5:15 PM - 6:30 PM
[SSS08-P17] Comparison of the spectral characteristics of deep low-frequency earthquakes with regular earthquakes around the Gifu-Nagano prefectural border
Keywords:source spectra
Source spectra provide essential information to understand the spatiotemporal evolution of the fault rupture during an earthquake. Aki (1967) proposed the omega-square source model, in which the high-frequency side of the source spectrum decreases by the square of the frequency, which is consistent with broad observations. The omega-square model is widely used in analyses in seismology as a standard model of the source spectrum.
However, the actual source spectra of earthquakes likely have some diversity. Some previous studies suggest that the fall-off of the source spectrum decays linearly with frequency for low-frequency earthquakes that occurred along the plate boundary (Ide et al., 2007) and inside the plate (Yoshida et al., 2020). In this study, we estimated the source spectra of regular and deep low-frequency earthquakes that occurred near the Gifu-Nagano prefectural border from April 1, 2020, to May 11, 2020, and examined their shapes in detail. The seismic observation network surrounds this region, and many deep low-frequency earthquakes occurred.
For the estimation of source spectra, we first estimated Q-1(f) around this region and the site-amplification factors A(f) of seismic stations independently by the coda normalization method (Aki and Chouet, 1975 ; Aki, 1980) following Takahashi et al. (2005) and Yoshida et al. (2017). We used the coda normalization method because this method does not assume the omega-square model. We used the waveforms of 44 earthquakes with Mjma≧3.0 for the estimation of Q-1(f) and A(f). We then used the derived Q-1(f) and A(f) to estimate the source spectra of 350 regular earthquakes (Mjma≧2.0) and 35 deep low-frequency earthquakes.
As a measure of the spectral shape, we used the exponent n of a source spectral. We determined seismic moment, corner frequency, and the exponent n by fitting the theoretical source spectra to the observed one. For the regular earthquakes, the mean value and the standard deviation of n are 2.05 and 0.25, respectively. For the deep low-frequency earthquakes, the mean value and the standard deviation of n are 1.93 and 0.17, respectively. The mean value of deep low-frequency earthquakes is slightly lower, but the difference is not significant. Both types of earthquakes basically follow the omega-square model. The estimated values of n have some diversity, which may reflect the diversity of the spatiotemporal evolution of earthquake rupture.
This study assumes that Q-1(f) is homogeneous in space. However, since deep low-frequency earthquakes occurred much deeper than regular earthquakes, the representative Q-1(f) may actually be different from regular earthquakes. The obtained spectral shape of deep low-frequency earthquakes may be systematically biased because the assumed Q-1(f) was derived from regular earthquakes. A future study taking the spatial variation of Q-1(f) into account may clarify the detail of the diversity of source spectra.
However, the actual source spectra of earthquakes likely have some diversity. Some previous studies suggest that the fall-off of the source spectrum decays linearly with frequency for low-frequency earthquakes that occurred along the plate boundary (Ide et al., 2007) and inside the plate (Yoshida et al., 2020). In this study, we estimated the source spectra of regular and deep low-frequency earthquakes that occurred near the Gifu-Nagano prefectural border from April 1, 2020, to May 11, 2020, and examined their shapes in detail. The seismic observation network surrounds this region, and many deep low-frequency earthquakes occurred.
For the estimation of source spectra, we first estimated Q-1(f) around this region and the site-amplification factors A(f) of seismic stations independently by the coda normalization method (Aki and Chouet, 1975 ; Aki, 1980) following Takahashi et al. (2005) and Yoshida et al. (2017). We used the coda normalization method because this method does not assume the omega-square model. We used the waveforms of 44 earthquakes with Mjma≧3.0 for the estimation of Q-1(f) and A(f). We then used the derived Q-1(f) and A(f) to estimate the source spectra of 350 regular earthquakes (Mjma≧2.0) and 35 deep low-frequency earthquakes.
As a measure of the spectral shape, we used the exponent n of a source spectral. We determined seismic moment, corner frequency, and the exponent n by fitting the theoretical source spectra to the observed one. For the regular earthquakes, the mean value and the standard deviation of n are 2.05 and 0.25, respectively. For the deep low-frequency earthquakes, the mean value and the standard deviation of n are 1.93 and 0.17, respectively. The mean value of deep low-frequency earthquakes is slightly lower, but the difference is not significant. Both types of earthquakes basically follow the omega-square model. The estimated values of n have some diversity, which may reflect the diversity of the spatiotemporal evolution of earthquake rupture.
This study assumes that Q-1(f) is homogeneous in space. However, since deep low-frequency earthquakes occurred much deeper than regular earthquakes, the representative Q-1(f) may actually be different from regular earthquakes. The obtained spectral shape of deep low-frequency earthquakes may be systematically biased because the assumed Q-1(f) was derived from regular earthquakes. A future study taking the spatial variation of Q-1(f) into account may clarify the detail of the diversity of source spectra.