- Copyright © 2003 Society of Exploration Geophysicists
Rapid advancements in multicomponent acquisition methods and processing techniques have led to numerous applications for converted wave (C-wave) data that are increasingly used for exploration and exploitation of oil and gas. However, the increased prevalence of multicomponent seismic applications means that interpreters must face many difficult challenges: how to register P-wave time to C-wave time, determine the best methodology for interpreting C-wave data, and/or how to apply the C-wave interpretation in assessing the risks of exploration and exploitation prospects. This paper addresses these issues and attempts to guide the interpreter through the multicomponent data interpretive process with numerous examples from two East Cameron gas fields in the Gulf of Mexico.
In the fall of 1999, PGS acquired the first commercial multicomponent 3D survey in the Gulf of Mexico. BP licensed the data in early 2000. The survey, approximately 80 km2 full fold, covers BP's East Cameron 261 and East Cameron 265 fields. The primary purpose of the survey was to address data degradation due to the presence of gas in the shallow sediments. Seismic imaging over these fields, which have more than 30 gas-bearing reservoirs in depths of less than 1000 ft to over 15 000 ft, has been hampered by severe velocity push-downs and gas cloud effects in the conventional P-wave data. The new 4C data, particularly the C-wave data, have unquestionably improved imaging in and below gas cloud areas. The fields were subsequently re-evaluated, and interpretation of new data resulted in a much more improved understanding of the hydrocarbon-bearing reservoirs which led to five consecutive successful exploitation wells in East Cameron 261.
Acquisition and processing
While it is not the intent of this paper to go into the details of acquisition and processing, the reader may be interested in how the survey was acquired and processed in order to compare to other multicomponent data with different acquisition designs and/or processing flows. The survey was acquired with four-component ocean-bottom cables, which were deployed without tension and not dragged. A wide-azimuth patch was used with shot-lines orthogonal to cables. Specific key acquisition parameters included:
Dual cable orthogonal patch acquisition
Receiver direction = north-south
Shot direction = east-west
Group interval = 50 m
Receiver line interval = 437.5 m (EC 265), 550 m (EC 261)
Source interval = 37.5 m
Source line interval = 420 m (EC 265), 540 m (EC 261)
Source depth = 7.5 m (EC 265), 5 m (EC 261)
Record length = 11 s
25 × 25 m bin (acquired)
It is important to note that two of the four blocks were acquired with narrower receiver and source line intervals. The results reveal that: (1) there is better shallow C-wave data coverage (smaller notches) within the narrower receiver and source line interval acquisition area as expected, and (2) the overall C-wave data quality improved noticeably with better reflection continuity and greater resolution. Examples from both areas will be pointed out later in the paper.
In seismic data processing, the following general sequence of preprocessing steps was first applied to all components: reformatting of field data, geometry assignment, designature of the air-gun array wavelet and instrument response, and datuming. On the P-wave data, hydrophone and vertical geophone components were combined to reduce water-bottom reverberations. Next gain recovery, deconvolution, and residual statics were applied. They were followed with prestack time migration for both PZ and PS data.
Background field information
East Cameron 261 Field (100% BP W.I.) and East Cameron 265 Field (50% BP W.I.) are offshore Louisiana in 160 ft of water, approximately 175 miles southeast of Houston, Texas. These gas fields were discovered in the 1960s, and the majority of the known reserves were produced during the 1970s-early 1990s. Recent production had fallen off to levels where economic viability of the fields was in question. Successful recent exploitation drilling has significantly increased reserves and the production rate while extending the life of East Cameron 261. The fields combined have produced over 650 BCFG to date.
Field pays range in depth from 1000–15 000 ft (with more than 30 producing reservoir sands), and they are typically represented by bright amplitudes on the P-wave seismic data. The reservoirs are Plio-Pleistocene in age and were deposited in a near-shore environment in predominately prograding deltaic sequences. The traps are usually faulted three-way or four-way structures in a growth fault environment. Hydrocarbon-bearing stratigraphic traps, although not common, are also present. Typical porosities range from 18% to slightly over 30%. Permeability is generally high (100s of millidarcies). Most production is in normally pressured sediments, with only the deepest 15 000 ft sand being overpressured. Figures 1 and 2 show 3D views of East Cameron 261 and East Cameron 265.
P-wave to C-wave time registration
Because of the lack of workstation software tools designed specifically for C-wave interpretation and industry's general inexperience in C-wave interpretation, geophysicists often face many challenges with multicomponent seismic data. The first and possibly the most important hurdle is caused by the slower arrival times for C-waves than for P-waves for the same subsurface horizon. The process of correlating P-wave reflectors to C-wave reflectors from identical subsurface horizons, called registration, is critical for interpreting multicomponent data. For example, if the average VP/VS ratio of sediments down to a reflector is 3, the C-wave time for that reflector will be twice that of P-wave time.
In this study, numerous attempts were made to visually correlate packages of reflectors from P to C after compressing the C-wave time to approximately match the P-wave time. More often than not, visual correlations were misleading because of differences in reflectivity responses from P to C. Also, a PS synthetic seismogram was created using dipole sonic and density log data, and an attempt was made to tie the PS synthetic seismogram to a C-wave section. This method also was unsuccessful because of the limited depth and poor quality of the dipole sonic log. Additionally, a “pseudo-PS synthetic seismogram” was created with velocities calculated using regression equations determined from P and S log data elsewhere on the Shelf Gulf of Mexico. This result also did not provide an adequate synthetic tie to allow us to confidently register P to C.
After much frustration in trying various methods to register the time from P to C, a new method was developed and successfully implemented into the project. This new technique utilizes fault planes to register the time; hence, it is called “the fault plane registration method.” The entire procedure was done on the workstation interactively using Landmark's SeisWorks and TDQ applications. Other similar software can just as easily be utilized. Described below is a detailed step-by-step explanation of this technique.
Load P-wave and C-wave seismic data (2D or 3D) into a workstation. Ideally, the C-wave data should already have undergone some initial squeezing to enable approximate visual correlation of seismic features from P to C; however, this is not imperative.
Carefully interpret the most obvious and visible faults, key horizons, and any peculiar features on a representative P-wave line.
Display the same interpretation on the corresponding C-wave line. Depending on how well the initial squeeze was done, there may be some noticeable discrepancies on fault placements and horizon interpretation on the C-wave line. The discrepancies should typically be on the order of a few milliseconds to a couple of hundred milliseconds vertically high or low if a reasonable initial squeeze was applied to the original C-wave data. Aside from the inherent expected differences, such as those resulting from gas cloud areas, the discrepancies are likely attributable to the imprecision associated with the initial squeezing of the C-wave data.
Choose a fault, preferably one that cuts from shallow to deep, and a fault that is well imaged on both P and C. By placing the cursor on the P-wave fault plane (note the P-wave time) and moving the cursor up or down along the same trace to the C-wave fault plane (note the C-wave time), one can accurately determine the amount of “misregistration” on C. Repeat this step from shallow to deep along the fault plane to get representative P and C time pairs.
Repeat step 4 at other locations (along other fault planes or any other events that can be easily correlated). In a small survey area, one function may work fine, but in large survey areas or areas where the geology is very complex, more functions will probably produce better results.
Apply the derived P and C time pairs to “depth convert” the C-wave data. Now, run a reverse operation (depth to time) using a one-to-one function to bring the C-wave data back to time.
One key advantage of the fault plane registration method is that an interpreter can do all the steps using a workstation interactively and immediately be able to view the results. The quality control is also done quickly by displaying the P-wave interpretation on the C-wave data and by noting the discrepancies of fault placements and interpreted horizons. Oftentimes, if the time registration is not done properly, P-wave interpreted horizons will not follow the C-wave reflectors. In those cases, the applied function needs to be revised and rerun. Figures 3 and 4 show application of this technique to the East Cameron 4C data set. Figure 5 demonstrates an application of a C-wave instantaneous phase section to further adjust the fault interpretation.
Delivery of the final processed seismic volumes from PGS occurred in early January 2000, a month prior to BP's drilling program within East Cameron 261. Obviously, there was not sufficient time for interpretation of the data before drilling. There were, however, some key factors that allowed us to drill a well with such a short time for interpretation and data analyses. First, comprehensive work had been done on previously acquired streamer 3D data in the area. Secondly, the P-wave volume of the new 4C data was processed first and delivered to us earlier, and further work was done using this new P-wave data prior to the delivery of the C-wave data in early January. Most of January was spent on the P to C time registration described earlier and interpretation of the C-wave data.
Careful planning and streamlined interpretation of the C-wave data was imperative due to the drilling schedule. Some key interpretation strategies that enabled us to meet the tight schedule were: (1) Most of the faults previously interpreted using the P-wave data were directly used for the C-wave data interpretation. This was possible because of successful P to C time registration. (2) All P-wave based interpreted horizons were directly displayed in the C-wave volume and were properly utilized. This allowed immediate identification of misinterpreted zones such as those in the gas cloud area. Even in correctly interpreted areas, the P-wave horizons did not align perfectly on C-wave peaks, troughs, or zero-crossings. Some causes for this misalignment are attributed to the inherent differences in P and C acoustic impedances, phase changes, lateral velocity variations, and imprecision of P to C time registration. Depending on the specific area and the cause of misalignment, some horizons were reinterpreted using the P-wave interpretation as a guide while others were tracked for maximum peaks, troughs, or zero-crossings using a small time analysis window to recreate new horizons that honor the C-wave data. (3) Finally, using the C-wave horizons produced from the work discussed above, various attribute maps were created which were further analyzed via appropriate cross-plots.
With the interpretation methodology described above, we were able to eliminate much duplication of effort and meet the tight deadlines imposed on our drilling program.
Acquiring new data at the late stage of prospect generation may perhaps be viewed as a huge burden, particularly if there is a set drilling schedule. However, in this project, the incorporation of the new 4C data into our interpretations undoubtedly contributed to the success of our drilling program.
Application: 3200-ft sand
It is a common knowledge that P-wave energy can be scattered and attenuated when traveling through gas-saturated reservoirs. This is particularly a problem when the objective zones lie below shallow gas sediments, which degrade imaging of deeper features. The 3200-ft sand is a typical Class III sand and it is fairly continuous reservoir, generally 20 ft thick. Approximately 40 billion ft3 of gas has been produced from three wells in two simple three-way fault traps. This reservoir lies beneath gas sands at 1000 ft, 1700 ft, and 2800 ft. The P-wave imaging of this reservoir is very poor because of distortion caused by the overlying gas sediments and overall background gas in the vicinity. The discontinuity of reflectors, velocity push-downs, and general structural ambiguity are quite obvious on the P-wave data. The 3200-ft sand P-wave time structure map also reveals problematic areas where there are lows in the middle of what should be an overall structural high trend. The C-wave imaging of this reservoir, on the other hand, is excellent. The gas cloud is not present and, instead, crisp continuous reflectors are in place along with clearly defined faults previously not well imaged on the P-wave data (Figure 6). A new 3200-ft sand map constructed using the C-wave data shows the three-way fault trap configuration very distinctly (Figure 7).
An additional step was taken to more precisely outline the gas cloud area that had an impact on this reservoir. A new map was created by taking the difference between the P and C (postregistration) two-way times from the 3200-ft sand time structure maps (Figure 7). The map shows that the time delta increases drastically from a just few milliseconds in the background to a few tens of milliseconds in the gas cloud area. Thus, the amount of velocity push-down on the P-wave volume due to shallow gas can be estimated using this map.
The amplitude extraction of the reservoir sand was created using both the P-wave and C-wave volumes. The P-wave amplitude extraction map (Figure 8) displays several bright anomalies that are indicative of gas. It is also apparent from well penetrations and volumetric calculation of the reservoir that a sizable portion of the gas-charged area does not appear bright on the P-wave amplitude map because of data deterioration due to shallow gas. The C-wave amplitude extraction map (Figure 8), however, shows most of the gas-charged area to be noticeably dim. In fact, the outline of three-way fault traps on the C-wave time structure map matches very well with the outline of dim areas on the C-wave amplitude extraction map. This observation is one of many examples cited that demonstrates the pattern of bright on P and dim on C responses for gas sands.
A plot of amplitude values from the P-wave and C-wave amplitude extraction maps was generated for further analyses. The entire mapped area was used as input to distinguish anomalous areas from the background. A zone was selected on the plot that corresponds to a cluster of points that represents bright amplitudes on P and dim amplitudes on C. These chosen points were then transferred into map view, which revealed that most of the points lie on the gas-charged area. (Figure 9). To make a contrast, a different zone was selected; only those points that are bright on P and C were highlighted. The corresponding map view exhibited that these may in fact represent “false” bright spots caused by lithologic anomalies and not by fluid (Figure 10). The key learning point here is that in this normally pressured Class III environment, gas sands have a characteristic seismic response of bright on P and dim on C.
Application: 9000-ft sand
The 9000-ft sand (Class II), locally called the B2 sand, is typically 80–130 ft in thickness but, in places, can exceed 250 ft. The sand is deposited as part of the massive, thick deltaic sequences seen throughout the region. The traps for this productive sand are highly faulted three-way and four-way closures. The structure is set up by a series of antithetic faults to a major down-to-the-south growth fault. There are several gas-charged traps for this sand, but the areal extent of each is much smaller in comparison to that of the 3200-ft sand discussed earlier.
The top of the B2 sand was independently mapped using both the P-wave and C-wave volumes (Figure 11). In constructing these time structure maps, faults interpreted from the P-wave data were primarily used. The overall appearance of both structure maps is very similar; however, there are subtle but very significant differences. Notice the shifts of the structural highs from one map to the other (areas circled in yellow on Figure 11). Well ties to these maps reveal that the C-wave map is, in fact, more accurate than the P-wave map. Moreover, with the anticlinal axis of the overall structural trend in mind, the C-wave map makes much more geologic sense. Representative time slices at this depth using P-wave and C-wave volumes (Figures 12) succinctly illustrate the trap configuration. (Note that the boxed southeastern corner corresponds to the B2 map area.) It is also observed that while the overall structure of the mapped area may be better defined on the C-wave time slice, the P-wave time slice visibly contains more details with better resolution. The C-wave data at this depth have narrower frequency bandwidth and, therefore, subtle faults and fine details are sometimes not very well illuminated.
The P-wave and C-wave amplitude extraction maps of the B2 sand (Figure 13) reveal a number of interesting anomalies. The areas outlined in black on the P-wave amplitude extraction map are known gas areas that are characterized by bright amplitude anomalies. In contrast, the same areas on the C-wave amplitude extraction map show that they have consistently dim amplitude anomalies identical to the conclusions derived from the 3200-ft sand study. Applying the same concept, several new prospects were generated, two of which are outlined in red.
A representative P-wave and the corresponding C-wave seismic line over Texaco well EC265 A5 (Figure 14), a B2 sand gas producer, are shown in Figure 15. Additionally, another P-wave and its C-wave equivalent seismic profile over Shell well EC266 1, a B2 wet sand penetration, are presented in Figure 16. Notice the consistent pattern of a bright on P and dim on C combination for the gas-charged B2 sand and a dim on P and bright on C combination for the wet B2 sand at the respective well locations. With a careful observation of the B2 reflector in both P and C (Figure 16), one can distinctly see the same pattern from fault block to fault block. (Known gas and prospect fault blocks are bright on P and dim on C; other fault blocks are exhibiting just the opposite behavior indicating wet sand as proven by the EC266 1 penetration.) A detailed study of additional pay zones throughout the field resulted in similar conclusions.
AVO modeling of the B2 sand was performed and substantiated that the amplitude strength of the PS reflector at the shale/sand boundary is weakened with the introduction of gas into the sand (Figure 17). It is presumed that the slope of this converted P-wave to S-wave AVO is primarily dependent on the S-wave impedance contrast at the interface. It is concluded that the rock properties in the study area is such that, even in a normally pressured Class II environment, the gas sands have bright on P and dim on C signatures.
Figure 18 shows a plot designed to separate gas sands from wet sands. The area that represents bright amplitude on P and dim amplitude on C is selected on this plot, and those points are once again transferred into map view. In contrast to the 3200-ft sand plot, the y-axis now measures the differences of P and C amplitude values. This type of plot exhibits a more linear trend of points, which facilitates choosing an analysis window. The obvious observation from Figure 18 is yet again that only the gas-bearing anomalies and identified prospects are highlighted on the map view when bright on P and dim on C amplitude points are selected from the plot.
More examples of interpreted and uninterpreted P-wave and C-wave seismic lines and time slices are presented here. Figures 19–22 are from data with wider source/receiver spacing; Figures 24–26 are from narrower source/receiver spacing. As mentioned, it is quite apparent that the overall quality of the C-wave data is noticeably improved on data with enhanced acquisition parameters. Also the reflectors are more continuous with broader bandwidth and higher resolution.
Some interpreters may just glance at the inferior quality of the C-wave data and decide to only use the P-wave data for prospecting; there is additional valuable information to be found in the C-wave data. Careful mapping and keen observation of the differences in P and C responses are absolutely essential in maximizing the uses of the multicomponent data.
As an example, let us analyze well EC261 A5ST2, one of the five recent discovery wells. Figure 19 is a P-wave and C-wave comparison of a line across this well. The well was drilled for multiple targets, with the primary target being the Valvulina (H)-4 sand. While the quality of the C-wave data may not allow accurate structural mapping, the overall C-wave response is dim for this reservoir. Incidentally, a distinct flat spot is seen on P-wave at the main part of the field pay, which is noticeably absent on C-wave. The strong P-wave flat spot response is primarily due to the large P-wave velocity contrast between gas and water-saturated sands. C-wave reflectivity, on the other hand, is due to contrasts in S-wave velocity and density. The density contrast alone is not sufficient in this case to produce an observable flat spot on the C-wave. An interesting seismic expression of the gas-water contact zone is also seen in Figure 20. Using careful observation, one can see that the P-wave Valvulina (H)-4 reflector goes from bright to dim and the corresponding C-wave reflector does just the opposite at the contact zone. Figures 21 and 22 show two more examples of P and C gas sand signature comparisons. Figure 21 compares P-wave and C-wave seismic profiles across the Valvulina (H) reservoirs; and Figure 22 is a representative time slice comparison of the Lenticulina gas sand. A 3D view of the Valvulina (H)-4 structure with a P-wave amplitude overlay is shown in Figure 23. The gas/water contact seen here corresponds to the flat spot on Figure 19.
A marked improvement of the C-wave data quality is quite evident with narrower source and receiver line spacing. A dip line over the gas cloud area, a strike line over the gas cloud area, and a dip line outside the gas cloud area are shown in Figures 24, 25, and 26, respectively. The superiority of C-wave imaging in the gas cloud area is quite obvious, allowing the interpreters to accurately pick faults, construct maps, and tie wells. Good quality C-wave data, such as these, not only allow better mapping of areas that are poorly imaged in P-wave, but also enable more technical analyses like those described in this paper. The authors firmly believe that considerable value was attained through the incorporation of multicomponent seismic interpretation in our exploration and exploitation efforts. Perhaps with even better C-wave data quality, the industry will begin to employ wider and more sophisticated applications of multicomponent data in the future.
Several conclusions are made from this study. Using the fault plane registration method, the original C-wave times are squeezed successfully to match the P-wave times. Without a doubt, the C-wave data within and below the gas cloud area yield far superior subsurface images than the P-wave data. This allows for more accurate well ties and hence better structure maps. Additionally, a comparison of the P-wave and C-wave amplitude extraction maps reveals that gas sands in the study area (both Class II and Class III) are consistently bright on P-wave and dim on C-wave. This dimming observed on C-wave may be an important criterion for distinguishing gas sand reflectors verses “false” bright spots.
“Useful approximations for converted-wave AVO” by Ramos and Castagna (Geophysics, 2001). “Imaging through gas clouds with converted waves” by Cafarelli et al. (World Oil, 2001). “Anisotropic 3D prestack depth imaging of the Donald Field with converted waves” by Nolte et al. (SEG 2000 Expanded Abstracts). “Characterizing reservoir by using jointly P- and S-wave AVO analyses” by Jin (SEG 1999 Expanded Abstracts). Combining P-wave and S-wave seismic data to improve prospect evaluation by Hardage (Bureau of Economic Geology, The University of Texas at Austin, Report of Investigations 237, 1996). “Amplitude-versus-offset variations in gas sands” by Rutherford and Williams (Geophysics, 1989).
The authors thank BP management for allowing publication of this paper. We are grateful for the enthusiasm and the strong support for the project given by Steve Decatur, the asset manager at the time of the study. We are indebted to the members of BP's Upstream Technology Group, particularly Hans Sugianto, Dan Ebrom, and Jerry Beaudoin for their technical assistance, constructive comments, and discussions. Additionally, sincere appreciation is given to BP's Wendy Kurek for help preparing the figures. Finally, we thank PGS for its great work in the acquisition and processing of these 4C data and its cooperation in delivering the data early enough to have an impact on the drilling program.
Coordinated by Rebecca B. Latimer