Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Earth science
Reexamination Certificate
1999-10-27
2001-05-01
McElheny, Jr., Donald E. (Department: 2862)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Earth science
C367S073000
Reexamination Certificate
active
06226596
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to Computer Aided Interpretation Systems and more particularly to analysis of graphical data. Even more particularly, the invention relates to analysis of three dimensional seismic data.
BACKGROUND OF THE INVENTION
Since the early 1980's, three dimensional seismic image technology has revolutionized subsurface geologic mapping and hydrocarbon exploration in the petroleum industry. However, interpretation of three dimensional seismic images has been mainly a subjective process involving a human interpreter extracting and discriminating information by visual inspection of reflection patterns. This approach is time consuming, the results depend on individual human interpreters, and the results usually are not repetitive. Also, human interpreters tend to inspect reflection patterns along either the inline or crossline direction but ignore the third dimension. In previous studies, numerous seismic attributes have been used in an attempt to quantify seismic interpretation. However, these attributes are generally based on one trace and depend on an interpreted horizon, or they have been specifically used to enhance certain aspects of seismic data, such as dip, azimuth, and faults, etc. These attributes have certain limitations in the identification and classification of seismic facies, for example, that are characterized by lateral relationships of amplitude in different orientations and from different perspectives.
Little has been published on quantitative seismic pattern recognition and hydrocarbon indicator detection in three dimension s using volume-based and horizon-independent seismic technology. Although image textures have been used in conventional two dimensional image processing, such as photographs, Landsat, and side-scan sonar images, etc., little attempt has been made to quantitatively analyze three dimensional seismic interpretation and hydrocarbon exploration data. Since seismic images differ fundamentally from other images in that seismic images consist of vertical traces with alternating peak and trough amplitudes, and the traces are aligned laterally in an orderly manner to form characteristic stratal nature of reflection events, a different approach is required to capture and process both stratigraphic and structural information in the subsurface out of the seismic volume. Although several authors have applied the texture concept to reflection seismic interpretation in recent years, textures have usually been extracted in two dimensions, which significantly limits their reliability and resolution.
Manipulating multiple attributes effectively is an important next process after numerous attributes have been extracted. Conventionally, false-color (RGB) mapping, principal component analysis (PCA), clustering, and supervised classification have been commonly used to reduce the dimensionality of attributes and to display multiple attributes. Such processes use human interpreters, and since interpreters generally have little idea about the clustering structure of attributes in multiple dimensions, they must subjectively determine the number of classes to be grouped before classification. That may lead to a classification volume in which two different classes may be grouped into one class, or one actual class is split into two or more different classes. Generally, human interpreters have little control and may not realize that misclassification occurred until they learn the actual facies classes in the data volume by running different sessions of classification using different numbers of classes.
Combining multiple textural attributes helps minimize the non-uniqueness in seismic pattern recognition and classification. However, generalizing and interpreting numerous attributes are generally difficult, which is particulary true with a large number of attributes. Previously, various attempts have been made to condense and visualize multiple attribute data sets. For example, false-color imaging technology has been used to accommodate three attributes at a time by forming one RGB false-color image. Principle Components Analysis (PCA) has been performed to reduce the data dimensionality. In addition, various classification algorithms (e.g., clustering/unsupervised, NNT, and supervised classifiers) have been commonly used to reduce multiple attribute volumes into one final thematic volume that consists of discrete categories (classes) based on their closeness in multiple attribute space. However, with increasing number of seismic attributes and the size of attribute volumes, computational efficiency becomes an important issue in multi-attribute classification process. Besides, resolution and reliability of these conventional classification technology depend on user-specified class number. In this context, the attribute-trace concept proposed here provides a new solution to improve the efficiency, resolution, and reliability for multi-attribute classification and interpretation.
It is thus apparent that there is a need in the art for an improved method of analyzing three dimensional seismic data. The present invention meets these and other needs.
SUMMARY OF THE INVENTION
It is an aspect of the present invention to analyze three dimensional seismic data.
It is another aspect of the invention to transform seismic amplitude data into seismic texture attribute data.
Another aspect of the invention is to define a volume texture element (texel) around each voxel of the three dimensional seismic data.
Another aspect of the invention is to create a Voxel Coupling Matrix (VCM) from the three dimensional seismic data, and to transform the VCM into a plurality of texture attributes.
A further aspect of the invention is to transform the texture attributes into an interpretation (thematic) volume using an attribute-trace classification method.
The above and other aspects of the invention are accomplished in a system that first optimizes the seismic amplitude data and then defines an evaluation window, called a texel, to be used in selecting data. The texel has both a size and a geometry, and is used to select data to be converted from the three dimensional input seismic amplitude data into a VCM. Once texel size and geometry is selected, the input data is evaluated voxel by voxel to create the output VCM. The VCM data is then processed to create several texture attributes, such as homogeneity, contrast and randomness as well as others. The covariance of attributes are evaluated to select the least correlated and these are then combined using an attribute-trace method to produce an interpretation volume, which is then displayed.
REFERENCES:
patent: 5966672 (1999-10-01), Knupp
patent: 6011557 (2000-01-01), Keskes et al.
Ebel Jack E.
Marathon Oil Company
McElheny Jr. Donald E.
LandOfFree
Method for analyzing and classifying three dimensional... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for analyzing and classifying three dimensional..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for analyzing and classifying three dimensional... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2482294