Signature Matching For Seismic Signal Identification

Authors

Pankaj Jadhav
Homi Bhabha National Institute Mumbai-400094, India
Debabrata Datta
Homi Bhabha National Institute Mumbai-400094, India
Siddhartha Mukhopadhyay
Homi Bhabha National Institute, Mumbai-400094, India

Synopsis

Seismic signals can be classified as natural or manmade by matching signature of similar events that have occurred in the past. Waveform matching techniques can be effectively used for discrimination since the events with similar location and focal mechanism have similar waveform irrespective of magnitude. The seismic signals are inherently non-stationary in nature. The analysis of such signals can be best achieved in multiresolution framework by resolving the signal using continuous wavelet transform (CWT) in time-frequency plane. In this paper similarity testing and classification of nuclear explosion and earthquake are exploited with correlation, continuous wavelet transform, cross-wavelet transform and wavelet coherence (WC) of P phase of seismogram. Clustering of seismic signals continuous wavelet spectra is performed using maximum covariance analysis. The proposed classifier has an average classification accuracy of 94%.

 

WREC21
Published
September 22, 2021
Online ISSN
2582-3922