5:15 PM - 6:45 PM
[U15-P65] Relationship between the coseismic slip distribution and its assumed fault geometry in the 2024 Noto Peninsula earthquake derived from very dense geodetic observation data
Keywords:The 2024 Noto Peninsula earthquake, GNSS, SAR, Slip distribution estimation, Bayesian inversion, Hamiltonian Monte Carlo
The Noto Peninsula earthquake that occurred on January 1, 2024, was a reverse fault earthquake with a northwest-southeast pressure axis, and large crustal deformation of up to 2 m horizontally and 4 m vertically has been obtained from various observations. Intense seismic swarms have been observed in the vicinity of the epicenter since the end of 2020, and surface displacement data have been obtained from GEONET and GNSS stations at cell phone base stations by SoftBank Corp., as well as temporal stations by Kyoto University and Kanazawa University.
In this presentation, we construct and discuss the static slip distribution model of the earthquake based on the surface displacement data obtained from these observation networks deployed near the epicenter. The Hamiltonian Monte Carlo method, a Markov chain Monte Carlo method, was used simultaneously to estimate the intensity parameter of Laplacian smoothing.
First, two rectangular faults were placed to match the aftershock distribution of the aftershocks relocated by the double-difference method. The two faults correspond to the Saruyama-oki and Wajima-oki segments and the Suzu-oki segments. We tested different fault geometries by changing the dip angles of the two faults within 20-70°. In the estimation, the standard deviation of the likelihood function was set larger than usual to consider the possibility of pillar tilting due to strong ground shaking and the uncertainty of Green's function. The results show that the slip distribution has two or three slip peaks with 4-7 m maximum slips in each dip angle case. Two peaks are located on the southwestern fault, corresponding to the prominent uplift in the northern coastal areas of Suzu and Wajima, as suggested by the SAR analysis. On the northeastern fault, slip was estimated at the junction of the two faults, and a small peak of 2-3 m of slip appeared on the northeast side only when the two faults were low-angle. However, the posterior distribution of slip in this area has a large confidence interval (the 60% credit interval > 5 m) due to its location far from the onshore observation network, and it is difficult to discuss whether the rupture indeed extended to this area or not.
Changes in the dip angle of the southwestern fault strongly affected the reproducibility of the displacement data. The low-angle southwestern fault explained nearby horizontal displacement well, while the high-angle fault reproduced vertical displacement better. The high-angle case (~50°) slightly overestimated the northwestward displacement observed at more distant stations, such as Toyama and Niigata prefectures. These results suggest that the source fault of the earthquake may not have a simple geometry and that more complex fault model assumptions (e.g., a listric fault with a larger dip angle at shallower depths) are needed. In addition, to improve the data resolution, especially in the northwestern part of the peninsula, we perform a joint inversion using the results of SAR Pixel Offset analysis and discuss the fault model in detail.
Acknowledgments: The SoftBank's GNSS observation data used in this study was provided by SoftBank Corp. and ALES Corp. through the framework of the "Consortium to utilize the SoftBank original reference sites for Earth and Space Science". We used open ALOS-2 data through JAXA EORC web site. The RINC software, which was developed by Dr. Taku Ozawa and provided through PIXEL, was used for ALOS-2 data processing.
In this presentation, we construct and discuss the static slip distribution model of the earthquake based on the surface displacement data obtained from these observation networks deployed near the epicenter. The Hamiltonian Monte Carlo method, a Markov chain Monte Carlo method, was used simultaneously to estimate the intensity parameter of Laplacian smoothing.
First, two rectangular faults were placed to match the aftershock distribution of the aftershocks relocated by the double-difference method. The two faults correspond to the Saruyama-oki and Wajima-oki segments and the Suzu-oki segments. We tested different fault geometries by changing the dip angles of the two faults within 20-70°. In the estimation, the standard deviation of the likelihood function was set larger than usual to consider the possibility of pillar tilting due to strong ground shaking and the uncertainty of Green's function. The results show that the slip distribution has two or three slip peaks with 4-7 m maximum slips in each dip angle case. Two peaks are located on the southwestern fault, corresponding to the prominent uplift in the northern coastal areas of Suzu and Wajima, as suggested by the SAR analysis. On the northeastern fault, slip was estimated at the junction of the two faults, and a small peak of 2-3 m of slip appeared on the northeast side only when the two faults were low-angle. However, the posterior distribution of slip in this area has a large confidence interval (the 60% credit interval > 5 m) due to its location far from the onshore observation network, and it is difficult to discuss whether the rupture indeed extended to this area or not.
Changes in the dip angle of the southwestern fault strongly affected the reproducibility of the displacement data. The low-angle southwestern fault explained nearby horizontal displacement well, while the high-angle fault reproduced vertical displacement better. The high-angle case (~50°) slightly overestimated the northwestward displacement observed at more distant stations, such as Toyama and Niigata prefectures. These results suggest that the source fault of the earthquake may not have a simple geometry and that more complex fault model assumptions (e.g., a listric fault with a larger dip angle at shallower depths) are needed. In addition, to improve the data resolution, especially in the northwestern part of the peninsula, we perform a joint inversion using the results of SAR Pixel Offset analysis and discuss the fault model in detail.
Acknowledgments: The SoftBank's GNSS observation data used in this study was provided by SoftBank Corp. and ALES Corp. through the framework of the "Consortium to utilize the SoftBank original reference sites for Earth and Space Science". We used open ALOS-2 data through JAXA EORC web site. The RINC software, which was developed by Dr. Taku Ozawa and provided through PIXEL, was used for ALOS-2 data processing.