Stratal Slices and Stratagrids: an Application for Seismic Geomorphology

By: Staffan Kristian Van Dyke1

1 Geomodeling Technology Corporation; Houston, TX, USA; email: staffan.vandyke@geomodeling.com

The process by which a seismic volume is flattened with reference to an interpolated, continuous 3-dimensional surface (lithologic boundary, disconformity, condensed section, etc.), defined as a horizon, is known as “stratal slicing.” This type of extraction differentiates from a typical time slice which is simply the manifestation of values along a purely horizontal cross-section taken across the seismic survey (Figure 01a). Although much information can be ascertained from such an extraction, the process of stratal slicing is a much preferred method for revealing the stratigraphic nature of an interval being studied, because it removes the variation of seismic attributes (e.g., amplitude) across different strata (Figure 01b). This unique approach was introduced into industry in the late 1990’s by Dr. Hongliu Zeng, then a Research Scientist for the Bureau of Economic Geology at the University of Texas, in Austin.

Figure 01: A Typical seismic time slice extraction of amplitude values showing the morphologic pattern of a channel, whereby variations of seismic amplitude reflects different strata or changes of acoustic properties within the same stratum (a); and the resultant stratal slice extraction (b) showing much more internal detail with regard to the channel’s continuity (b).

The example used in this study comes from a public dataset known as the Stratton Field dataset and it is located in south Texas, in the northwestern portion of the Gulf Coast Basin. The main sedimentary package being studied for this paper is the Oligocene Frio Formation; it is characterized by a high sediment-supplied fluvial system distinguished by rapid deposition and high subsidence rates (Levey et al, 1994). Most prominent in the 3D seismic survey is a large

syndepositional growth fault capped by a progradational sequence exhibiting the characteristics of a meandering stream environment. This can all be ascertained directly from stratal slices extracted at different intervals above and below the referenced horizon(s).

The software used in this study is AttributeStudioTM; all accompanying diagrams and figures derive from this source. AttributeStudioTM is a seismic attribute interpretation platform that integrates both qualitative and quantitative workflows, such as attribute generation, visualization, classification, and property prediction. We introduce “strata-grid” as the major object for analyzing attributes in the interactive attribute workflow. The attribute analysis process using AttributeStudioTM is made to be very interactive, largely because the strata-grid is used as one of the major attribute analysis objects. The strata-grid makes the “interactive” possible, not only for visualizing the data, but also for quantitatively analyzing attributes for large 3-D seismic surveys – without handling the whole data volume (Wen, 2011). Stratal slicing within a strata-grid is much simpler than having to go through the arduous process of volume flattening, rather, phantom horizons are calculated at user-defined intervals organized to follow paleo-depositional surfaces; this process is much better than being limited by time slice sample unit range/subvolume interval range/etc., as an analog signal can theoretically be sampled infinite times, as can be done with strata-grids (Figure 02).

Figure 02: The concept of stratal slices and their associated attribute extractions along adjacent traces (Wen, 2011).

The idea of stratal slicing is simple, as one of Steno's Laws states, the Principle of Original Horizontality, depositional sequences are laid down in a horizontal manner. Since an interpreted horizon is a representation of a coeval 3-dimensional surface, it is in its nature to undulate in space as it follows the wavelet pick from trace to trace [trough trough, (+/-) zero-crossing (+/-) zero-crossing, or peak peak]. It is a horizon that is of such importance

when doing stratal slices since a time slice simply cuts stratigraphy horizontally. Because of this, no events of interest can be ascertained for reservoir characterization purposes unless they are purely parallel to the horizontal. And as most of us know, this is rarely the case; so a horizon, which captures all of the detail from one coeval surface, can then have all of its data referenced in accordance to that horizon via phantom horizons. The resultant strata-grid sub-volume is then a better representation of the depositional history for the interval being studied, at least many tens to hundreds of sample units beneath the flattened horizon (i.e., either depth [m] or time [s]). Alternately, AttributeStudioTM allows the user even greater flexibility by allowing two or more horizons to be used as input (one stratigraphically higher unit and the stratigraphically lower units as controls); these are then “sliced” as proportional slices, top-conformable slices, or bottom-conformable slices (Figure 03); each being uniquely suited to handle different depositional contacts and bedding geometries.

Figure 03: AttributeStudioTM Strata-grids: Proportional, Top Conformable, or Bottom Conformable.

Interactive analysis workflows based on seismic attributes calculated from or extracted to a “Strata-grid” are demonstrated as more effective and powerful than traditional linear workflows (Wen, 2011). For the purpose of this study, multiple horizons were interpreted to capture the seismic characteristics of a progradational sequence showing the EOD (Environment of Deposition) of a meandering stream environment. The picked horizons named were prefaced “CS” for Channel System, and suffixed with an “ip” for interpolated (Figure 04 and 05). Now that these events have been picked and they are represented by irregular topological surfaces, again, known as horizons, the real work can begin.

Figure 04: Seismic vertical section view showing prominent growth fault at base of interval capped by a progradational unit topped by the horizon CS_05 ip in pink with the horizon CS_01 ip at its base in purple.

Figure 05: CS_01 ip – CS_05 ip shown in 3D space.

Firstly, measure the amount of time/depth that you would like to flatten your volume by (this is not necessary when bounding by two or more horizons); generally a good rule of thumb is to backstrip the depositional history of the seismic reflectors either above or below your reference horizon allowing you to pin down other time-equivalent stratigraphic events, which may be used as controls for the Strata-grid(s). The resultant dataset will be far more accurate in respect to showing the internal architectural elements of the interval being studied.

The next process is to make the Strata-grid; choose the appropriate type of lithologic stacking pattern, proportional slices were used in this study since the stratigraphy of the reflectors exhibit a subparallel nature between the control horizons. 800 slices were extracted at equally spaced intervals for a roughly 400ms time sample range (Figure 06). The next step is to extract all available seismic attributes (instantaneous/geometrical/response/energy) to the Strata-grid, such extractions as instantaneous phase, response frequency, most negative curvature, reflection strength, and many others were studied in great detail in order to illuminate the seismic reservoir characteristics and morphologic patterns found hidden within each seismic attribute. In accordance with all typical attributes, spectral decomposition was also run. Multiple realizations of spectral decomposition were executed to evaluate the differing responses from the varied algorithms (discrete Fourier transform [DFT]; continuous wavelet transform [CWT]; time–frequency continuous wavelet transform [TFCWT]; S-transform [ST]; and continuous wavelet packet transform [CWPT]). The resultant seismic attribute extractions, including the spectral decomposition results, produced spectacular results (Figures 07, 08, and 09).

Figure 06: Proportionally sliced Strata-grid with 800 equally spaced intervals; poststack amplitude response of the top of the Strata-grid is shown

Figure 07: A mud-filled channel can be seen from iso-frequency Strata-grid slices calculated from the five different spectral decomposition algorithms, each exhibiting varying characteristics of the deposit (Wen, 2011).

Figure 08: Channel morphology clearly imaged with the sweetness attribute.

Figure 09: Internal channel architecture of a beautiful meandering fluvial system showing channel thalweg bounded by proximal levees (seismic attribute: semblance).