09:15 〜 09:30
[SCG45-14] Spatiotemporal Evolution of Aseismic Slip Preceding the 2011 Tohoku-Oki Earthquake by Reprocessed OBPR Data
キーワード：海底水圧観測、2011年東北地方太平洋沖地震、非地震性すべり
Approximately two days before the Tohoku-oki earthquake (M 9.0, 2011/03/11 14:46, JST; Mainshock), the largest foreshock (M 7.3, 2011/03/09 11:45; Foreshock #1) occurred in the northeastern side of Mainshock hypocenter. After that, there was remarkable foreshock activity, including the second-largest foreshock (M6.8 2011/03/10 6:23; Foreshock #2). Additionally, afterslip following Foreshock #1 has been identified. While such precursory activity has been confirmed, this area has experienced relatively large (M~7) earthquakes with high frequency, such as “Miyagi-oki earthquake” occurring at approximately 30-year intervals in the downdip side of Mainshock, and an M8.1 earthquake in 1981 on the northern side of Mainshock. However, none of these past earthquakes led to a rupture of the M9 earthquake. To understand the differences between the foreshocks in 2011 and past M~7 earthquakes, it is necessary to capture the activity following Foreshock #1 in more detail and estimate its influence on Mainshock rupture. The objective of this study is to estimate the spatiotemporal evolution of slip on the plate interface, including aseismic and coseismic slip of Foreshocks #1 and #2, using OBPR data processed with the latest noise reduction techniques, and to clarify what happened before Mainshock.
The afterslip distribution following Foreshock #1 using GNSS and OBPR data previously, but a time-invariant slip distribution was assumed due to the high noise level in OBPR data. Subsequent studies have revealed that the similarity of non-tidal ocean fluctuation (ocean noise) in OBPR data strongly depends on the depths of stations. Furthermore, a new ocean was developed, which has a higher spatial resolution than the model used for noise reduction in previous studies. Taking these advancements in the understanding of ocean noise characteristics, we reanalyzed the OBPR data to enhance tectonic signals associated with afterslip.
From the afterslip signal and the coseismic vertical displacement recorded in OBPR data for Foreshock #1 and #2, we estimated the seismic and aseismic slip distributions. Since the number of observed data points was smaller than the number of model parameters, we used L1-norm regularized least square method, a type of sparse modeling, to estimate the slip magnitude for each subfault. Using displacement at each time step, we estimated a spatiotemporal evolution of the cumulative slip. As a result, it was successfully imaged that aseismic slip expanded southward from Foreshock #1, and also occurred downdip side of Mainshock hypocenter.
A comparison between the spatiotemporal evolution of aseismic slip and the seismic slip distributions of Foreshock #1 and #2 revealed that the coseismic slip of Foreshock #2 occurred at the edge of the aseismic slip that developed around Foreshock #1. Subsequently, aseismic slip occurred around the Mainshock hypocenter, indicating a cascading development of seismic and aseismic slip. While previous studies suggested this behavior from seismicity, this study is the first to provide evidence using geodetic data. Furthermore, the aseismic slip reaccelerated approximately seven hours before Mainshock.
In this study, we revealed three major behaviors before Mainshock such as aseismic slip downdip of Mainshock, the southward expansion of aseismic slip from the Foreshock #1 toward Mainshock, and the reacceleration of aseismic slip immediately before Mainshock. In contrast, for the 2005 Miyagi-oki earthquake (Mw 6.8), aftershock activity and repeating earthquake analyses suggested that aseismic slip occurred in the northern region of Mainshock hypocenter. However, aseismic slip was not suggested in the downdip side of Mainshock hypocenter, or the updip side of the 2005 event. Therefore, stress loading from both northern and downdip side of Mainshock hypocenter, as well as the reacceleration of slip may have contributed to the occurrence of the M9 earthquake.
The afterslip distribution following Foreshock #1 using GNSS and OBPR data previously, but a time-invariant slip distribution was assumed due to the high noise level in OBPR data. Subsequent studies have revealed that the similarity of non-tidal ocean fluctuation (ocean noise) in OBPR data strongly depends on the depths of stations. Furthermore, a new ocean was developed, which has a higher spatial resolution than the model used for noise reduction in previous studies. Taking these advancements in the understanding of ocean noise characteristics, we reanalyzed the OBPR data to enhance tectonic signals associated with afterslip.
From the afterslip signal and the coseismic vertical displacement recorded in OBPR data for Foreshock #1 and #2, we estimated the seismic and aseismic slip distributions. Since the number of observed data points was smaller than the number of model parameters, we used L1-norm regularized least square method, a type of sparse modeling, to estimate the slip magnitude for each subfault. Using displacement at each time step, we estimated a spatiotemporal evolution of the cumulative slip. As a result, it was successfully imaged that aseismic slip expanded southward from Foreshock #1, and also occurred downdip side of Mainshock hypocenter.
A comparison between the spatiotemporal evolution of aseismic slip and the seismic slip distributions of Foreshock #1 and #2 revealed that the coseismic slip of Foreshock #2 occurred at the edge of the aseismic slip that developed around Foreshock #1. Subsequently, aseismic slip occurred around the Mainshock hypocenter, indicating a cascading development of seismic and aseismic slip. While previous studies suggested this behavior from seismicity, this study is the first to provide evidence using geodetic data. Furthermore, the aseismic slip reaccelerated approximately seven hours before Mainshock.
In this study, we revealed three major behaviors before Mainshock such as aseismic slip downdip of Mainshock, the southward expansion of aseismic slip from the Foreshock #1 toward Mainshock, and the reacceleration of aseismic slip immediately before Mainshock. In contrast, for the 2005 Miyagi-oki earthquake (Mw 6.8), aftershock activity and repeating earthquake analyses suggested that aseismic slip occurred in the northern region of Mainshock hypocenter. However, aseismic slip was not suggested in the downdip side of Mainshock hypocenter, or the updip side of the 2005 event. Therefore, stress loading from both northern and downdip side of Mainshock hypocenter, as well as the reacceleration of slip may have contributed to the occurrence of the M9 earthquake.