Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Earth science
Reexamination Certificate
2002-08-29
2003-12-09
McElheny, Jr., Donald E. (Department: 2857)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Earth science
Reexamination Certificate
active
06662112
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates generally to the field of geophysical prospecting. More particularly, the invention relates to the field of seismic data processing. Specifically, the invention is a method for classifying AVO anomalies in near-offset and far-offset seismic data volumes.
2. Description of the Related Art
AVO (Amplitude Versus Offset) analysis of multiple three-dimensional seismic data volumes requires that the volumes are properly aligned and that AVO anomalies are identified and distinguished from background trends. Recently, neural networks have been employed to assist in the identification and classification of AVO anomalies. Sun et al., “AVO Inversion by Artificial Neural Networks”, SEG Calgary 70th Annual Meeting, 2000, discuss a method for training a neural network to predict the far offset response, then using the network to predict the far-offset response everywhere and compare this to the actual far-offset response. They also mention the need to align neural network-predicted and actual far-offsets with cross correlation before making the comparison. There are no details given in their paper regarding the use of sub-sample interpolation, or exploiting or generating time shift and correlation volumes in the process to filter the time shift volume. Furthermore, they do not discuss the training of the neural network on the near and far volumes, derivative AVO volumes, or the use of the cross-correlation volume for AVO classification methodology.
The conventional approach to AVO classification uses the familiar AVO cross plotting methodology to discriminate AVO anomalies from background. The most powerful feature of cross plotting is the ability to interact with the data volumes with dynamic linking to better understand the sensitivity of the spatial and geologic extent of AVO anomalies relative to the region selected in the cross plot. Several related techniques have been used in the oil industry to automate or enhance the calculation of AVO anomalies.
DeGroot, “A Method for Transforming One or More Seismic Input Cubes to One or More Seismic Output Cubes by Way of Neural Network Mapping”, EAGE Conference, 1999, states that information can be obtained from multiple volumes including near-offset and far-offset stacks, gradients and intercept, acoustic impedance or 4-D volumes (3-D volumes over time). However, DeGroot does not disclose the methods necessary to perform these operations, nor the methods for dealing with residual alignment between offset volumes.
Balz and Pivot, “Fast Identification of AVO Anomalies Using Classification of Pre-Stack Waveforms”, discuss a methodology using self-organizing maps or K means clustering to classify AVO response. (SEG 2000 Expanded Abstracts, Society of Exploration Geophysicists International Exposition and Seventieth Annual Meeting, Calgary, Alberta, Aug. 6-11, 2000.) Their method is designed to work with pre-stack seismic data and for a specific interval defined through horizon interpretation. Their method is interval-based, not volume-based. There is no mention of building a 3D consistent time shift volume to time align AVO cubes, or exploiting multiple attributes for AVO classification, such as cross correlation and near and far product with difference. Additionally, they do not mention the use of a probabilistic neural network approach with user defined training.
Thus, there exists a need to generate, in a computationally efficient manner, a rapid method for classifying AVO anomalies in pairs of near-offset and far-offset seismic data volumes. This process must also mimic the process employed by and results obtained manually by the seismic interpreter.
SUMMARY OF THE INVENTION
The invention is a method for classifying AVO anomalies in pairs of near-offset and far-offset seismic data volumes. First, a plurality of initial AVO seismic attributes are calculated that are representative of the offset seismic data volumes. A probabilistic neural network is constructed from the calculated initial AVO seismic attributes. AVO anomaly classifications are calculated in a portion of the offset seismic data volumes. The preceding steps are repeated until the calculated AVO anomaly classifications in the portion of the offset seismic data volumes are satisfactory. Finally, AVO anomaly classifications are calculated throughout the offset seismic data volumes using the constructed probabilistic neural network.
REFERENCES:
patent: 5465323 (1995-11-01), Mallet
patent: 5859925 (1999-01-01), Yaeger et al.
patent: 5940777 (1999-08-01), Keskes
patent: 6044328 (2000-03-01), Murphy et al.
patent: 6236942 (2001-05-01), Bush
patent: 6438493 (2002-08-01), West et al.
patent: 6560540 (2003-05-01), West et al.
patent: WO 97/11392 (1997-03-01), None
patent: WO 00/16125 (2000-03-01), None
Sun et al., 2000, “AVO Inversion by Artificial Neural Networks”, SEG Calgary 70th Annual Meeting, Aug. 6-11, 2000, pp. 130-133.
Baltz and Pivot, 2000, “Fast Identification of AVO Anomalies Using Classification of Pre-Stack Waveforms”, SEG 2000 Expanded Abstracts, SEG Int'l Expo. and 70th Annual Meeting, Calgary, Alberta, Canada, Aug. 6-11, 2000.
DeGroot, 1999, “A Method for Transforming One or More Seismic Input Cubes to One or More Seismic Output Cubes by Way of Neural Network Mapping”, EAGE Conference, 1999.
Johnston, D. H, Eastwood, J. E., and Shyeh, J., 2000, “Using Legacy Sesimic Data in an Integrated Time-Lapse Study: Lena Field, Gulf of Mexico”, The Leading Edge, Mar. 2002, pp. 294-302.
Matlock, R. J. and Asimakopoulos, G. T., 1986, “Can Seismic Stratigraphy Problems be Solved Using Automated Pattern Analysis and Recognition?” The Leading Edge, Geophys. Explor., vol. 5, No. 9, pp. 51-55.
Vinther, R., Mosegaard, K., et al., 1995, “Seismic Texture Classification: A Computer-Aided Approach to Stratigraphic Analysis”, SEG International Expo and 65thAnnual Meeting, paper SL 1.4, Oct. 8-13, pp. 153-155.
Vinther, R., Mosegaard, et al., 1996, “3D Seismic Texture Classification”, SPE 35482, pp. 57-63.
Meldahl, P., Heggland, R., de Groot, P. and Bril, B., 1999, “The Chimney Cube, an Example of Semi-Automated Detection of Seismic Objects by Directive Attributes and Neural Networks: Part I-Methodology and Part II—Interpretation”, SEG International Expo and 69thAnnual Meeting, Houston.
Gao, D., 1999, “The First-Order and the Second-Order Seismic Textures: Implications for Quantitative Seismic Interpretation and Hydrocarbon Exploration”, AAPG Annual Mtg, San Antonio, TX, Apr. 11-14, 1999, p. A45.
Dawson Dwight C.
Eastwood John E.
Gross Michael D.
Johnston David H.
West Brian P.
ExxonMobil Upstream Research Company
McElheny Jr. Donald E.
Plummer J. Paul
LandOfFree
Method for classifying AVO data using an interpreter-trained... 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 classifying AVO data using an interpreter-trained..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for classifying AVO data using an interpreter-trained... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3148948