--> --> Large Volume Reconnaissance Using 3-D Visualization Software, Desert Springs Field, Sweetwater County, Wyoming, by Robert Bunge, #40103 (2003).
[First Hit]

Datapages, Inc.Print this page

  Click to view article in PDF format.

 

Large Volume Reconnaissance Using Previous Hit3-DNext Hit Visualization Software, Desert Springs Field, Sweetwater County, Wyoming*

By

Robert Bunge1

 Search and Discovery Article #40103 (2003)

 

*Adapted from “extended abstract” for presentation at the AAPG Annual Meeting, Salt Lake City, Utah, May 11-14, 2003. 

1Anadarko Petroleum Corporation, The Woodlands, Texas ([email protected])

 

Introduction 

Three-dimensional visualization software provides an effective tool for reconnaissance mapping of large volume Previous Hit3-DNext Hit seismic datasets. It is particularly useful for evaluating areas of poor data quality and areas with high structural or stratigraphic complexities. A workflow process is presented to demonstrate how prospective seismic anomalies can be identified more rapidly and with greater accuracy than by conventional software interpretation schemes. 

These techniques are demonstrated using Desert Springs Field in south-central Wyoming (Figure 1), which has produced over 226 BCFG equivalent from the Upper Cretaceous Lewis Formation since 1958. The Lewis reservoir is a series of turbidite sandstones deposited at the base of a southward prograding slope wedge. This occurred during the final transgressive phase of the Cretaceous Western Interior Seaway. The trap is created by the updip pinch out of sandstones along the east flank of the Rock Springs Uplift. 

Seismic data in the Rockies is subject to a variety of data problems associated with geology, acquisition, and processing that make mapping amplitude anomalies related to stratigraphic targets difficult to map. Geology-related difficulties include items such as dipping strata, structural complexity, and small, scattered targets. Acquisition and processing related issues include items such as noise and data skips.

 

uIntroduction

uFigure captions

uWorkflow

uSummary

uAcknowledgments

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

uIntroduction

uFigure captions

uWorkflow

uSummary

uAcknowledgments

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

uIntroduction

uFigure captions

uWorkflow

uSummary

uAcknowledgments

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

uIntroduction

uFigure captions

uWorkflow

uSummary

uAcknowledgments

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

uIntroduction

uFigure captions

uWorkflow

uSummary

uAcknowledgments

 

 

Figure Captions

Figure 1. Index map of Laramide basins of Wyoming (after Surdam et al., 1997) and location map of Desert Springs Field, with generalized structural map (after Ryder et al., 1981).

Figure 2: Reconnaissance scan of the volume. Snapshot A shows the general structure, including a large fault, acquisition footprint, and a prospective anomaly (Desert Springs Field). Snapshot B shows a close-up. This is interpreted to be a mounded feature at the toe of a clinoform. The green arrow marks an anomaly of interest.

Figure 3: Autopick of continuous horizon. Frame A shows the first iteration of the autopick. Frame B shows the second iteration, with a Previous Hit3-DNext Hit Slice showing the relation of the horizon to the anomaly (marked by the arrow). The horizon is several hundred ms below the anomaly, but the reflections are relatively conformable. Frame C shows the final horizon after three iterations of the autopicker and an interpolation. Subtle, low offset fault patterns can be interpreted.

Figure 4: Frame A shows the flattened volume with a Previous Hit3-DNext Hit slice. Frame B shows the flattened Previous Hit3-DNext Hit slice trimmed to the anomaly. Frame C shows a perspective view of the Previous Hit3-DNext Hit trim volume. The trim volume is 56 ms thick.

Figure 5: Map view of the trim volume in Figure 3C with the opacity curve to the left applied. The prospective anomaly can be seen the lower right-center. Often low amplitude patterns are more clear where opacity has removed the data. The white dashed line follows a low amplitude trend related to the toe of a prograding complex. Some low amplitude meanders may be seen as well (one is highlighted in green).

Figure 6: Using multibody detection, all of the voxels with an amplitude greater than 12,000 are highlighted, and a histogram of the sizes of detected bodies is created. Bodies can then be eliminated based on size. In the middle frames only the 10 largest anomalies are still highlighted. On the right, only the largest anomaly in the trim volume is left.

Figure 7: Horizon fit to multibody detection. On the left, the volume has been unflattened, and a horizon has been fit to maximum amplitude in the detection. On the right, the horizon is seen in the anomaly on a Previous Hit3-DNext Hit slice.

Figure 8: Structure map of anomaly with Desert Springs wells shown. The insert shows the pay sand on an SP curve overlain on the seismic. Although the field is larger than the mapped anomaly, current mapping is sufficient for exploration purposes. The downdip edge of the anomaly shows the gas-water contact (dashed, blue line), and the updip edge is coincident with the 40-ft gross sand isopach (red line). The divergence from the 40-ft contour on the southern end is likely due to a facies change in the reservoir. Field development would require more detailed mapping and advanced seismic techniques.

 

Workflow 

There are several ways of addressing data problems during interpretation. I have found the workflow below to be fast and effective. Visualization software makes interpretation much easier by allowing data quality and data problems to be quickly assessed. Prospective anomalies are isolated through data elimination. Removing data not related to prospective anomalies takes away distraction and noise. Data can be removed both by paring down a volume to the prospective zone and by using opacity. 

1) Scan the Previous Hit3-DNext Hit volume to identify any prominent isolated anomalies (Figure 2). Make notes not only about anomalies, but also about strong continuous events close to them. Also note data quality issues such as skips and spatially varying amplitudes related to surface or processing problems. 

2) Map continuous events nearest the anomalies of interest are using an autopicker (Figure 3). Three-dimensional visualization allows rapid quality check of the seismic picks and structural interpretation by scrolling through the volume. The threshold should be set very high (90-95%) to prevent bleeding. After the autopicker has mapped the coherent event, interpolate the horizon between the autopicked points. 

3) Flatten on the horizon of interest (Figure 4). Although the anomalies themselves may be difficult to map, nearby continuous events provide a structural picture. They can also be used to back-out structure to improve the ease of making stratigraphic interpretation. 

4) Reduce the volume to just the anomalous areas (Figure 5). Trim the volume as close as possible to anomalies in time and cut down the area to the traces of interest. Be aware of ‘anomalies’ that may be related Previous HitresidualTop structure after flattening the volume. 

5) Adjust the opacity to highlight significant anomalies and to eliminate background data (Figure 5). The opacity curve can be manipulated in any number of ways to show the anomaly better. Also play with different color-bars to see if they add to the visualization. 

6) Multi-body detection is employed to isolate the larger, more prospective anomalies and to remove smaller, insignificant anomalies (Figure 6). If an economic limit related to prospect size has been determined, it can be used as a threshhold during anomaly elimination. Voxels highlighted during body detection will remain highlighted if the volume is unflattened (Figure 7). This advantageous for viewing the anomalies and fitting horizons to the anomalies in their proper structural position.

 

Summary 

Through this workflow an interpreter can quickly identify and isolate targets that would otherwise be extremely time consuming and difficult to interpret. Three-dimensional visualization is the best method to map prospects faster, easier, and more precisely (Figure 8).

 

Acknowledgments 

Thanks to Anadarko Petroleum Corporation for supporting this project and allowing the work to be shared. Thanks to Lee Shannon for assisting with information on Desert Springs Field. WesternGeco graciously allowed the Continental Divide data set to be used in this paper. Figures 2-7 are screen captures from VoxelGeo by Paradigm Geophysical. Figure 8 is a screen capture from SeisWorks by Landmark Geophysical.

Return to top.