Seismic Pattern Recognition.
Abstract
The paper applies various pattern recognition techniques to the problem of discriminating underground nuclear explosion seismic signals from those of natural earthquakes. The data set consists of 186 short period seismograms recorded by the Large Aperture Seismic Array in Montana. Enough seismic and pattern recognition theory is developed to allow easy interpretation of the results. The techniques tested run the gamut from simple, linear classifiers to a complex, adaptive, nonlinear classifier. The best results were 99.4% correct identification of earthquakes and 94.2% correct identification of explosions. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1972
- Accession Number
- AD0757877
Entities
People
- Ronald D. Bouvier
Organizations
- Air Force Institute of Technology