9:00 AM - 9:15 AM
[SSS07-13] Estimation of source properties with foreshocks on a 4-meter-long laboratory fault using synthetic Green's function
★Invited Papers
Keywords:Rock friction experiment, inversion of source mechanisms, waveform modeling
The local seismic events within the simulated fault occur during the stick-slip experiments, many of which can be categorized as foreshocks. Those events could be triggered by a precursory aseismic slip as a byproduct of the nucleation (Yamashita et al., 2021, Nat. Commun.), while some of them may contribute to the nucleation of main fault rupture as a part of the cascade-up process (McLaskey, 2019, JGR). The source properties of foreshocks are thus of great interest to clarify their interaction with the evolution of aseismic slip and the nucleation of main ruptures.
To investigate them, we conducted stick-slip experiments using biaxial rock friction apparatus with a 4-meter-long metagabbro fault to reproduce the foreshocks. We applied the normal stress up to 6 MPa and increased the shear stress to cause stick-slip events. The acoustic signals were recorded using 32 broadband AE sensors (Olympus V103-RM) installed 70mm below the simulated fault surface aligned along the fault. The signal was amplified by 40 dB and then sampled at 10MHz. The sensor gains were estimated using a signal from ball-drop sources, which was generated by a steel ball with a diameter of 3 mm dropped from 500 mm of height onto the fault surface (Okubo et al., 2021, AGU). We calibrated the recorded electrical signals using the estimated gains and then obtained the ground velocity generated by foreshocks although the frequency response is assumed to be flat in this calibration.
We then computed Green’s function associated with each pair between a foreshock and AE sensors using the reciprocity theorem for the sake of computational efficiency. We used 3D FDM-based software, OpenSWPC (Maeda et al., 2017, EPS) to compute Green’s functions. We discretized the domain with a grid size of 0.5 mm enough to resolve the frequency range of our analysis (~0.3 MHz). We synthesized the ground velocity waveform with hypothetical magnitudes and slip directions of the foreshock assuming the slip surface is constrained on the fault plane.
The location of foreshocks was determined using four AE sensors. Although a spatial error of ~40 mm arises due to the uncertainty in the pick of first arrival time, we used the most likely source location to compute Green’s functions for the vertical component on the side surface of the rock specimen. Given the slip direction parallel to that of shear loading, the comparison of a spectrum between the synthesized and observed waveforms provides the feasible estimation of seismic moment M0 within the range of laboratory foreshocks (e.g. Mw = -5.5). We further investigate the potential of evaluation in the source time function and the moment tensor inversion from the fitting of the waveform in the time domain. This analysis will help address the characteristics of foreshocks which would be governed by the local evolution of aseismic slip as well as evaluate its contribution to the release of the concentrated stress caused by the non-uniform stress distribution on the fault.
To investigate them, we conducted stick-slip experiments using biaxial rock friction apparatus with a 4-meter-long metagabbro fault to reproduce the foreshocks. We applied the normal stress up to 6 MPa and increased the shear stress to cause stick-slip events. The acoustic signals were recorded using 32 broadband AE sensors (Olympus V103-RM) installed 70mm below the simulated fault surface aligned along the fault. The signal was amplified by 40 dB and then sampled at 10MHz. The sensor gains were estimated using a signal from ball-drop sources, which was generated by a steel ball with a diameter of 3 mm dropped from 500 mm of height onto the fault surface (Okubo et al., 2021, AGU). We calibrated the recorded electrical signals using the estimated gains and then obtained the ground velocity generated by foreshocks although the frequency response is assumed to be flat in this calibration.
We then computed Green’s function associated with each pair between a foreshock and AE sensors using the reciprocity theorem for the sake of computational efficiency. We used 3D FDM-based software, OpenSWPC (Maeda et al., 2017, EPS) to compute Green’s functions. We discretized the domain with a grid size of 0.5 mm enough to resolve the frequency range of our analysis (~0.3 MHz). We synthesized the ground velocity waveform with hypothetical magnitudes and slip directions of the foreshock assuming the slip surface is constrained on the fault plane.
The location of foreshocks was determined using four AE sensors. Although a spatial error of ~40 mm arises due to the uncertainty in the pick of first arrival time, we used the most likely source location to compute Green’s functions for the vertical component on the side surface of the rock specimen. Given the slip direction parallel to that of shear loading, the comparison of a spectrum between the synthesized and observed waveforms provides the feasible estimation of seismic moment M0 within the range of laboratory foreshocks (e.g. Mw = -5.5). We further investigate the potential of evaluation in the source time function and the moment tensor inversion from the fitting of the waveform in the time domain. This analysis will help address the characteristics of foreshocks which would be governed by the local evolution of aseismic slip as well as evaluate its contribution to the release of the concentrated stress caused by the non-uniform stress distribution on the fault.