Seismic stratigraphic surface classification

• US 10,139,507 B2
• Filed: 01/29/2016
• Issued: 11/27/2018
• Est. Priority Date: 04/24/2015
• Status: Active Grant

• Current Assignee: Exxonmobil Upstream Research Company (Exxon Mobil Corporation)
• Original Assignee: Exxonmobil Upstream Research Company (Exxon Mobil Corporation)
• Inventors: Wahrmund, Leslie A., Paiva, Antonio R. C., Hanson-Hedgecock, Sara E.
• Primary Examiner(s): Lee, Kyoung
• Application Number:
  US15/011,135
• Publication Number:
  US 20160313463A1
• Time in Patent Office: 1,033 Days
• CPC Class Codes: G01V 1/28 Processing seismic data, e.g. for interpretation or for event detection G01V1/48 takes precedence
• CPC Class Codes: G01V 1/30 Analysis G01V1/50 takes precedence
• CPC Class Codes: G01V 1/301 for determining seismic cross-sections or geostructures
• CPC Class Codes: G01V 2210/64 Geostructures, e.g. in 3D data cubes
• CPC Class Codes: G01V 99/00 Subject matter not provided for in other groups of this subclass
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

A method to classify one or more seismic surfaces or surface patches based on measurements from seismic data, including: obtaining, by a computer, a training set including a plurality of previously obtained and labeled seismic surfaces or surface patches and one or more training seismic attributes measured or calculated at, above, and/or below the seismic surfaces; obtaining, by the computer, one or more unclassified seismic surfaces or surface patches and one or more seismic attributes measured or calculated at, above, and/or below the unclassified seismic surfaces; learning, by the computer, a classification model from the previously obtained and labeled seismic surfaces or surface patches and the one or more training seismic attributes; and classifying, by the computer, the unclassified seismic surfaces or surface patches based on a comparison between the classification model and the unclassified seismic surfaces or surface patches.

Claims

1. A method to classify one or more seismic surfaces or surface patches based on measurements from seismic data, comprising:

obtaining, by a computer, a training set including a plurality of previously obtained and labeled seismic surfaces or surface patches and one or more training seismic attributes measured or calculated at, above, and/or below the seismic surfaces;

obtaining, by the computer, one or more unclassified seismic surfaces or surface patches and one or more seismic attributes measured or calculated at, above, and/or below the unclassified seismic surfaces, wherein the one or more seismic attributes includes a single boundary measure of attribute contrast above and below a seismic surface or surface patch of the plurality of previously obtained and labeled seismic surfaces or surface patches, such that the single boundary measure represents the collapsing of stratigraphically diagnostic seismic facies information into said single boundary measure;

learning, by the computer, a classification model from the previously obtained and labeled seismic surfaces or surface patches and the one or more training seismic attributes, wherein the classification model is based at least in part on the single boundary measure of attribute contrast; and

classifying, by the computer, the unclassified seismic surfaces or surface patches based on the application of the classification model to the unclassified seismic surfaces or surface patches.

2. The method of claim 1, wherein the method further includes quantifying a degree of confidence in a classification of the unclassified seismic surfaces or surface patches.

3. The method of claim 1, wherein the classifying includes labeling the unclassified seismic surfaces or surface patches with a label that differentiates between stratigraphic classes.

4. The method of claim 3, wherein the labeling differentiates between sequence boundaries and flooding surfaces.

5. The method of claim 3, wherein the classifying uses a relationship between surfaces to further differentiate flooding surfaces into maximum flooding surfaces or transgressive flooding surfaces.

6. The method of claim 1, wherein the learning includes learning the classification model implicitly from the plurality of previously obtained and labeled seismic surfaces or surface patches and the one or more training seismic attributes.

7. The method of claim 1, wherein the single boundary measure is derived from one or more of the following attributes above versus below the seismic surface or surface patch:

amplitude;

dip;

frequency;

chaos;

attributes that highlight thinning, pinchouts, and/or terminations; and

attributes that highlight surface position, extent, continuity, and curvature.

8. The method of claim 1, wherein the classifying includes eliminating redundant attributes from amongst the one or more seismic attributes measured or calculated at, above, and/or below the unclassified seismic surfaces using a single-link hierarchical dendogram.

9. The method of claim 1, wherein the classifying is by hard assignment.

10. The method of claim 1, wherein the classifying is by soft assignment.

11. The method of claim 1, wherein the classifying includes segmenting the one or more unclassified seismic surfaces or surface patches.

12. The method of claim 1, wherein the classifying includes individually classifying segments of the one or more unclassified seismic surfaces or surface patches.

13. The method of claim 11, wherein the segmenting results in approximately equal segment sizes.

14. The method of claim 11, wherein the segmenting includes using the one or more seismic attributes to determine segment size, wherein at least one segment has a different size than another segment.

15. The method of claim 2, further comprising differentiating between different flooding surfaces of the classification of the unclassified seismic surfaces or surface patches.

16. The method of claim 1, further comprising using the classification of the unclassified seismic surfaces or surface patches to manage the production of hydrocarbons.

17. The method of claim 1, wherein the classification model is learned in an incompletely labeled training dataset.

18. The method of claim 1, wherein the classifying includes classifying surfaces above but not below AVO.