10:45 AM - 12:15 PM
[SCG45-P42] Estimation of rake angle of MTLAFS by quasi-dynamic earthquake sequence simulation: development of a model based on stress field and GNSS strain-velocity fields and validation using paleoseismic surveys
Keywords:MTLAFS, quasi-dynamic earthquake sequence simulation
The Median Tectonic Line Fault Zone (MTLFZ) is the longest active fault zone in Japan with predominantly right-lateral strike-slip faulting (Ikeda et al., 2009). It is uncertain whether faulting along the MTLFZ occurs on the vertical or high-angle dipping surface (Median Tectonic Line Active Fault System, MTLAFS), as indicated by its lateral faulting and surface displacement (Tsutsumi and Okada, 1996, etc.), or on the north-dipping surface (Material Boundary, MBMTL), as observed by seismic reflection (Ito et al., 1996, Sato et al., 2015, etc.). Uchide et al. (2022) calculated the slip tendency Ts'=|τ|/(μ|σ|) (Yukutake et al. (2015)) distribution for all of Japan using the J-SHIS database, which records the location, strike, and dip angle of active faults and source mechanisms of about 220,000 events (Mj>0.5) of small earthquakes. For the area around Shikoku in the MTLFZ, they obtained Ts'>0.95 when the dip angle was assumed to be 90°, and 0.6<=Ts'<=0.7 when the dip angle was assumed to be 40. This study aims to estimate the fault plane dip angle of MTLAFS by performing a quasi-dynamic earthquake sequence simulation using the stress and strain fields obtained from observations as input parameters.
We newly constructed two three-dimensional non-planar fault geometry models of a vertical fault plane and a north-dipping fault plane of the MTL, and performed quasi-dynamic earthquake sequential simulations on both models. The mean recurrence interval and mean displacement velocity obtained from the simulations are then compared with their observed values from geomorphological and geological studies (e.g., Okada, 1970, Goto et al., 2003) to evaluate the validity of the respective shape models. In the simulations, the rake angle at each fault plane is calculated from the stress field obtained by stress inversion (Uchide et al., 2022). The stress loading rates in the normal and tangential directions of the fault plane are calculated from the strain-velocity distribution obtained from GNSS data, from which the contribution of elastic deformation due to interplate adhesion has been removed (Nishimura, 2022). Numerical computations were performed using the highly efficient quasi-dynamic boundary element method code HBI with the lattice H-matrix implemented in Ozawa et al. (2022), using 144 nodes of Wisteria/BDEC-01 for about 5 hours, with a size of about 200,000 elements and 300,000 time steps.
The results show that the mean horizontal and vertical displacement velocities for the vertical model were consistent with their observed value from the trench survey, differing by at most a factor. On the other hand, in the case of the north-dipping model, the absolute value of the mean vertical displacement velocity was about 2 m/kyr in the western part of the southern margin of the Sanuki Mountains and 3 m/kyr in the Negoro segment, which is about an order of magnitude higher than the observed values of 0.11-0.4 m/kyr (Okada, 1978) and 0.4 m/kyr (Okada & Samukawa, 1978), respectively. Therefore, the results of this simulation show that the simulated fault activity of the vertical model is closer to the actual fault activity than the north-dipping model when comparing the mean displacement velocities. The mean displacement velocity in the eastern part of the Iyonada segment is 1-2 m/kyr based on a trench survey (Okada et al., 1998a), which is smaller than the mean horizontal displacement velocity of 5-6 m/kyr (Goto, 1996) at a trench survey point in the western segment of the Ishizuchi Mountains north margin segment located about 50 km to the east. Simulation results assuming that the two segments are continuous in the plane also gave in faster mean horizontal displacement velocities at the former trench point than the observed value. A simulation using a model with both segments as discontinuous, as the small mean horizontal displacement velocity at the eastern end of the Iyo segment was considered to be an effect of the segment boundary, the mean horizontal displacement velocity at the eastern end of the Iyo segment was smaller due to the boundary condition, showed that in agreement with the trench survey results.
We newly constructed two three-dimensional non-planar fault geometry models of a vertical fault plane and a north-dipping fault plane of the MTL, and performed quasi-dynamic earthquake sequential simulations on both models. The mean recurrence interval and mean displacement velocity obtained from the simulations are then compared with their observed values from geomorphological and geological studies (e.g., Okada, 1970, Goto et al., 2003) to evaluate the validity of the respective shape models. In the simulations, the rake angle at each fault plane is calculated from the stress field obtained by stress inversion (Uchide et al., 2022). The stress loading rates in the normal and tangential directions of the fault plane are calculated from the strain-velocity distribution obtained from GNSS data, from which the contribution of elastic deformation due to interplate adhesion has been removed (Nishimura, 2022). Numerical computations were performed using the highly efficient quasi-dynamic boundary element method code HBI with the lattice H-matrix implemented in Ozawa et al. (2022), using 144 nodes of Wisteria/BDEC-01 for about 5 hours, with a size of about 200,000 elements and 300,000 time steps.
The results show that the mean horizontal and vertical displacement velocities for the vertical model were consistent with their observed value from the trench survey, differing by at most a factor. On the other hand, in the case of the north-dipping model, the absolute value of the mean vertical displacement velocity was about 2 m/kyr in the western part of the southern margin of the Sanuki Mountains and 3 m/kyr in the Negoro segment, which is about an order of magnitude higher than the observed values of 0.11-0.4 m/kyr (Okada, 1978) and 0.4 m/kyr (Okada & Samukawa, 1978), respectively. Therefore, the results of this simulation show that the simulated fault activity of the vertical model is closer to the actual fault activity than the north-dipping model when comparing the mean displacement velocities. The mean displacement velocity in the eastern part of the Iyonada segment is 1-2 m/kyr based on a trench survey (Okada et al., 1998a), which is smaller than the mean horizontal displacement velocity of 5-6 m/kyr (Goto, 1996) at a trench survey point in the western segment of the Ishizuchi Mountains north margin segment located about 50 km to the east. Simulation results assuming that the two segments are continuous in the plane also gave in faster mean horizontal displacement velocities at the former trench point than the observed value. A simulation using a model with both segments as discontinuous, as the small mean horizontal displacement velocity at the eastern end of the Iyo segment was considered to be an effect of the segment boundary, the mean horizontal displacement velocity at the eastern end of the Iyo segment was smaller due to the boundary condition, showed that in agreement with the trench survey results.