Accession Number:

ADA257769

Title:

Data to Test and Evaluate the Performance of Neural Network Architectures for Seismic Signal Discrimination: Data Sets 2-3. Volume 1

Descriptive Note:

Final rept. 1 Sep 1990-31 Aug 1992

Corporate Author:

SCIENCE APPLICATIONS INTERNATIONAL CORP SAN DIEGO CA

Report Date:

1992-08-31

Pagination or Media Count:

80.0

Abstract:

This study describes a data set that was developed to test and evaluate the performance of neural networks for automated processing and interpretation of regional seismic data. This data set may also be valuable for other applications related to seismic monitoring at regional distances, and it is available at the Center for Seismic Studies CSS in an Oracle database or in UNIX tar format on exabyte tapes. It consists of waveform and parametric data from 500 regional events recorded by the short-period elements of the NORESS and ARCESS arrays in Norway, and the GERESS array in Germany the Oracle database at CSS also included data from the FINESA array in Finland and a 3- component station in Poland called KSP. The epicentral distances are primarily 50-2000 km, and the magnitudes are primarily 1.0-5.0. Most of the events are mining explosions in the western part of the CIS, Sweden, Finland, Poland, and Germany. Also included are 22 presumed underwater explosions, and 51 earthquakes in Fennoscandia that were identified in a regional bulletin produced by the University of Helsinki. Other presumed earthquakes for which independent bulletin information was not available include events in the Alps and Mediterranean region that were recorded by GERESS. The Oracle database is in CSS 3.0 format, and the exabyte tapes include waveforms in SAC binary format and parametric data in ASCII tables. Detailed documentation has been developed for each event, and is included in a 13-volume report at the CSS.

Subject Categories:

  • Seismology
  • Cybernetics
  • Explosions

Distribution Statement:

APPROVED FOR PUBLIC RELEASE