3:45 PM - 4:00 PM
[SGD02-07] Geometry of Median Tectonic Line (MTL) in eastern Shikoku based on GNSS and InSAR time series analysis
Keywords:Median Tectonic Line (MTL), fault, inter-seimic displacement, Global Navigation Satellite System (GNSS), Interferometric Synthetic Aperture Radar (InSAR)
Median tectonic Line (MTL) is one of the longest active fault zones in south-western Japan (Okada,1973). Although understanding fault movement is essential to assess earthquake potential, the fault movement of the MTL hasn’t been revealed completely.
In geodetic approach, the spatial resolution of deduced fault geometry depends on that of observed surface deformation. Today, InSAR (Interferometric Synthetic Aperture Radar) is known as a powerful remote sensing technique which can observe spatial extent of surface deformation with high spatial resolution. However, it has’t been accomplished to detect the inter-seismic displacement of the MTL by InSAR, since highly variable topography and dense vegetation induce the significant noise. Fortunately, L-band satellite ALOS-2, which are less affected by vegetation due to its long wavelength, has been operated from 2014, and we can use sufficient ALOS-2 images to conduct time series analysis. Also, atmospheric delay correction methods become more sophisticated. In this study, we tried detecting inter-seismic displacement of the MTL using GNSS (Global Navigation Satellite System) and InSAR and estimating the MTL’s geometry in eastern Shikoku.
In the GNSS analysis, we used GEONET daily coordinates (F5 solution) from 2005 to 2009 which less contain the effect of post-seismic process of Tohoku-oki earthquake and slow slip event near Bungo channel. Assuming crustal deformation around Shikoku had consisted in the inter-seismic displacement of the Nankai Trough subduction zone and that of the MTL, and assuming they hadn’t been changed in time, we calculated GNSS linear velocity in the south-western Japan. In the InSAR analysis, we used SAR data in 4 frames (Ascending: 128-670, 129-660, Descending: 21-2930, 21-2940) obtained by Stripmap mode observation by ALOS-2. We made interferograms for all possible interferometric pairs. We also applied the ionospheric phase correction (Gomba et al., 2015) and the neutral atmospheric correction (Kinoshita, 2022). We corrected the post-seismic deformation of Tohoku-oki earthquake based on Tobita (2016). After correction, we conducted time series analysis using LiCSBAS2 (Morishita, 2020) to obtain the LOS (Line of Sight) velocity fields. We fixed the reference of GNSS and InSAR derived velocity fields to the Amur micro plate using MORVEL (Demets et al., 2010) plate model. After that, we could derive InSAR velocity fields, which was consistent with GNSS velocities with RMSE less than 2 mm/yr.
Next, we corrected the effect of plate coupling along the Nankai Trough subduction zone to extract only the inter-seismic displacement of the MTL. We estimated the relative block motion rate of MTL and the slip deficit rate distribution on the Nankai Trough subduction zone by the linear least squares inversion with the rectangular dislocation model (Okada, 1992). We used 261 GNSS stations’ horizontal velocities around Shikoku except for stations near the MTL fault zone. The estimated MTL relative block motion rate in right-lateral direction was 4.3 mm/yr. After correcting the effect of the Nankai Trough based on estimated slip deficit distribution, we could recognize that LOS velocity decreases as the distance from the MTL increase in the northern side of the MTL for descending frames. This result was roughly consistent with the velocity field based on a simple forward model (lower limit of locking depth: 15 km, dip angle: 60 deg). On the other hand, on the southern side of MTL, and a whole scene of ascending frames, we couldn’t detect any clear MTL-related signal.
We estimated the MTL geometry (lower limit of locking depth and dip angle) using InSAR (Descending) derived velocity on the northern side of the MTL. We approximated the MTL fault plane by a rectangular plane. Changing the lower limit of locking depth and dip angle little by little, we estimated optimal parameters with the smallest residual that can explain the InSAR velocity in northern side of the MTL well. The upper edge of the fault plain was a prior fixed at the surface (depth: 0 km). We calculated the standard deviation of “InSAR based velocity – fault model-based velocity” for all fault models, and we found that the value became minimum when locking depth was 16 km and dip angle was 48 deg. According to geological study, locking depth and dip angle of “Sanuki-Sanmyaku Nanen Tobu segment”, which is one of the MTL segment and located in eastern Shikoku, are 10~15 km, 40~45 degree respectively. This roughly accords to our result. On the other hand, it was also reported that MTL seems to be vertical faults rather than north-dipping shallow faults based on fault quasi-dynamic boundary element method (Muramatsu et al. 2023, AGU).
In geodetic approach, the spatial resolution of deduced fault geometry depends on that of observed surface deformation. Today, InSAR (Interferometric Synthetic Aperture Radar) is known as a powerful remote sensing technique which can observe spatial extent of surface deformation with high spatial resolution. However, it has’t been accomplished to detect the inter-seismic displacement of the MTL by InSAR, since highly variable topography and dense vegetation induce the significant noise. Fortunately, L-band satellite ALOS-2, which are less affected by vegetation due to its long wavelength, has been operated from 2014, and we can use sufficient ALOS-2 images to conduct time series analysis. Also, atmospheric delay correction methods become more sophisticated. In this study, we tried detecting inter-seismic displacement of the MTL using GNSS (Global Navigation Satellite System) and InSAR and estimating the MTL’s geometry in eastern Shikoku.
In the GNSS analysis, we used GEONET daily coordinates (F5 solution) from 2005 to 2009 which less contain the effect of post-seismic process of Tohoku-oki earthquake and slow slip event near Bungo channel. Assuming crustal deformation around Shikoku had consisted in the inter-seismic displacement of the Nankai Trough subduction zone and that of the MTL, and assuming they hadn’t been changed in time, we calculated GNSS linear velocity in the south-western Japan. In the InSAR analysis, we used SAR data in 4 frames (Ascending: 128-670, 129-660, Descending: 21-2930, 21-2940) obtained by Stripmap mode observation by ALOS-2. We made interferograms for all possible interferometric pairs. We also applied the ionospheric phase correction (Gomba et al., 2015) and the neutral atmospheric correction (Kinoshita, 2022). We corrected the post-seismic deformation of Tohoku-oki earthquake based on Tobita (2016). After correction, we conducted time series analysis using LiCSBAS2 (Morishita, 2020) to obtain the LOS (Line of Sight) velocity fields. We fixed the reference of GNSS and InSAR derived velocity fields to the Amur micro plate using MORVEL (Demets et al., 2010) plate model. After that, we could derive InSAR velocity fields, which was consistent with GNSS velocities with RMSE less than 2 mm/yr.
Next, we corrected the effect of plate coupling along the Nankai Trough subduction zone to extract only the inter-seismic displacement of the MTL. We estimated the relative block motion rate of MTL and the slip deficit rate distribution on the Nankai Trough subduction zone by the linear least squares inversion with the rectangular dislocation model (Okada, 1992). We used 261 GNSS stations’ horizontal velocities around Shikoku except for stations near the MTL fault zone. The estimated MTL relative block motion rate in right-lateral direction was 4.3 mm/yr. After correcting the effect of the Nankai Trough based on estimated slip deficit distribution, we could recognize that LOS velocity decreases as the distance from the MTL increase in the northern side of the MTL for descending frames. This result was roughly consistent with the velocity field based on a simple forward model (lower limit of locking depth: 15 km, dip angle: 60 deg). On the other hand, on the southern side of MTL, and a whole scene of ascending frames, we couldn’t detect any clear MTL-related signal.
We estimated the MTL geometry (lower limit of locking depth and dip angle) using InSAR (Descending) derived velocity on the northern side of the MTL. We approximated the MTL fault plane by a rectangular plane. Changing the lower limit of locking depth and dip angle little by little, we estimated optimal parameters with the smallest residual that can explain the InSAR velocity in northern side of the MTL well. The upper edge of the fault plain was a prior fixed at the surface (depth: 0 km). We calculated the standard deviation of “InSAR based velocity – fault model-based velocity” for all fault models, and we found that the value became minimum when locking depth was 16 km and dip angle was 48 deg. According to geological study, locking depth and dip angle of “Sanuki-Sanmyaku Nanen Tobu segment”, which is one of the MTL segment and located in eastern Shikoku, are 10~15 km, 40~45 degree respectively. This roughly accords to our result. On the other hand, it was also reported that MTL seems to be vertical faults rather than north-dipping shallow faults based on fault quasi-dynamic boundary element method (Muramatsu et al. 2023, AGU).