Fall-off rate of amplitude spectra of deep low-frequency tremors around Kii Peninsula, Japan, estimated from array observation

Slow earthquakes have been seismologically and geodetically observed in subduction zones around the world since 2000s [e.g. Obara, 2002]. Several studies have investigated source characteristics of slow earthquakes from amplitude spectra. However, the fall-off rate of amplitude spectra of slow earthquakes has been controversial, that is, f^-1 [Ide et al., 2007] or f^-2 [Fletcher et al., 2011; Zhang et al., 2011].

We analyzed velocity waveform data recorded from February 18, 2011 to November 9, 2016 by a seismic array in Kii Peninsula, Japan, which was composed of 39 seismographs operated by AIST, Japan. The sampling frequency of waveform data is 200 Hz. To improve S/N of waveforms, we conducted semblance analysis of velocity waveforms filtered at 2-8 Hz and stacked those aligned based on the slowness with the maximum semblance. Then, we converted the stacked velocity waveform to displacement one by integrating in time domain. For the spectral analysis of deep low-frequency tremors, we used a time window of 10 s, which includes the maximum amplitude. We estimated amplitude spectra by applying FFT to the displacement waveform with a correction of anelastic attenuation (Q = 300) [Yabe and Ide, 2014]. We examine the shape of amplitude spectra in the range of 1 to 10 Hz, where the tremors spectra are generally higher than noise one.

The amplitude spectra of tremors show a falloff as f^-1 rather than f^-2. We also estimated amplitude spectra of regular and low-frequency earthquakes, showing a falloff as f^-2 and f^-1, respectively. These results suggest that the source mechanism of tremors is different from that of regular earthquakes. However, the falloff as f^-2 is obtained at frequencies higher than 3-8 Hz for Cascadia tremors [Zhang et al., 2011]. Further discussion needs the precise estimation of tremor source spectra in a broader frequency band, especially at frequencies higher than 10 Hz.