11:30 AM - 11:45 AM
[SGD03-10] Detection of afterslip in the 2024 Noto Peninsula Earthquake using InSAR analysis
Keywords:InSAR, afterslip, 2024 Noto Peninsula Earthquake
In large earthquakes, post-seismic deformation occurs after the event and continues slowly. The causes include afterslip, viscoelastic relaxation, and poroelastic rebound. Afterslip refers to slow fault slip following an earthquake, typically lasting months to years. Following the 2024 Noto Peninsula earthquake (Mw 7.6) on January 1, GNSS observations detected post-seismic deformation over a wide area centered on the Noto Peninsula (Geographical Survey Institute). However, the cause of this deformation remains unclear. While afterslip has been reported in past major earthquakes, its location, scale, and relationship with aftershock distribution is not well understood. In this study, we use InSAR (Interferometric Synthetic Aperture Rader), a space geodetic technique detecting ground displacement accurately and spatially, to estimate the afterslip distribution and elucidate fault movement mechanisms. If we extract displacement spatially, we may capture afterslip undetectable by GNSS, a point observation technique.
We created InSAR images using 14 SLC (Single Look Complex) images from ALOS-2 SM1-mode, acquired between January 12 and August 23, 2024. InSAR images contain noise from orbital, topography, ionosphere, and atmosphere, in addition to surface displacement signals. To improve detection accuracy, we applied noise corrections. The Split-Spectrum method (Gomba et al., 2016) was used for ionospheric correction. Atmospheric delay corrections followed a GNSS-based approach (Kinoshita, 2022). We set a reference point in Anamizu Town, located south of the Noto Peninsula, where afterslip effects were assumed to be minimal. Displacements along the coastline near the fault were extracted and analyzed as a time-series graph. The reference and displacement points were selected from pixels with coherence ≧0.5 in the first interferometric pair (January 12 and January 26, 2024), and the average displacement was calculated within 200 m radius around these points. A forward model for viscoelastic relaxation was created using a program by Dr. Yukitoshi Fukahata (Kyoto University) and evaluated by comparison with displacement maps in GNSS observations made using the F5 solution.
The time-series graph shows 1 cm ~ 3 cm displacement away from the satellite between January 12 and April 19, 2024. Since time-series graphs decay logarithmically and afterslip follows a logarithmic function (Stamps & Kreemer, 2024), it is possible that displacement due to afterslip was extracted, though viscoelastic relaxation cannot be ruled out. As this earthquake is a reverse-fault event with a NW-SE pressure axis, if afterslip occurred in the coseismic slip direction, displacement should be toward the satellite given the northbound orbit and rightward observation. To clarify the cause of the displacement, we developed a forward model of post-seismic deformation due to viscoelastic relaxation. Toyama Bay, over 80 km from the epicenter fault, is assumed to be due to viscoelastic relaxation, and the model parameters were adjusted to match these displacements. The fault dip angle varied from 35° to 60° in 5° increments, and RMSE (Root Mean Square Error) were calculated for each case. At 35°, RMSE was 8.705 mm (EW), 10.640 mm (NS), and 19.467 mm (vertical). At 60°, RMSE was 8.856 mm (EW), 13.373 mm (NS), and 16.173 mm (vertical). Smaller dip angles fit better in EW and NS, while larger dip angles matched better in the vertical component. However, Toyama Bay displacements remain inconsistent, requiring further fault parameter adjustments.
In the future, we will refine the viscoelastic relaxation model and subtract its displacement from InSAR results to evaluate whether the logarithmic decay signal in the time-series graphs is due to afterslip.
We created InSAR images using 14 SLC (Single Look Complex) images from ALOS-2 SM1-mode, acquired between January 12 and August 23, 2024. InSAR images contain noise from orbital, topography, ionosphere, and atmosphere, in addition to surface displacement signals. To improve detection accuracy, we applied noise corrections. The Split-Spectrum method (Gomba et al., 2016) was used for ionospheric correction. Atmospheric delay corrections followed a GNSS-based approach (Kinoshita, 2022). We set a reference point in Anamizu Town, located south of the Noto Peninsula, where afterslip effects were assumed to be minimal. Displacements along the coastline near the fault were extracted and analyzed as a time-series graph. The reference and displacement points were selected from pixels with coherence ≧0.5 in the first interferometric pair (January 12 and January 26, 2024), and the average displacement was calculated within 200 m radius around these points. A forward model for viscoelastic relaxation was created using a program by Dr. Yukitoshi Fukahata (Kyoto University) and evaluated by comparison with displacement maps in GNSS observations made using the F5 solution.
The time-series graph shows 1 cm ~ 3 cm displacement away from the satellite between January 12 and April 19, 2024. Since time-series graphs decay logarithmically and afterslip follows a logarithmic function (Stamps & Kreemer, 2024), it is possible that displacement due to afterslip was extracted, though viscoelastic relaxation cannot be ruled out. As this earthquake is a reverse-fault event with a NW-SE pressure axis, if afterslip occurred in the coseismic slip direction, displacement should be toward the satellite given the northbound orbit and rightward observation. To clarify the cause of the displacement, we developed a forward model of post-seismic deformation due to viscoelastic relaxation. Toyama Bay, over 80 km from the epicenter fault, is assumed to be due to viscoelastic relaxation, and the model parameters were adjusted to match these displacements. The fault dip angle varied from 35° to 60° in 5° increments, and RMSE (Root Mean Square Error) were calculated for each case. At 35°, RMSE was 8.705 mm (EW), 10.640 mm (NS), and 19.467 mm (vertical). At 60°, RMSE was 8.856 mm (EW), 13.373 mm (NS), and 16.173 mm (vertical). Smaller dip angles fit better in EW and NS, while larger dip angles matched better in the vertical component. However, Toyama Bay displacements remain inconsistent, requiring further fault parameter adjustments.
In the future, we will refine the viscoelastic relaxation model and subtract its displacement from InSAR results to evaluate whether the logarithmic decay signal in the time-series graphs is due to afterslip.