Finding channel sands with seismic facies analysis and litho-seismic modeling

March 1, 1999
W - E in-line #634 from Winters pinchout 3D survey with interpreted horizons. Reference horizon T1 (red) is used for turbidite deposits analysis, reference horizon C1 (blue) is used for the channel play. Well pay zones: A - 26 ft, B - 15 ft, C - 70 ft, D - dry. Images courtesy of Flagship Geosciences [54,578 bytes].

A new exploration methodology

Manuel Poupon
Flagship Geosciences

Kostia Azbel
CGG Geoscience

George Palmer
GeoNexus Interpretive Modeling Corp

The Winters pinchout 3D survey is located in the southern part of the Sacramento basin in Solano County, California. It extends over 52 sq miles and covers the Elkhorn Slough field. This survey is from 3D spec data acquired by CGG in 1990-1995.

The Winters pinchout 3D survey is located on the edge between the Winters structural play in the west and the Upper Winters sands pinchout in the east. This latter Upper Cretaceous play has been described as "turbiditic in nature, sands being transported through channels incised into the shelf and deposited into deepwater fans surrounded by shales" (K. Lanning & G. Cambois, 1998).

Oil exploration in this area was performed in several phases. The first drilling campaign started in 1992 and was based on 2D seismic interpretation and regional study of net sand isopachs. Amerada Hess drilled three wells in partnership with Enron Oil and Gas and encountered Winters sands at approximately 8,600 ft. These wells showed mixed results.

The second drilling campaign started in 1997 after a detailed interpretation of the Winters pinchout 3D survey including AVO, Coherency Cube, and volume factor analysis. This campaign boosted estimated reserves to 14-18 bcf recoverable.


The discovery and development of the Elkhorn Slough field is a good example of continuing development of E&P technology driving drilling, from 2D seismic and well mapping techniques up to 3D interpretation and sophisticated seismic attributes and AVO analyses. For that reason, this field was selected as our pilot project to test the validity of this new approach.

Using Stratimagic™, a 3D stratigraphic interpretation system, combined with NexModel™, a litho-seismic modeling package, a new interpretation was performed based on seismic facies analysis (using neural network applied to seismic trace shape) and 1D-modeling of seismic responses from well curves.

This methodology uses an interval approach. Seismic traces come from a user-defined seismic interval. Boundaries of the interval can be parallel to one seismic event (constant time interval) or limited by two events (seismic interval with variable thickness). The definition of the interval is crucial for the seismic facies analysis and therefore great care should be put into the interpretation and selection of the reference horizon(s).

Reference horizons

The first phase of the work was to interpret the appropriate seismic events to be used as reference horizons. A model-oriented auto-tracking tool was used to interpret two horizons: a trough at the top of the turbidite package (horizon T1, red) and a peak within the turbidites (horizon C1, blue), previously identified as the top of the Winters sands. Horizon attributes (amplitude, dip and azimuth maps) were calculated and used to control the quality of the auto-tracked picks and identify the area of interest.

Using a horizon-slicing tool we were able to identify different intervals of interest. First, a constant time interval corresponding to a 60-ms interval below the T1 horizon was used to characterize the turbidite package over the entire 3D survey. A neural network analyzed trace shapes within this interval and a series of synthetic traces (representing the shape variation within the interval) was generated and sorted in a model.

Each trace shape from our seismic interval was compared with the 18 synthetic traces and assigned a model trace color, based on maximum correlation. The resulting seismic facies map shows similarity between actual traces and a set of model traces generated by the neural network. This first analysis, performed in 5 minutes, was done in an unsupervised mode that does not require any seismic pre-processing or any well data. As a result, we were able to identify the limit of the turbiditic deposits as well as a channel system oriented NE-SW and incising turbidite sequences oriented NW-SE. This channel system corresponds to the Elkhorn Slough Field.

Seismic facies map

To characterize the Winters sands, a second area was defined corresponding to the 60-ms seismic interval below the C1 horizon and the analysis was limited to the channel area only. A "supervised" approach was used by replacing one model trace by the actual seismic trace at Well C. This seismic response, corresponding to 70 ft of gas sands, was used as an indicator of Winters reservoirs.

The resulting seismic facies map indicates differences in seismic trace shapes between wells. Conventional amplitude-based maps and coherency slice tend to group all these wells in a bright spot zone. Amplitude maps could be good indicators of gas presence, but could not differentiate between the reservoir zone (Well C) and non-economic gas accumulation or "fizzy" water (Well D). Trace shape variation proved to be a good indicator of gas-charged sand thickness.

After seismic facies maps were interpreted, a conventional interval attribute analysis was performed over the same intervals. This allowed us to improve our understanding of the channel system by combining the results of our seismic facies analysis with interval attribute maps using the mixed map tool. On these mixed maps, attribute maps are used as background for the seismic facies to highlight zones with amplitude anomalies.

Litho-seismic modeling

The first part of our work was to confirm that geological features identified on our seismic facies maps were related to the Winters sands. If Winters reservoirs have a seismic signature (both in shape and amplitude) that can be modeled from well curves, then our seismic facies maps could be calibrated to sand thickness and/or petrophysical properties such as porosity or fluid content.

To analyze only one variable at a time, the litho-seismic model was based only on shape. It was determined from Elkhorn field that 6 ft of gas sand was giving an amplitude response similar to a 70-ft reservoir. Using well curves (DT, RHOB and GR) Winters sands were identified. A synthetic seismogram was generated and compared with the seismic trace at Well C. Analysis showed that gas-charged Winters sands could be modeled. Next we examined well parameters to better understand the seismic shape signature in terms of geological properties.

Decreasing the sand thickness from 70 ft to 30 ft changed the trace shape by shifting the trough from the middle of the interval to the top and decreasing the wavelet length. Variation of the porosity from 20% to 10% did not affect the trace shape significantly.

However, decreasing gas saturation by 50% affected the trace shape causing a decrease in the relative amplitude of the trough in the middle of the interval. This behavior exactly replicates the variation of shapes observed between Well C and Well D. This analysis confirmed that the seismic facies map was a good indicator of sand thickness as well as gas saturation.


The mixed map is a perfect example of integration between the seismic facies approach and conventional interval attribute analysis. This map highlights thick gas sands that were poorly defined by interpretation techniques based only on amplitude. When the seismic facies map is used in conjunction with litho-seismic modeling, the interpreter can speculate on lithologic characteristics of untested areas.

The idea of analyzing a seismic facies map based on the variation of seismic trace shapes comes from the assumption that changes in lithology, rock properties and fluid content should affect seismic response in both amplitude and shape. Amplitude analysis has always been a key element in oil exploration. Although this technique combined with seismic inversion and coherency cube technology has proven successful, it is often time-consuming, requires the expertise of several geoscientists and can mislead the interpreter.

The direct study of the variation of trace shapes has been neglected in the oil exploration due to the lack of appropriate tools to accurately map these changes. The lack of a tool to quantify these changes in trace shape controlled by petrophysics was also a major limitation to this technique.

Based on the results obtained in the Elkhorn Slough field, seismic facies analysis combined with litho-seismic modeling of well data is an accurate, cost-effective and quick methodology. It revealed subtle geological features expressed in the shape of the seismic trace. This methodology combined with conventional seismic attribute analysis allowed a better understanding of the Winters sands that could be used in future field development/appraisal or in conjunction with other techniques such as seismic inversion.


The authors wish to thank CGG-Americas (Paul Ettinger, CGG 3D spec and Jim Hallin, CGG-Geoscience) for providing us with the data and the Flagship Geoscience staff (Randy Phillips) for reviewing this paper. Stratimagic and NexModel are trademarks of CGG-Petrosystems. Stratimagic incorporates the SISMAGE technology under license from Elf Aquitaine.


Lanning, K., Cambois, G., "Case Study of the Elkhorn Slough Field, Solano Co. California. Risk Reduction using State-of-the-art 3D tools,"1998 SEG, Expanded Abstracts.

Nickerson, R., Cambois, G., Meyer, J., "A Successful AVO Case Study from Marginal 3D Land Data," SEG 1997 Expanded Abstracts, 167-170.


Manuel Poupon is Director of Technology with Flagship Geosciences in Houston, Texas.

Kostia Azbel is a Consulting Geophysicist at CGG Geoscience in Houston, Texas working with the Flagship Geosciences Team.

George Palmer is General Manager of GeoNexus Interpretive Modeling Corp and provider of dynamic interpretive modeling software.

Copyright 1999 Oil & Gas Journal. All Rights Reserved.

Courtesy Finder Energy's QUARTERLY REPORT, 31 January 2024
Photo 218944302 © Cagkan Sayin |
Photo 160704710 © Riccardo Mojana |