Since the advent of digital seismic, amplitude analysis has assisted resource exploration. Though the earlier results using amplitude analysis were promising, it was gradually realized that more awareness was required for its accurate implementation. Consequently, the technique application was not always successful, casting doubt on its validity. Nowadays, the limitations are better understood, and the calibration of amplitude anomalies before and after drilling is standard.
Near-, mid- and far-angle stacks clearly detect one type of amplitude anomaly, predominantly amplitude-versus-offset class III (see the June 1999 Geophysical Corner for more information on AVO), which exhibit a negative reflection coefficient at the near-angle stack, which becomes more negative at the far-angle stack range. Class III anomalies are primarily associated with unconsolidated sediments, such as high-porosity sandstones in the Gulf of Mexico or the Barents Sea. Nonetheless, near-, mid- and far-angle stack generation should be properly documented and constrained to a specific zone of interest. As interpreters, we should carefully quality control partial stacks generated at the processing center since, most of the time, partial stack generation is done without any specific prospect in mind, and the product is later examined by seismic interpreters, who will decide if a detailed AVO analysis is required. However, what happens if the definition of the angle range bypasses amplitude anomalies outside the defined angle distribution?
In this article, we aim to show the limitations of using partial angle stacks to interpret an amplitude anomaly in the Gulf of Mexico. As we examine the geophysical concepts from a practical perspective, we focus on showing how partial angle stacks are created and present how partial angles are used. The examples are from the Gulf of Mexico using two datasets, one with the partial angle stacks generated and provided without angle definition, and the second pre-stack seismic gathers in the offset domain, which are transformed to the angle domain and stacked. Both will help to discuss the importance of carefully choosing specific angles based on the geologic objective and how fit-to-purpose stacks can change the perception of a seismic interpreter based on amplitude anomalies.
Partial angle stacks are created by choosing a range of angles in common-mid-point angle gathers. Figure 1 shows a CMP gather and its transformation from the offset domain to the angle-of-incidence domain. The transformation of pre-stack seismic gathers from the offset to angle domain can be accomplished using the established mathematical relationship between the offset, angle of incidence and the velocity information field interval.
The Gulf of Mexico is a well-known oil and gas basin; herein, two datasets are used to demonstrate how the appropriate choice of angle range benefit interpreters. We included borehole data and a rock-physics model to approximate an S-wave data since synthetic gather generation requires both velocity profiles (compressional and shear).
The first dataset has three partial stack angle stacks; near-, mid-, and far-angle; each partial stack has unique characteristics in terms of amplitude and frequency distribution, which helps to visualize the structural and stratigraphic features in seismic. However, once the stacks are generated, the angle distribution is fixed for all depths, casting doubt on the validity of the angle distribution and target. In this dataset, downloaded from the National Archive of Marine Seismic Surveys, we identified an amplitude anomaly that we calibrated with a borehole. Figure 2a shows an isometric view to portray the structural framework of the field and the amplitude distribution in the frequency domain as well as extraction of minimum amplitude and on the surface of interest. The drilled amplitude anomaly shows a good to excellent conformance of amplitude decrease associated with the structural high, and the production of the borehole confirmed substantial gas production.
Analyzing the amplitude maps and the strengthening of a negative anomaly on the far stacks, we interpret the amplitude anomaly corresponds to a class III. Nonetheless, which angle distribution was used to create a far-angle stack? We do a disservice to ourselves by not including the angle range to which each angle-stack corresponds.
Does the angle distribution hinder other amplitude anomalies? Can forward modeling provide further insight into the matter? Is there an optimum angle distribution?
To answer these questions, borehole calibration, modeling, and analysis help quantify the AVA effect (figure 3). The amplitude analysis in figure 3b shows a well-defined AVA class III. The amplitude starts weakly negative amplitude for near-angle traces (figure 3b), becoming more negative for the far-angle traces. Green boxes suggest the range of angle stacks where the amplitude trend is displayed, and the effect is synthesized. Looking into the amplitude values in the seismic survey, we can infer that the ranges used for creating the angle stack are 0-14, 12-26 and 24-36 degrees, represented by the cyan triangles.
The present angle range distribution holds valid for strong amplitude anomalies. However, amplitude anomalies with a different distribution, for example, a strong negative amplitude in near angles, develop slightly negative at far angles of incidence. To present this hypothesis, a three-layered model with different parameters shows a case where the present angle distribution will no longer hold valid.
As it can be observed, between the two cases, there are slight lithological and fluid content differences; both portray the case of a good reservoir rock with high hydrocarbon content, yet, in case 2, the amplitude anomaly remains slightly negative up to 40 degrees. The presented synthetic example is like the second dataset available, where seismic offset gathers are available along borehole wireline data, including gamma ray, resistivity, neutron porosity, bulk density and compressional sonic.
Combining seismic and borehole data, we show how the incorrect angle definition could be enough for one amplitude anomaly but insufficient for a nearby amplitude anomaly. In figure 5, the seismic gathers in the offset domain are transformed to the angle domain using the required mathematical transformation. It can be observed that the range of angles reaches a different amplitude minimum at different angles (white arrow), which suggests the need to separate the angle stack definition for both amplitude anomalies. Between the two amplitude anomalies, there are 750 meters (based on the resistivity log response), and changes in the amplitude due to compaction are discarded. However, stacking using the shallow anomaly as a reference fades the amplitude anomaly for the more profound amplitude anomaly. The origins of said difference could be explained in the forward modeling above; the amplitude changes are most likely due to different reservoir properties, either porosity, clay content or hydrocarbon saturation.
Analyzing the amplitude information of the shallow anomaly (about 1.9 seconds), the amplitude values at near offsets compared to the far offsets are entirely different, which suggests the presence of an amplitude anomaly. The AVO analysis for this anomaly suggests an AVO class III, a negative intercept and a strong negative gradient. In contrast, amplitude anomaly around 2.5 seconds has a stronger negative intercept, i.e., a more negative amplitude value at the near offsets and a more negative amplitude at the far offsets. Nonetheless, the gradient is subtle in comparison with the shallow anomaly. The AVO plot of the deep anomaly shows an AVO class III with different distribution of angles, if we compare the AVO effect for the first 38-40 degrees the shallow anomaly indicated with the red line has a stronger AVO effect than the deep anomaly shown with the green curve. Nonetheless, we ought to remember that the green amplitude curve initially has a more extended angle distribution, whereas at far angles, the amplitude is more negative.
Transforming into partial angle stacks, the angle gathers presented in figure 5c, using the boxes defined to show the method’s weakness, two partial angle stacks are generated with the near-angle stack from 4-12 degrees and far angle stack from 28 to 38 degrees. An interpretation following the most negative for the shallow (red interface) and deep (green interface) anomaly is generated, and amplitude extraction on both surfaces is calculated. Each partial angle stack amplitude extraction on both near and far angles shows opposite results for both anomalies, as seen in figure 6. While the shallow surface displays a sharp increase in amplitude response from near angle to far angle (figure 6a and b), the deeper anomaly shows subtle to no amplitude change (figure 6c and d), which an interpreter could analyze as a nonexistent AVO/AVA anomaly. A solution to this problem would be to change the angle distribution for the deeper anomaly and have two sets of partial angle stacks.
The generation of partial angle stacks benefits from tailored studies, which include using borehole data, petrophysics, rock-physics models and forward modeling. Using partial angle stacks is beneficial for interpreting the geological strati-structural information in the form of amplitude variation with angle of incidence. Such an exercise must be carried out with caution as different geological causes could translate into ambiguous amplitude variations with the angle of incidence. To avoid such complications, it is important to calibrate the seismic partial angle stacks with modeled angle gathers generated using well data and associated with the prospect using amplitude tools. Processing centers need to be made aware of interpreters needs. It may be noted that partial angle stacks generated during processing provide average responses observed at different depths, as the angles may not represent the intervals of interest.
The authors would like to thank U.S. Geological Survey and the U.S Department of the Interior for making partial stacks available, Schlumberger for the Petrel license to make the 3-D interpretation and geospatial integration, Enverus for the license to download the historical archive of borehole information, and CGG and Hampson and Russell for the license and training dataset. Also, the editors for the careful reviews and candid discussion.