Global Seismic Noise Wavelet-based Measure of Nonstationarity

Abstract

The properties of low-frequency global seismic noise, represented by continuous records for 24 years, 1997–2020, are investigated at 229 broadband stations located around the world. The property of waveforms, known as the Donoho–Johnston threshold, which separates the absolute values of the orthogonal wavelet coefficients into small and large is analyzed. The ratio of the number of large coefficients to their total number is determined by the dimensionless DJ index, which takes values from 0 to 1. The DJ index is considered as a measure of the nonstationarity of noise: the larger is the DJ index, the more is nonstationary the waveform. For each station, daily DJ index values are calculated. An auxiliary network of 50 reference points is introduced, the positions of which are determined by clustering the positions of seismic stations. For each reference point, a time series is constructed with a time step of 1 day, which is calculated as the median of daily DJ index values from the 5 nearest operable stations. For all pairs of reference points, the coherence between the DJ index values is estimated in a sliding time window of 365 days with an offset of 3 days, and the maximum values of the coherence function and the frequency at which the maximum coherence is reached are determined. The average values of the maximum coherences show strong growth after 2003, and the maximum distances between the reference points, for which the maximum coherence exceeded the threshold of 0.9, undergo an explosive increase in values after 2012. By extrapolating and averaging the DJ index values at the reference points, the region of concentration of maximum DJ index values was determined at the North-East Siberia. The bursts in the mean value of the maximum coherences between the day length and the DJ index values at the control points precede the release of seismic energy with a delay of about 530 days.