Application of Self-Organizing Feature Maps to Reservoir Characterization

  • C. A. Link
  • J. Conaway
Part of the Studies in Fuzziness and Soft Computing book series (STUDFUZZ, volume 80)


Neural networks are becoming an increasingly popular analysis tool in a wide variety of applications. One application of considerable interest is reservoir characterization. Calculating reservoir properties has typically been a process of extrapolation or interpolation from a sparse number of locations using geostatistical methods. A neural networks approach allows estimation of reservoir parameters from not only the sparse set of well locations, but also an auxiliary data set such as a complete grid of 3-D seismic data.

This chapter discusses the novel application of unsupervised neural networks for reservoir characterization. The unsupervised approach is compared to conventional supervised neural network approaches to illustrate the fundamental differences. Two examples of supervised neural network parameter estimation are presented followed by an introduction and application of unsupervised neural networks. The three cases can be summarized as: 1) estimation based on spatial information only, 2) estimation guided by seismic data, and 3) analysis using an unsupervised neural network approach.

A numerical study described in the first approach shows that it compares reasonably well with conventional geostatistical approaches. The second or guided approach is a practical way to associate seismic data with well data over the entire area of a prospect. Validation results for this method are also good, and the example shown emphasizes that results are highly data dependent. The third or unsupervised approach comprises a novel method for analyzing a 3-D seismic data volume in a relatively unbiased manner. This technique gives two useful results: first, class or correlation maps showing similarities of data traces over the prospect area and second, output or weight vectors that represent prototype seismic traces associated with a particular output class. These prototype traces have proved to be very useful for further analysis of the data volume. An example of the application of the unsupervised approach to a producing oil field in Montana is described in which a prediction for a well location based on the neural network analysis proved successful on drilling.


Weight Vector Input Vector Seismic Data Seismic Attribute Reservoir Characterization 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. An, Ping and Moon, Wooil M., 1993, Reservoir characterization using feedforward neural networks: 63rd Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, 93, 258–262.Google Scholar
  2. An, Ping, 1994, The effect of random noise in lateral reservoir characterization using feedforward neural networks: 64th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, 94, 787–790.Google Scholar
  3. Cybenko, G., 1988, Continuous valued neural networks with two hidden layers are sufficient: Technical Report, Department of Computer Science, Tufts University, Medford, MA.Google Scholar
  4. Cybenko, G., 1989, Approximation by superpositions of a sigmoidal function: Mathematics of Control, Signals and Systems 2, 303–314.CrossRefGoogle Scholar
  5. de Groot, P. F. M., Campbell, A. E., Kavli, T. and Melnyk, D., 1993, Reservoir characterization from 3D seismic data using artificial neural networks and stochastic modeling techniques: 55th Mtg. Eur. Assoc. Expl Geophys., Abstracts, 93, Session:B047.Google Scholar
  6. Demuth, H. and Beale, M., 1994, Neural network toolbox user’s guide: MATLAB® — The Mathworks, Inc.Google Scholar
  7. Demuth, H. and Beale, M., 2000, Neural network toolbox user’s guide: MATLAB® — The Mathworks, Inc.Google Scholar
  8. Hagan, M.T., Demuth, H.B., and Beale, M., 1996, Neural network design: PWS Publishing Co.Google Scholar
  9. Hertz, J., Krogh, A., and Palmer, R.G., 1991, Introduction to the theory of neural computation: Addison-Wesley Publishing Company.Google Scholar
  10. Himmer, P. and Link, C.A., 1997, Reservoir porosity prediction from 3-D seismic data using neural networks: 67th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, 97, 842–845.Google Scholar
  11. Hornik, K., Stinchcombe, M., and White, H., 1989, Multilayer feedforward networks are universal approximators, Neural Networks 2, 359–366.CrossRefGoogle Scholar
  12. Johnston, David H., 1993, Seismic attribute calibration using neural networks: 63rd Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 93, 250–253.Google Scholar
  13. Kohonen, T., 1989, Self-organization and associative memory: 3rd Ed., Berlin: Springer-Verlag.CrossRefGoogle Scholar
  14. Nelson, M.M. and Illingworth, W.T., 1991, A practical guide to neural nets: Addison-Wesley Publishing Company, Inc.Google Scholar
  15. Ronen, S., Hoskins, J., Schultz, P. S., Hattori, M. and Corbett, C., 1994, Seismic-guided estimation of log properties, part 2: Using artificial neural networks for nonlinear attribute calibration*: The Leading Edge, 13, no. 6, 674–678. (* Erratum in TLE-13–8-822)CrossRefGoogle Scholar
  16. Rummelhart, D. E., Hinton, G. E., and Williams, R. J., 1986, Learning representations by back-propagating errors: Nature, 323, 533–536.CrossRefGoogle Scholar
  17. Taner, M. T., Koehler, F. and Sheriff, R. E., 1979, Complex seismic trace analysis*: Geophysics, 44, no. 6, 1041–1063.CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2002

Authors and Affiliations

  • C. A. Link
    • 1
  • J. Conaway
    • 2
  1. 1.Dept. of Geophysical EngineeringMontana Tech of The University of MontanaButteUSA
  2. 2.Conoco CenterUSA

Personalised recommendations