15:00 〜 15:15
[SCG45-06] Abrupt Change in Stress Field Near the Deep Plate Interface: Investigation Using Short-Period Moment Tensor Inversion
キーワード：モーメントテンソルインバージョン、低強度プレート境界、2003年宮城沖地震、プレート境界地震
Seismic waveforms provide information about both earthquake sources and wave propagation paths. For periods longer than several tens of seconds, the effects of wave propagation paths are generally well modeled, and source information is estimated through moment tensor inversion and slip process inversion. In such long-period waveform analysis, the minimum magnitude for which moment tensors can be estimated is approximately Mw3.5. If this minimum magnitude could be lowered to about Mw2.0, the amount of usable data would increase more than tenfold, potentially leading to significantly improved spatial resolution of fault structures and stress fields.
This study examines the possibility of estimating moment tensor solutions for Mw2.0-4.0 earthquakes through short-period waveform fitting. We used earthquake clusters near the source region of the 2003 Miyagi-oki Mw7.0 earthquake. Being deep earthquake clusters, their seismic waveforms are relatively interpretable even at short periods. The JMA catalog lists 3,476 M2-4 earthquakes from 2003 to July 2022.
We perform moment tensor inversion in the 2-5 Hz frequency band for direct P- and S-wave portions (0.6s). Initially, to explore the feasibility of using the 2-5 Hz frequency band, we compared observed and synthetic waveforms for events with known moment tensor solutions. The attenuation and density structures were preliminarily determined from empirical relationships using 1D Vp and Vs structures, similar to Yoshida et al. (2024, JGR). Comparison of observed and synthetic waveform amplitudes showed that observed amplitudes tended to be about half of the synthetic ones, though there was little variation among stations.
We modified the Q model and station correction values using least squares method to minimize the discrepancy between observed and synthetic amplitude ratios. Using the revised Q model and station corrections, we performed moment tensor inversion for 3,476 M2-4 earthquakes.
The analysis yielded reliable moment tensor solutions for 2,789 earthquakes. The obtained moment tensor solutions showed clear depth dependence. Near the plate interface, focal mechanisms are characterized by low-angle thrust-type earthquakes with nodal planes parallel to the plate boundary, while strike-slip earthquakes dominate in the hanging wall just above. Below the plate interface, mechanism characteristics change significantly, showing reverse-fault earthquakes with P-axes parallel to the plate boundary over a depth range of about 15 km. Further deeper, focal mechanisms dramatically change to normal-fault earthquakes with T-axes parallel to the plate boundary. This change in focal mechanisms aligns well with the predicted transition from plate boundary earthquakes to down-dip compression and down-dip tension stress fields within the slab, supporting the validity of our estimated mechanisms.
According to the obtained focal mechanisms, the stress field becomes nearly parallel to the plate boundary just 1 km below the interface. This suggests that the stress field within the slab is determined primarily by internal deformation accumulated during its own subduction process, with minimal influence from contact with the overriding plate. Under such a stress field, little shear stress should act on the plate boundary. The overlying plate boundary showed little activity before the Tohoku earthquake but experienced frequent repeating earthquakes after the event. Nevertheless, it remains predominantly in a creeping state. Our results may indicate that this deep plate boundary accumulates very little strain. Even in such an environment, stress might concentrate on a small number of seismic patches, leading to frequent repeating earthquakes.
This study examines the possibility of estimating moment tensor solutions for Mw2.0-4.0 earthquakes through short-period waveform fitting. We used earthquake clusters near the source region of the 2003 Miyagi-oki Mw7.0 earthquake. Being deep earthquake clusters, their seismic waveforms are relatively interpretable even at short periods. The JMA catalog lists 3,476 M2-4 earthquakes from 2003 to July 2022.
We perform moment tensor inversion in the 2-5 Hz frequency band for direct P- and S-wave portions (0.6s). Initially, to explore the feasibility of using the 2-5 Hz frequency band, we compared observed and synthetic waveforms for events with known moment tensor solutions. The attenuation and density structures were preliminarily determined from empirical relationships using 1D Vp and Vs structures, similar to Yoshida et al. (2024, JGR). Comparison of observed and synthetic waveform amplitudes showed that observed amplitudes tended to be about half of the synthetic ones, though there was little variation among stations.
We modified the Q model and station correction values using least squares method to minimize the discrepancy between observed and synthetic amplitude ratios. Using the revised Q model and station corrections, we performed moment tensor inversion for 3,476 M2-4 earthquakes.
The analysis yielded reliable moment tensor solutions for 2,789 earthquakes. The obtained moment tensor solutions showed clear depth dependence. Near the plate interface, focal mechanisms are characterized by low-angle thrust-type earthquakes with nodal planes parallel to the plate boundary, while strike-slip earthquakes dominate in the hanging wall just above. Below the plate interface, mechanism characteristics change significantly, showing reverse-fault earthquakes with P-axes parallel to the plate boundary over a depth range of about 15 km. Further deeper, focal mechanisms dramatically change to normal-fault earthquakes with T-axes parallel to the plate boundary. This change in focal mechanisms aligns well with the predicted transition from plate boundary earthquakes to down-dip compression and down-dip tension stress fields within the slab, supporting the validity of our estimated mechanisms.
According to the obtained focal mechanisms, the stress field becomes nearly parallel to the plate boundary just 1 km below the interface. This suggests that the stress field within the slab is determined primarily by internal deformation accumulated during its own subduction process, with minimal influence from contact with the overriding plate. Under such a stress field, little shear stress should act on the plate boundary. The overlying plate boundary showed little activity before the Tohoku earthquake but experienced frequent repeating earthquakes after the event. Nevertheless, it remains predominantly in a creeping state. Our results may indicate that this deep plate boundary accumulates very little strain. Even in such an environment, stress might concentrate on a small number of seismic patches, leading to frequent repeating earthquakes.