, and seismic velocities may decrease above (and below) the reservoir, delaying reflections from outside the reservoir. Maximum traveltime delays may be expected just above the depleting reservoir.

With limited accuracy, we can presume that the spatially filtered traveltime shifts obtained as part of the crossequalization described earlier are the results of overburden stretch caused by reservoir compaction.

Figure 13a shows the time shifts between the legacy and phase I data, using inline 3500. Legacy data are set as the reference. Green indicates a delay in arrival times from legacy to phase I, and blue indicates early arrival. The figure shows that there are negligible time shifts everywhere other than at the location of reservoir A in which significant delays in arrival times are observed in the overburden above it. Within and below the reservoir, the delay is less than above. Stretching of the overburden because of reservoir compaction delays the reflections in the overburden. Compaction of reservoir A and its possible decrease in reservoir thickness and increase in velocity act to counter the delay caused by overburden stretching, and we observe much less delay within the reservoir and below the reservoir. Time shifts are generally cumulative and do not fluctuate, but in Figure 13a, we see that the time shifts disappear in places. A closer look at the figure shows that time shifts become negligible as the seismic amplitudes reach zero. The time shifts are based on correlation between the two data sets, which becomes meaningless when the amplitudes are negligible. Although the accuracy of the time shifts is limited by data quality, they are consistently associated with the reservoir structure and location. We see negligible time shifts at locations away from the reservoir, and we have confidence that these time shifts are associated with the reservoir. The mechanism that can best explain this is reservoir compaction.

Figure 14 shows time shifts on a time slice with similarity to highlight the static shifts with respect to the faults separating the reservoirs. A display of squared instantaneous amplitude is added to show the relative position of reservoir A.

Figure 13b shows the time shifts observed during the late production period. Phase I data are set as the reference for this case. The time delay apparent in the overburden during the early production period is absent in the later production period. One possible explanation is that the stress arch in the overburden had developed completely by phase I, with no additional expansion afterward. However, a time delay is apparent in the later production period beneath the producing reservoir, perhaps caused by stretching in the underburden, as observed elsewhere by Hawkins et al. (2007). Because the OBC data have been processed to match for time-lapse analysis, the lack of time shifts between the OBC data sets could also be a result of those processing steps. Because a primary focus of the time-lapse study was reservoir A, less attention may have been paid to the underburden, perhaps explaining the “noisy” time shifts there. Although the accuracy of the time shifts is limited by the data quality, the results suggest that reservoir A has undergone some compaction as a result of pressure depletion. Variations in seawater velocity because of changes in salinity and weather conditions are also considered a primary source of time shifts in time-lapse analysis. Bertrand and MacBeth (2003) perform seismic modeling and discuss the effects of spatial and seasonal variations in seawater velocity that might not be corrected by routine data processing. They demonstrate that such changes in seawater velocity may introduce time shifts between the time-lapse data sets. The time shifts observed at Teal South do not seem to be associated with the variations in seawater velocity for two reasons: First, all the time shifts observed at Tea South are very localized and directly above the reservoir. Second, the time shifts are quite significant for only an 85-m water depth.

Time-lapse data from the Teal South oil field was analyzed to explore the effects of production on the producing reservoir and neighboring, unproduced, reservoirs. The reservoirs are high-porosity, unconsolidated, and overpressured sandstones, containing light oil just above the bubble point, which respond strongly to fluid saturation and pressure changes. Three sets of seismic data were acquired over the field at three different times. The first set of data, the legacy data, was acquired using streamer cables prior to production, and the other two data sets (time-lapse data phases I and II) were acquired using OBCs after eight and 30 months of production, respectively. The legacy streamer data were crossequalized in this study with the time-lapse OBC data to allow studies to include the early period of production.

To define the time-lapse changes in amplitude and to emphasize the visual impact, the difference in squared instantaneous amplitude is computed between consecutive surveys, and two difference volumes are generated: one between the legacy and phase I surveys and the other between phases I and II. These difference volumes support the results from previous studies and add the following observations due to greater emphasis on detail allowed by the visualization technique used to include the legacy data set:

1. Water encroachment at the base of the downdip end of the producing reservoir A is evidenced by a dimming of amplitudes in the later production period.

2. Neighboring reservoirs that are separated from the producing reservoir by one or more faults show time-lapse changes that imply pressure changes can be communicated easily through water sands that are juxtaposed across faults.

3. Reservoir B, previously reported as a leaking reservoir, is only developed as an oil reservoir after production from reservoir A, as a result of fluid expanding from reservoir C and leaking through its spill point. Specifically,

   • Legacy data show that the sand of reservoir B was water filled at the time of first production in reservoir A.

   • We propose that the pressure drop from production in reservoir A is communicated to reservoir C, which was an oil reservoir at the time of discovery. Because of this pressure drop, gas in reservoir C came out of solution and the hydrocarbon volume increased.

   • Oil within reservoir C is then pushed past the spill point toward reservoir B where it is trapped again. With the continued drop in pressure, more oil is added to reservoir B, where it also experiences expansion.

   • After some period, oil in reservoir B reaches its spill point, escaping into the surrounding rock.

4. Some confusing observations can be accounted for by complicated, but reasonable, scenarios:

   • For example, different intensities of brightening in different portions of reservoir C at different times may be explained by a trade-off between the effects of frame stiffening and fluid expansion (and local variations in lithology), resulting in negligible time-lapse changes on stacked data in the oil zone; but in the gas cap, the fluid effects dominate.

   • Two possible paths for pressure communication between reservoirs A and C are established. A tiny reservoir, surrounded by faults, is identified between these two reservoirs, demonstrating time-lapse changes on both difference volumes, although those changes include unexplained dimming in the later production period. The communication of pressure changes likely passed through the tiny reservoir to B and on to C.

5. Possible compaction effects are identified through traveltime shifts, indicating compaction of reservoir A that resulted in stretching of the overburden during the early production period and of the underburden in the later production period.

The original project at Teal South was made possible due to the foresight of geophysicists at Texaco and with ERCH; we appreciate the extraordinary efforts made by those people and the members of the ERCH Consortium. The legacy data set was provided by Diamond Geophysical. The software used in this study for analysis and display was OpendTect, provided by dGB Earth Sciences, whom we gratefully acknowledge. We very much appreciate the input provided by the anonymous reviewers and the associate editor, which significantly improved the quality of the manuscript.

Dr. Nayyer Islam received a B.S. in geologic engineering from UET, Lahore, Pakistan, and joined that institution as a lecturer in 2008. He held that position until coming to Michigan Technological University, where he received M.S. (2011) and Ph.D. (2014) degrees. Currently, he is working with BP America as a petrophysicist. His research areas of interest include reflection seismology, rock physics modeling, and petrophysics (i.e., seismic petrophysics).

Mohamed A. Ezawi received a B.S. (2008) in geophysics from the University of Tripoli, Libya, and an M.S. (2012) in geophysics from Michigan Technological University. He has been a seismic processing specialist at Schlumberger’s PetroTechnical Services (PTS) segment in Libya since late 2012, with experience in surface and borehole seismic data. He also served as the manager of the Quality Steering Committee at PTS. His research interests include seismic attribute analysis and seismic petrophysics.

Wayne D. Pennington received degrees in geology and geophysics from Princeton University, Cornell University, and the University of Wisconsin-Madison. He is the dean of the College of Engineering at Michigan Technological University, where he has worked since 1994. Previously, he was at Marathon Oil Company’s Petroleum Technology Center in Littleton, Colorado, following several years on the faculty at the University of Texas at Austin. He has served as the first vice-president of SEG, as the president of the American Geosciences Institute, and as a Jefferson Science Fellow at the U.S. Agency for International Development.