[S-CG45] Science of slow-to-fast earthquakes

[SCG45-P44] Quasi-dynamic simulations of subduction zone earthquake sequences considering viscoelastic relaxation in asthenosphere

*Ryoya Matsushima¹, So Ozawa², Ryosuke Ando¹ (1.Graduate School of Science, University of Tokyo, 2.Stanford University)

Keywords:Viscoelastic relaxation, Quasi-dynamic earthquake sequences simulation, The 2011 off the Pacific coast of Tohoku Earthquake

Crustal deformation was observed over a wide area of Japan after the 2011 off the Pacific coast of Tohoku Earthquake. For example, the long-wavelength component of the strain rate in the northern Niigata-Kobe Tectonic Zone (NKTZ) turned from contraction to elongation after the earthquake, while the short-wavelength component continued contracting (Meneses-Gutierrez and Sagiya, 2016). Fukahata et al. 2020 pointed out that viscoelastic relaxation in the asthenosphere contributed to the long-wavelength component of crustal deformation, while ductile shear deformation at the deep part of faults contributed to the short-wavelength component. Therefore, it is necessary to consider the viscoelastic effect to simulate crustal deformation after a huge earthquake. However, previous works using the boundary integral equations method (BIEM) cannot consider the 3D structure of the viscoelastic media. For example, Hashima et al. 2008 used viscoelastic Green’s function, however, this function is only for the deformation in a multilayered half-space. In addition, few studies dealt with the mechanical interaction between plate boundary slip and inland faults in the framework of simulations of sequences of earthquakes and aseismic slip (SEAS). Thus, it is impossible to precisely incorporate the 3D structure of the subduction zone into a simulation.
Therefore, we suggest a new method that put faults with rate- and state-dependent friction (RSF) law and lithosphere-asthenosphere boundary (LAB) with viscous resistance in the elastic media and calculate by quasi-dynamic simulation using BIEM. This approach is similar to the methods in which bulk viscoelasticity is considered as viscoelastic relaxation at the boundary in elastic media (e.g. Duan and Oglesby, 2005, and Miyake and Noda, 2019). To validate, the results from this method were also compared with the analytical solution of Savage and Prescott, 1978.

Methods
In elastic half-space, a vertical transform fault was placed in the 0~60km depth and a horizontal surface imitating LAB was placed at the depth of 60km to compare with the analytical solution of Savage and Prescott, 1978 which modeled seismic fault slip and resulting afterslip due to bulk viscoelasticity. Viscous resistance proportional to velocity (e.g. Ando et al. 2012) was set as the stress boundary condition for the LAB where v is the displacement velocity at the upper surface of the LAB and w is the width of the region where deformation occurs in the asthenosphere. An instantaneous left-lateral slip was introduced on the fault as a backslip and simulated the media displacement, i.e., the LAB displacement. Displacement at the free surface was then calculated and compared with the analytical solution of Savage and Prescott, 1978.

Results
A preliminary analysis using the simulation code hbi (Ozawa et al. 2022) to calculate surface displacements under the same boundary condition as Savage and Prescott, 1978 produced spatial patterns qualitatively similar to their solution.
A more detailed comparison with the analytical solution will be presented on the day. The presentation will also show the simulation results of a model that takes into account the 3D structure of the subduction zone using this method.