The statistics of the time segments of low-frequency microseisms: Trends and synchronization

The problem of identifying the effects of synchronization in the parameters of low-frequency microseismic noise from the data of 77 stations belonging to the F-net broadband seismograph network in Japan for the period from the beginning of 1997 through August, 2009 is considered. The vertical components measured initially with a sampling rate of one second and subsequently converted into the signals sampled at 1 minute intervals by means of averaging and decimation are used in the analysis. Six statistics are taken as the parameters: the support width of the multifractal singularity spectrum; the generalized Hurst exponent; the asymmetry coefficient of the spectrum of singularity; the logarithmic variance; the spectral exponent; and the linear predictability index. These parameters are calculated from the realizations contained within consecutive daily time intervals. When using the moving time window with a width of one year for evaluating the multiple correlation, the daily variations in the median values of the statistics of the noise measured at five spatial clusters of stations exhibit a stable increase in the synchronization not long before the Hokkaido earthquake (September 25, 2003; M = 8.3), subsequently passing to the new level of high synchronization. Based on the analysis of the trends in the index of linear predictability it turned out possible to estimate the beginning of the enhancement in the synchronization with rather high accuracy as the middle of 2002. The effect revealed for the variations in the different parameters of microseisms is an independent argument for the earlier conclusion about the synchronization in the field of the microseismic noise on the Japan Islands.