17:15 〜 18:45
[SCG40-P19] Seismic moment tensor estimation for the virtual earthquake in real-scale numerical rockbox simulations
キーワード：Discrete Element Method(DEM), Numerical Simulation, Seismic moment tensor
This study analyzes virtual earthquakes within a real-scale numerical rockbox experiment based on the Discrete Element Method (DEM). DEM simulations in 3D can directly investigate the granular layer's internal structure (Furuichi et al., 2018). They can reproduce geological-scale structures like the accretionary prisms and earthquake-like events within the structure (Furuichi et al., 2023). The quantification of events in such simulations will enhance understanding of the deformation in the virtual environment, thereby lessening the differences in replicating natural phenomena.
The current model simulates interactions between ~5.6 million single particle grains of radius 12.5m, treated as distinct entities undergoing the horizontal shortening of the granular rock layer of thin 3D geometry (100 km x 2 km x 250 m), which generates a thrust-like accretionary wedge. The plate convergence process is simulated in the longitudinal direction with the periodic boundary condition imposed in the transverse direction to minimize the influence of the thin boundary. Starting with nearly homogeneously layered structures, geological scaled structures, including the complex fault system, are created through a fast (0.5 m/s) convergence stage, after which the convergence speed is reduced to a slow (5e-4 m/s) rate. During this slow convergence, we observe several localized rapid deformation events with wave radiation resembling earthquake rupture (Furuichi et al., 2024). The hypocentres of these events are near the fault zone, developed within the geological structure.
To comprehend and quantify the source strength and orientation of the virtual earthquake, the seismic moment tensor M estimation plays a key role. However, since the moment tensor is a concept based on continuum media, a continuum representation of granular simulation results is required. Then, we developed a method to analyze the displacement gradient tensors across the model segments based on granular data to calculate the strain, thereby determining elastic stress. The model space is clipped 2500 m in x- and y-directions around the hypocentre and divided into grids of 50 m x 50 m x 50 m. The offset of the stress of the model from this corresponding elastic stress is defined as stress glut Γ(x,t). The seismic moment tensor can be calculated from the stress glut tensor by multiplying each tensor component with the model segment's volume. In this presentation, we will show the flow for the estimation of seismic moment tensor for such virtual earthquakes and preliminary results.
The current model simulates interactions between ~5.6 million single particle grains of radius 12.5m, treated as distinct entities undergoing the horizontal shortening of the granular rock layer of thin 3D geometry (100 km x 2 km x 250 m), which generates a thrust-like accretionary wedge. The plate convergence process is simulated in the longitudinal direction with the periodic boundary condition imposed in the transverse direction to minimize the influence of the thin boundary. Starting with nearly homogeneously layered structures, geological scaled structures, including the complex fault system, are created through a fast (0.5 m/s) convergence stage, after which the convergence speed is reduced to a slow (5e-4 m/s) rate. During this slow convergence, we observe several localized rapid deformation events with wave radiation resembling earthquake rupture (Furuichi et al., 2024). The hypocentres of these events are near the fault zone, developed within the geological structure.
To comprehend and quantify the source strength and orientation of the virtual earthquake, the seismic moment tensor M estimation plays a key role. However, since the moment tensor is a concept based on continuum media, a continuum representation of granular simulation results is required. Then, we developed a method to analyze the displacement gradient tensors across the model segments based on granular data to calculate the strain, thereby determining elastic stress. The model space is clipped 2500 m in x- and y-directions around the hypocentre and divided into grids of 50 m x 50 m x 50 m. The offset of the stress of the model from this corresponding elastic stress is defined as stress glut Γ(x,t). The seismic moment tensor can be calculated from the stress glut tensor by multiplying each tensor component with the model segment's volume. In this presentation, we will show the flow for the estimation of seismic moment tensor for such virtual earthquakes and preliminary results.