Materials of the International Conference
50th Anniversary of the International Geophysical Year
and Electronic Geophysical Year

16-19 September 2007 • Suzdal, Russia

Complex electromagnetic monitoring of Corinth Gulf seismic area (Greece): Fuzzy logic algorithms time series pattern recognition

J. Zlotnicki1, A. Gvishiani2, J. Le Mouël3, M. Rodkin2, S. Agayan2, Sh. Bogoutdinov2, and G. Vargemezis4

1UMR 6524, CNRS-OPGC, Clermont-Ferrand, France

2Geophysical Center, Russian Academy of Sciences, Moscow, Russia

3Laboratoire de Gèomagnvtisme, IPGP, France

4Aristotvles Laboratory, University, Thessaloniki, Greece

Abstract

During several decades numerous studies focused on the search of electric, EM and/or magnetic precursors of earthquakes. In particular, electric signals in the ULF band (less than 10 Hz) are used in search of possible precursor anomalies. This study is devoted to examination of such data obtained at La Corinth Gulf seismic area monitoring system. This monitoring system is recognized as the Earth ground components of the CNES DEMETER Program aimed at the search and evaluation of pre-seismic signals. The La Corinth Gulf monitoring system is used to validate EM signals recorded on board of Demeter satellite. The system of monitoring installed in La Corinth Gulf seismic area gives an enormous flow of magnetic, electric, EM, and seismic data that hardly can be examined manually. Besides, the character of the data obtained is very complicated, and the non-formal criteria and even intuition are used in experts' data processing. Thus, an elaboration of a methodology of data processing and studies based on artificial intelligence approach is needed. We introduce the approach providing an alternative to the visual analysis of data by experts. The used algorithms of data processing are elaborated on the basis of artificial intelligence, fuzzy-logics and statistics. These algorithms carry out the morphological examination of time series with the purpose of allocating in them the segments with the needed (arbitrary) character features of a signal. Both, a sample of a real record or some ideal pattern formulated by expert, can be used to form the procedure of recognition. Results of the application of these algorithms had shown an opportunity of allocation of anomalies in electric and magnetic data, and distinguishing between anomalies of different morphology corresponding to different physical influences. These algorithms can be used in systems of monitoring (for (semi)automation of work of expert) and/or with the research purposes – for revealing characteristic morphological sequences in long rows of data.

Citation: J. Zlotnicki, A. Gvishiani, J. Le Mouël, M. Rodkin, S. Agayan, Sh. Bogoutdinov, and G. Vargemezis (2007), Complex electromagnetic monitoring of Corinth Gulf seismic area (Greece): Fuzzy logic algorithms time series pattern recognition, in: Materials of the International Conference '50th Anniversary of the International Geophysical Year and Electronic Geophysical Year', GC RAS, Moscow, doi:10.2205/2007-IGY50conf.

© 2007 Geophysical Center RAS and authors


Webmaster