Zeng, Hongliu, Tucker F. Hentz, and Lesli J. Wood

Bureau of Economic Geology, The University of Texas at Austin, Austin, TX

Stratal slicing of the Vermilion Block 50-Tiger Shoal 3-D volume revealed numerous high-quality, satellite-picturelike, depositional patterns in the middle Miocene-Pliocene section. Through pattern recognition guided by seismic-lithology relationships, modern depositional models, and sequence-stratigraphic analysis, depositional facies and depositional history can be studied with reservoir-level resolution (fourth and fifth order). Some imaged facies (depositional elements) that are important to reexploration and infield drilling in the area include (1) fluvial channels, (2) incised valleys, and (3) highstand deltas.

The Offshore Secondary Gas Recovery (OSGR) project is an ongoing joint research venture between the Bureau of Economic Geology and the U.S. Department of Energy. The goal of the OSGR is to research new techniques for defining the structure, stratigraphy, and hydrocarbon character in a major field in the northern Gulf of Mexico and to utilize these multidisciplinary techniques to identify additional gas resources, as well as to predict regional trends in hydrocarbon accumulation. As an industry partner, Texaco contributed the well data from its Starfak (Vermilion Block 50) and Tiger Shoal fields in the Vermilion and South Marsh Island Areas, offshore Louisiana, and, more important, a merged 3-D seismic volume in the area.

The 3-D seismic volume, 23 € 14 mi, records an interval of Miocene-Pliocene sediments in a 0.4- to 4.0-s section, including all major hydrocarbon-producing zones in the area. Except for local Starfak and Tiger Shoal fields, however, most of the area has no well control. To reach the project goal of finding additional gas reserves between and beyond wells, workers must improve 3-D seismic interpretation so that sequence-stratigraphic and reservoir study can be extrapolated to the entire area, throughout the section of interest. An important step is to optimize depositional-facies mapping by automating depositional-surface picking in the 3-D volume and to link seismic-attribute patterns directly to sedimentary rocks and depositional processes.

Seismic-Lithology Relationship

Analysis of sonic and density logs is essential to revealing the acoustic relationship between sandstone and shale. A segment (upper Miocene) of sonic and density logs in a type well was first converted to an impedance log (Z, product of velocity € bulk density) and then plotted against the effective porosity (f_e, density porosity with clay effect removed, Fig. 1). f_e is here an indicator of lithology, with the highest f_e (>0.3) being clean sandstone, the lowest f_e (<0.1) being shale, and intervening values being shaly sandstone and sandy shale. Notice in Figure 1 that, statistically, Z linearly decreases with f_e. In other words, sandstone is characterized by a Z lower than shale. If bounded by thick shale, the top of the sandstone would be reflected as a seismic trough (zero-phase wavelet in SEG reverse polarity).

Figure 1. Acoustic impedance as an indicator of lithology (effective porosity). Data are calculated from sonic, density, and GR logs from the 2,000- to 2,700-m section (upper Miocene) in a Starfak well.

After careful depth-to-time conversion and seismic-phase adjustment, this acoustic separation of sandstone and shale can best be seen on a well-to-seismic tied section (Fig. 2). More than 90% of the sandstones (middle Miocene-Pliocene) are tied to seismic trough events (red) without much ambiguity. Not only are regionally continuous sandstones confined to continuous seismic trough events (e.g., B, I, and N sands, Fig. 2), but many lenticular sand bodies also show excellent correlation to patchy seismic events (e.g., a, b, and c, Fig. 2). In contrast, most of the thin shales are characterized by seismic peak events (black). The only exception is some thick shaly units in the middle Miocene section that are reflected as complex peak-trough couplets, an indication of internal heterogeneity. The seismic amplitude (trough) can be safely used as an interpretation guideline to indicate sandstone directly. It can also be used to tie seismic to the sandstones in area wells.

Stratal Slicing

Seismic attributes must be picked on a depositional surface (geologic time surface) if they are to represent a genetic depositional unit. Such a seismic-surface display is called a stratal slice (Zeng, 1994). Time slices and horizon slices are currently the most commonly used seismic-surface displays to extract stratigraphic information. For depositional-facies analysis, however, both methods have limitations. Their strict application conditions have prevented them from being used in complicated (e.g., wedged) seismic sequences. Stratal slicing improves seismic-surface display mainly by making slices proportional between geologic time-equivalent seismic-reference events (e.g., flooding surfaces). A comparison among different approaches will be presented in the poster. For more details refer to Zeng (1994), Zeng et al. (1995), Posamentier et al. (1996), and Zeng et al. (1998a and b).

A stratal-slice volume has been generated among 13 reference seismic events in the middle Miocene-Pliocene section. Among 776 stratal slices in the roughly 3.0-s data interval, 46 are highlighted on the well-to-seismic section (Fig. 2) to show positions of representative stratal slices in two-way time and their relationship with seismic events, lithostratigraphic units, and well log characteristics. The highlighted slices demonstrate typical depositional facies images in different stages of basin development and will be shown in a stratal-slice movie in the poster. However, only selected slices are discussed herein for highlighting the methodology and the geologic interpretation of stratal slices.

Pattern Recognition and Facies Analysis

The interpretation of stratal slices is an integrated study of all available data. The key process is the pattern recognition of seismic-attribute (most commonly amplitude) images guided by depositional models. In some cases the patterns alone are enough for accurate facies identification (e.g., a meandering fluvial system). In other situations, however, well-derived information is always helpful, especially when seismic patterns are complex. In this study, the following pieces of information are used in seismic-pattern/depositional-facies conversion:

1. Pattern geometry: channel (shoestring), lobe, digitate, sheet/random, etc.

2. Pattern texture: smooth, patchy (variable), wormlike, etc.

3. GR/SP log pattern: upward fining, upward coarsening, blocky, serrate, straight, etc.

4. Amplitude (seismic-lithology relationship): negative (thick/blocky sand), positive (shale/condensed section), low/variable (variable polarity, thin sand/shale), highly negative (gas sand).

5. Relationships: regional setting, structure, sea level/systems tract inferred by sequence-stratigraphic study, etc.

For example, slice 1160 (Fig. 3a) illustrates some moderately sinuous, channellike features. Internal amplitudes are more smoothed as compared with those surrounding the features. On the basis of seismic-lithology modeling, negative amplitudes indicate thick sandstone trends. Wells that penetrate these features show that they are characterized by an upward-fining log pattern, supporting evidence of a fluvial channel fill. The slice is in the Pliocene section, an interval interpreted by sequence-stratigraphic study (Hentz et al., this meeting) as developing primarily in a coastal-plain environment. Therefore, the best interpretation of these features would be coastal streams.

Slice 2248 (Fig. 3b) shows a markedly different channel type. Most of these features are subregional in size and typically straight with local bifurcation downstream. Internally smoothed, negative-amplitude patterns are visually striking as compared with surrounding sheetlike, mostly positive amplitude patterns. These features are composed of mostly blocky and blocky-serrate sandstones in the penetrating wells and can be interpreted as thick sandstone belts. Stratigraphically below the Pliocene coastal-plain deposits, these Miocene features represent incised valley fills (IVFÌs) that incise exposed shelf facies. The IVFÌs contain deposits of lowstand and transgressive systems tracts (Hentz et. al, this meeting). Similar IVFÌs can be seen terminating updip in the coastal plain (e.g., Fig. 3), indicating that at least some of the IVFÌs also drain the landwardmost portion of the lowstand coastal plain.

Figure 3. Amplitude stratal slices showing (a) a Pliocene coastal plain, (b) upper Miocene incised valley fill, and (c) an upper Miocene highstand shelf delta system. See Figure 2 for stratigraphic position. CH-channel; FP=floodplain; IVF=incised valley fill; SH=shelf; BS I-IV=bright spots.

Slice 2472 (Fig. 3c) highlights generally lobate and digitate aerial geometries that have negative amplitudes that are inferred to be thick sandstone bodies. Each finger- or channellike feature grades downdip to low/variable-amplitude lobes, which have been confirmed by wells as being composed of thin, serrate sandstones. On the basis of its highstand position in the sequence-stratigraphic framework (Hentz et. al, this meeting), the system has been interpreted as a highstand shelf delta.

In the interval studied, a gas sandstone commonly shows as a bright spot. However, condensed sections and salt layers sometimes also generate bright spots that confuse interpreters. Facies analysis on stratal slices can help in reducing the ambiguity. For instance, F1 sand in Tiger Shoal field is gas producing in several wells. The reservoir is interpreted as an IVF sandstone and is reflected in Figure 3b as a bright spot labeled BS I. Although in some incised-valley deposits, thickness changes can cause amplitude variability, here there is no indication (on the basis of combined seismic and well data) that this sand is characterized by any significant thickness change throughout the field area. Therefore, the amplitude change is interpreted to be a function of fluid content. Several other examples of undrilled, high-amplitude IVFÌs are shown labeled BS II-IV. These examples, associated with the F1 sand interval, appear bounded updip by faults, creating possible structural/stratigraphic gas traps.

Through this process, stratal slices can be used to map sequential depositional elements and to aid in the evaluation of exploration and infield drilling targets. The next step is to automate the procedure by applying Neural Networks in facies-pattern recognition.

Seismic Sedimentology

Like an air-photo or satellite picture of a modern depositional system, stratal slices can be used in sedimentological study of ancient depositional systems. Stratal slices make full use of horizontal seismic resolution (limited to seismic bin size or a quarter seismic wavelength, whichever is larger), enough to depict most, if not all, depositional-facies changes meaningful in well-to-well scale reservoir detection/description. Depositional facies cannot only be mapped areally in detail (as shown in Fig. 3), but they can also be studied in geologic time to gain a sense of depositional history.

In the studied interval, expected stratigraphic resolution (at which two consecutive stratal slices show significant pattern changes) is around 5 m. In fact, as will be shown in the stratal slice movie, all fourth-order (30 to 130 m) and many fifth-order (5 to 30 m) depositional systems that can be recognized in well log correlation have been resolved by stratal slices. For example, Figure 4 depicts four IVFÌs in a 100- to 200-m section all developed during fourth-order lowstand periods. Slices 2544 and 2560 are stratigraphically separated by a shale only 3 to 15 m thick. The four slices were selected from 28 stratal slices sliced proportionally through the interval, many of them depicting significant facies changes. The four-slice series highlights the migration of IVFÌs with geologic time by portraying their size, direction, and spatial distribution. Such generic information on the character and behavior of Miocene-Pliocene depositional systems can also be used to gain more regional understanding of targeting Mio-Pliocene exploration in the Gulf of Mexico.

Figure 4. A four-stratal-slice series showing the migration of IVFÌs in geologic time (M-K sands). See Figure 2 for stratigraphic position. Vertical axis not on scale.

References

Posamentier H.W., G.A. Dorn, M.J. Cole, C.W. Beierle, and S.P. Ross, 1996, Imaging elements of depositional systems with 3-D seismic data: a case study: GCSSEPM Foundation 17th Annual Research Conference, p. 213-228.

Zeng, Hongliu, 1994, Facies-guided 3-dimensional seismic modeling and reservoir characterization: Ph.D. dissertation, The University of Texas at Austin, 164 p.

Zeng H., M.M. Backus, K.T. Barrow, and N. Tyler, 1995, Three-dimensional seismic modeling and seismic facies imaging: GCAGS Transactions, v. 45, p. 621-628.

Zeng H., M.M. Backus, K.T. Barrow, and N. Tyler, 1998a, Stratal slicing, part I: realistic 3-D seismic model: Geophysics, v. 63, no. 2, p. 502-513.

Zeng H., S.C. Henry, and J.P. Riola, 1998b, Stratal slicing, part II: real seismic data: Geophysics, v. 63, no. 2, p. 514-522.