5:15 PM - 6:45 PM
[SCG40-P02] SSE activity before the 2011 Tohoku Earthquake:
A revisit based on the reprocessed OBPR data
Keywords:Ocean Bottom Pressure-gauge, Slow Slip Events, Tohoku Earthquake
We analyzed data recorded by the ocean bottom pressure recorder (OBPR) off Miyagi, Japan, prior to the 2011 Tohoku earthquake (M 9.0), and investigated the characteristics of the slow slip activity in detail. During the period, two slow-slip events have been reported by previous studies: Afterslip of the largest (M 7.3) foreshock on March 9 (Ohta et al., 2012) and a spontaneous slow slip event (SSE) starting in mid-February on the up-dip side of the rupture zone of the M 7.3 foreshock (Ito et al., 2013).
To extract crustal deformation components from OBPR data, non-tidal oceanic deformation components (ocean fluctuation), with close time constants of tectonic signals and with amplitudes of ~ hPa (equivalent to ~ cm in vertical seafloor motions), need to be properly processed. In several previous studies, including Ito et al. (2013), took pressure differences across nearby stations to remove the fluctuations assuming their spatial coherency. However, Fredrickson et al. (2019) and Inoue et al. (2023) found that the similarity of seafloor pressure time series depends more strongly on the depth difference than on the horizontal distance between stations. In the Tohoku OBPR observation, the water depth difference between TJT1 and GJT3, located at the trenchward side, is very large and the differential pressure method may not work well in the reduction of the ocean fluctuation. Alternatively, we used a noise processing method combining an ocean model and principal component analysis (PCA) (Otsuka et al., 2023a). This method removes ocean fluctuation by calculating the difference between the output of the ocean model and the observed data, and then removes the common mode components remaining due to model misfit by subtracting principal components. This method allows us to discuss by absolute pressure (vertical displacement) rather than relative pressure (tilt), in terms of seafloor deformation.
For ocean models, MRI.COM-JPN (MRI model) was used in the present study. The model is based on multi-layer modeling with high horizontal resolution (Sakamoto et al., 2019) and its better performance over other models has been confirmed by Otsuka et al. (2023b). Hirata et al. (2023) pointed out that its good reproducibility in the long (> days) period fluctuations helps to improve the sensor drift components, often misestimated due to the presence of long-period pressure variations.
Based on the OBPR data processed with the new method, we analyzed the postseismic deformation associated with the largest foreshock on March 9. The signal-to-noise ratio of the OBPR records was improved from those used by Ohta et al. (2012) to investigate the temporal variation of the postseismic changes in OBPR data more in detail. At several OBPR stations, the postseismic pressure changes seem to be delayed from the origin time of the foreshock. We found a correlation between the onset timing of the pressure change and the epicentral distance from the foreshock, suggesting propagation of the afterslip from the epicenter. Using the reprocessed data, we will investigate the spatiotemporal evolution of the afterslip in detail, considering the M 6.8 event, which may affect the evolution of aseismic processes during Mar 9 and Mar, 11, when the mainshock happened.
Prior to the largest foreshock, possible tectonic signals exceeding the noise level were detected at the stations different from those, on which Ito et al. (2013) identified SSE related signals. Pressure changes interpreted as seafloor uplift were observed at P09 and P08, close to the rupture area of the M7.3 foreshock, whereas subsidence was observed at P07 and P03, downdip side of the M 7.3 source. The new observation indicates that the slow slip responsible for the deformation occurred more on the down dip side compared to the location of the SSE estimated by Ito et al. (2013). Notably, the source location of the SSE suggested by the reprocessed OBPR data is close to the epicenter of the mainshock of the Tohoku earthquake.
To extract crustal deformation components from OBPR data, non-tidal oceanic deformation components (ocean fluctuation), with close time constants of tectonic signals and with amplitudes of ~ hPa (equivalent to ~ cm in vertical seafloor motions), need to be properly processed. In several previous studies, including Ito et al. (2013), took pressure differences across nearby stations to remove the fluctuations assuming their spatial coherency. However, Fredrickson et al. (2019) and Inoue et al. (2023) found that the similarity of seafloor pressure time series depends more strongly on the depth difference than on the horizontal distance between stations. In the Tohoku OBPR observation, the water depth difference between TJT1 and GJT3, located at the trenchward side, is very large and the differential pressure method may not work well in the reduction of the ocean fluctuation. Alternatively, we used a noise processing method combining an ocean model and principal component analysis (PCA) (Otsuka et al., 2023a). This method removes ocean fluctuation by calculating the difference between the output of the ocean model and the observed data, and then removes the common mode components remaining due to model misfit by subtracting principal components. This method allows us to discuss by absolute pressure (vertical displacement) rather than relative pressure (tilt), in terms of seafloor deformation.
For ocean models, MRI.COM-JPN (MRI model) was used in the present study. The model is based on multi-layer modeling with high horizontal resolution (Sakamoto et al., 2019) and its better performance over other models has been confirmed by Otsuka et al. (2023b). Hirata et al. (2023) pointed out that its good reproducibility in the long (> days) period fluctuations helps to improve the sensor drift components, often misestimated due to the presence of long-period pressure variations.
Based on the OBPR data processed with the new method, we analyzed the postseismic deformation associated with the largest foreshock on March 9. The signal-to-noise ratio of the OBPR records was improved from those used by Ohta et al. (2012) to investigate the temporal variation of the postseismic changes in OBPR data more in detail. At several OBPR stations, the postseismic pressure changes seem to be delayed from the origin time of the foreshock. We found a correlation between the onset timing of the pressure change and the epicentral distance from the foreshock, suggesting propagation of the afterslip from the epicenter. Using the reprocessed data, we will investigate the spatiotemporal evolution of the afterslip in detail, considering the M 6.8 event, which may affect the evolution of aseismic processes during Mar 9 and Mar, 11, when the mainshock happened.
Prior to the largest foreshock, possible tectonic signals exceeding the noise level were detected at the stations different from those, on which Ito et al. (2013) identified SSE related signals. Pressure changes interpreted as seafloor uplift were observed at P09 and P08, close to the rupture area of the M7.3 foreshock, whereas subsidence was observed at P07 and P03, downdip side of the M 7.3 source. The new observation indicates that the slow slip responsible for the deformation occurred more on the down dip side compared to the location of the SSE estimated by Ito et al. (2013). Notably, the source location of the SSE suggested by the reprocessed OBPR data is close to the epicenter of the mainshock of the Tohoku earthquake.