skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Seismic event classification system

Abstract

In the computer interpretation of seismic data, the critical first step is to identify the general class of an unknown event. For example, the classification might be: teleseismic, regional, local, vehicular, or noise. Self-organizing neural networks (SONNs) can be used for classifying such events. Both Kohonen and Adaptive Resonance Theory (ART) SONNs are useful for this purpose. Given the detection of a seismic event and the corresponding signal, computation is made of: the time-frequency distribution, its binary representation, and finally a shift-invariant representation, which is the magnitude of the two-dimensional Fourier transform (2-D FFT) of the binary time-frequency distribution. This pre-processed input is fed into the SONNs. These neural networks are able to group events that look similar. The ART SONN has an advantage in classifying the event because the types of cluster groups do not need to be pre-defined. The results from the SONNs together with an expert seismologist's classification are then used to derive event classification probabilities.

Inventors:
 [1];  [2];  [3]
  1. (Castro Valley, CA)
  2. (Brentwood, CA)
  3. (Livermore, CA)
Publication Date:
Research Org.:
Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
OSTI Identifier:
869651
Patent Number(s):
US 5373486
Assignee:
United States Department of Energy (Washington, DC) LLNL
DOE Contract Number:
W-7405-ENG-48
Resource Type:
Patent
Country of Publication:
United States
Language:
English
Subject:
seismic; event; classification; computer; interpretation; data; critical; step; identify; example; teleseismic; regional; local; vehicular; noise; self-organizing; neural; networks; sonns; classifying; events; kohonen; adaptive; resonance; theory; useful; purpose; detection; corresponding; signal; computation; time-frequency; distribution; binary; representation; finally; shift-invariant; magnitude; two-dimensional; fourier; transform; 2-d; fft; pre-processed; input; fed; look; similar; sonn; advantage; types; cluster; pre-defined; results; expert; seismologist; derive; probabilities; seismic data; seismic event; neural network; fourier transform; neural networks; frequency distribution; neural net; /367/181/702/

Citation Formats

Dowla, Farid U., Jarpe, Stephen P., and Maurer, William. Seismic event classification system. United States: N. p., 1994. Web.
Dowla, Farid U., Jarpe, Stephen P., & Maurer, William. Seismic event classification system. United States.
Dowla, Farid U., Jarpe, Stephen P., and Maurer, William. Sat . "Seismic event classification system". United States. doi:. https://www.osti.gov/servlets/purl/869651.
@article{osti_869651,
title = {Seismic event classification system},
author = {Dowla, Farid U. and Jarpe, Stephen P. and Maurer, William},
abstractNote = {In the computer interpretation of seismic data, the critical first step is to identify the general class of an unknown event. For example, the classification might be: teleseismic, regional, local, vehicular, or noise. Self-organizing neural networks (SONNs) can be used for classifying such events. Both Kohonen and Adaptive Resonance Theory (ART) SONNs are useful for this purpose. Given the detection of a seismic event and the corresponding signal, computation is made of: the time-frequency distribution, its binary representation, and finally a shift-invariant representation, which is the magnitude of the two-dimensional Fourier transform (2-D FFT) of the binary time-frequency distribution. This pre-processed input is fed into the SONNs. These neural networks are able to group events that look similar. The ART SONN has an advantage in classifying the event because the types of cluster groups do not need to be pre-defined. The results from the SONNs together with an expert seismologist's classification are then used to derive event classification probabilities.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Sat Jan 01 00:00:00 EST 1994},
month = {Sat Jan 01 00:00:00 EST 1994}
}

Patent:

Save / Share:
  • In the computer interpretation of seismic data, the critical first step is to identify the general class of an unknown event. For example, the classification might be: teleseismic, regional, local, vehicular, or noise. Self-organizing neural networks (SONNs) can be used for classifying such events. Both Kohonen and Adaptive Resonance Theory (ART) SONNs are useful for this purpose. Given the detection of a seismic event and the corresponding signal, computation is made of: the time-frequency distribution, its binary representation, and finally a shift-invariant representation, which is the magnitude of the two-dimensional Fourier transform (2-D FFT) of the binary time-frequency distribution. Thismore » pre-processed input is fed into the SONNs. These neural networks are able to group events that look similar. The ART SONN has an advantage in classifying the event because the types of cluster groups do not need to be pre-defined. The results from the SONNs together with an expert seismologist's classification are then used to derive event classification probabilities. 21 figures.« less
  • An automated seismic processing system and method are disclosed, including an array of CMOS microprocessors for unattended battery-powered processing of a multi-station network. According to a characterizing feature of the invention, each channel of the network is independently operable to automatically detect, measure times and amplitudes, and compute and fit Fast Fourier transforms (FFT's) for both P- and S- waves on analog seismic data after it has been sampled at a given rate. The measured parameter data from each channel are then reviewed for event validity by a central controlling microprocessor and if determined by preset criteria to constitute amore » valid event, the parameter data are passed to an analysis computer for calculation of hypocenter location, running b-values, source parameters, event count, P- wave polarities, moment-tensor inversion, and Vp/Vs ratios. The in-field real-time analysis of data maximizes the efficiency of microearthquake surveys allowing flexibility in experimental procedures, with a minimum of traditional labor-intensive postprocessing. A unique consequence of the system is that none of the original data (i.e., the sensor analog output signals) are necessarily saved after computation, but rather, the numerical parameters generated by the automatic analysis are the sole output of the automated seismic processor.« less
  • An automated seismic processing system and method are disclosed, including an array of CMOS microprocessors for unattended battery-powered processing of a multi-station network. According to a characterizing feature of the invention, each channel of the network is independently operable to automatically detect, measure times and amplitudes, and compute and fit Fast Fourier transforms (FFT's) for both P-and S-waves on analog seismic data after it has been sampled at a given rate. The measured parameter data from each channel are then reviewed for event validity by a central controlling microprocessor and if determined by preset criteria to constitute a valid event,more » the parameter data are passed to an analysis computer for calculation of hypocenter location, running b-values, source parameters, event count, P- wave polarities, moment-tensor inversion, and Vp/Vs ratios. The in-field real-time analysis of data maximizes the efficiency of microearthquake surveys allowing flexibility in experimental procedures, with a minimum of traditional labor-intensive postprocessing. A unique consequence of the system is that none of the original data (i.e., the sensor analog output signals) are necessarily saved after computation, but rather, the numerical parameters generated by the automatic analysis are the sole output of the automated seismic processor.« less
  • The specification discloses a method and system wherein an operator interacts with an automatic data processing and display system to classify and sort subsurface reflective surfaces or segments in a multidimensional context. The technique utilizes a properly programmed general purpose digital computer in combination with a plurality of storage tube display screens. Seismic reflection data in the form of computer picked reflection segments, hereinafter referred to as seismic segment data, produced from previous computer processing steps is entered into the computer. Displays of the velocity, amplitude, dip and length of the segments in a space gate are then provided onmore » the display screens. A data responsive surface having a network of conductive wires and a detecting stylus are utilized by the operator to operate upon the displayed segment data to classify primary segments and to define boundaries relative to each of the parameter displays. The computer sorts the data according to the defined boundaries and segments are displayed which meet the defined boundary conditions. Provision is made to enable easy alteration of the displayed segment parameters and designated boundaries.« less
  • In the computer interpretation of seismic data, the critical first step is to identify the general class of an unknown event. For example, the classification might be: teleseismic, regional, local, vehicular, or noise. We have studied Self Organizing Neural Networks (SONNs) for classifying such events. Both Kohonen and Adaptive Resonance Theory (ART) SONNs were developed and tested with a moderately large set of real seismic events. Given the detection of a seismic event and the corresponding signal, we compute the time-frequency distribution, its binary representation, and finally a shift-invariant representation, which is the magnitude of the two-dimensional Fourier transform (2-Dmore » FFT) of the binary time-frequency distribution. This preprocessed input is fed into the SONNs. The overall results based on 111 events (43 training and 68 test events) show that SONNs are able to group events that look'' similar. We also find that the ART algorithm has an advantage in that the types of cluster groups do not need to be predefined. When a new type of event is detected, the ART network is able to handle the event rather gracefully. The results from the SONNs together with an expert seismologist's classification are then used to derive event classification probabilities. A strategy to integrate a SONN into the interpretation of seismic events is also proposed.« less