10:15 AM - 10:30 AM
[S08-13] Evaluation of frequency-dependent attenuation factor for accurate estimation of source parameters in the laboratory earthquakes
To constrain the source mechanics of the seismic events caused by the asperity patch, we conducted a stick-slip experiment using a 4-meter-long biaxial rock friction apparatus by placing thin circular gouge patches on the simulated fault (Okubo et al., 2023AGU). The non-self-similar scaling, which is characterized by a nearly constant source duration with varying seismic moments (e.g., Harrington and Brodsky, 2009; Lin et al, 2016), was observed in the foreshocks and aftershocks generated by the gouge patch, where the estimated source duration was ~3.0 μs for the different events. This could simulate rupture mechanisms associated with the asperity patch in nature (e.g., Sammis and Rice, 2001), whereas the artifact caused by the attenuation of rock specimen needs to be corrected to adequately evaluate the moment-duration scaling using acoustic emission (AE) waveforms. For this purpose, we estimated the frequency-dependent attenuation factor of the rock specimen using the ball-drop impact onto the fault surface and then we corrected the observed waveforms, in addition to the instrumental response of the AE sensor.
The force-time history applied onto the surface of a rock sample caused by the ball-drop impact can be analytically approximated using the Hertzian impact theory (Reed 1985; McLaskey and Glaser, 2009). The AE waveforms radiated from the ball-drop impact were used to calibrate the amplitude factor of the sensor coupling in the previous study (Okubo et al., 2022AGU). We reanalyzed them to evaluate the attenuation factor by fitting them to the synthesized waveforms. Along the fault, we dropped a steel ball with a diameter of 2 mm from a height of 500 mm at 32 locations on the surface and then recorded the waveforms using the 32 AE sensors (V103-RM, EVIDENT; resonant frequency 1 MHz) installed near the source locations. We selected the source-sensor pairs showing a high similarity in the P wave pulse enough to evaluate the attenuation, especially in the high-frequency components (> 300 kHz). After the correction of the instrumental response, we estimated the frequency-dependent attenuation factor Q-1(f) by comparing the P wave spectra between the observed and the synthesized waveforms. Assuming that the attenuation factor is homogeneous over the rock specimen, we stacked the Q-1(f) inferred from the 68 source-sensor pairs to obtain the averaged attenuation model. The representative values of the estimated Q were approximately 120 and 30 for the 300 kHz and 700 kHz, respectively.
We deconvolved this attenuation model from the source time function (STF) of the observed seismic events generated by the gouge patch and evaluated the half maximum amplitude width of the STF, which was a metric of the half source duration proposed by Lin et al. (2016). The source durations after the correction of the attenuation factor remained nearly constant at ~2.5 μs with the seismic moment ranging from 0.5 to 1.0 Nm. Therefore, although the attenuation could affect the estimation of the source durations, we conclude that the observed non-self-similar scaling in the seismic events of the gouge patch was not caused by the artifact due to the attenuation with the assumption that the attenuation factor is uniform regardless of the amplitude of the waveforms. This analysis helps minimize the uncertainty associated with the path effect on the AE waveforms to investigate the source mechanisms of the laboratory earthquakes.
The force-time history applied onto the surface of a rock sample caused by the ball-drop impact can be analytically approximated using the Hertzian impact theory (Reed 1985; McLaskey and Glaser, 2009). The AE waveforms radiated from the ball-drop impact were used to calibrate the amplitude factor of the sensor coupling in the previous study (Okubo et al., 2022AGU). We reanalyzed them to evaluate the attenuation factor by fitting them to the synthesized waveforms. Along the fault, we dropped a steel ball with a diameter of 2 mm from a height of 500 mm at 32 locations on the surface and then recorded the waveforms using the 32 AE sensors (V103-RM, EVIDENT; resonant frequency 1 MHz) installed near the source locations. We selected the source-sensor pairs showing a high similarity in the P wave pulse enough to evaluate the attenuation, especially in the high-frequency components (> 300 kHz). After the correction of the instrumental response, we estimated the frequency-dependent attenuation factor Q-1(f) by comparing the P wave spectra between the observed and the synthesized waveforms. Assuming that the attenuation factor is homogeneous over the rock specimen, we stacked the Q-1(f) inferred from the 68 source-sensor pairs to obtain the averaged attenuation model. The representative values of the estimated Q were approximately 120 and 30 for the 300 kHz and 700 kHz, respectively.
We deconvolved this attenuation model from the source time function (STF) of the observed seismic events generated by the gouge patch and evaluated the half maximum amplitude width of the STF, which was a metric of the half source duration proposed by Lin et al. (2016). The source durations after the correction of the attenuation factor remained nearly constant at ~2.5 μs with the seismic moment ranging from 0.5 to 1.0 Nm. Therefore, although the attenuation could affect the estimation of the source durations, we conclude that the observed non-self-similar scaling in the seismic events of the gouge patch was not caused by the artifact due to the attenuation with the assumption that the attenuation factor is uniform regardless of the amplitude of the waveforms. This analysis helps minimize the uncertainty associated with the path effect on the AE waveforms to investigate the source mechanisms of the laboratory earthquakes.