Research Outline
Research Themes
- Postseismic viscoelastic response
- mantle dynamics
- interplate coupling
Summary of Research to Date
Our laboratory consistently tackles difficult problems in subduction zone research via quantitative approaches using numerical methods such as the finite element method and finite difference method, inversion analysis, and other analytical methods. The following is a summary of recent progress of our research published or submitted mainly to international journals. The number in parenthesis indicates the reference number in the list of research achievements (peer-reviewed).
1. Displacement, strain, and stress fields associated with megathrust earthquakes
To quantitatively evaluate the relationship between different earthquakes that occur in relation to megathrust earthquakes, we simulated the 1944 Tonankai (M7.9) and 1946 Nankai (M8.0) earthquakes, which occurred off the Kii Peninsula and Shikoku Island, and their subsequent earthquakes. We constructed a three-dimensional elastic finite element model that considers the complex geometry of the Philippine Sea plate subducting beneath southwest Japan, calculated the strain and stress fields associated with faulting, and compared them with the observed data. We quantitatively evaluated the abrupt change in the compressive stress field ranging from the Kii Channel to the eastern part of Shikoku Island from the north-south to the east-west direction after the Tonankai and Nankai earthquakes based on stress changes. We elucidated the spatiotemporal range where the stress field is dominant. We also quantitatively assessed the effects of the Tonankai and Nankai earthquakes on the stress accumulation process of large inland earthquakes along the Median Tectonic Line and clarified the mechanism for the increase in seismic activity in the Mikawa region and the sudden decrease in the seismic activity around Wakayama due to the Tonankai earthquake (2,3).
Furthermore, we constructed a three-dimensional viscoelastic finite element model to investigate the detailed effects of the three-dimensional plate geometry on the displacement, strain, and stress fields associated with faulting. The results of this study significantly differed from those of conventional horizontally stratified models, demonstrating the importance of considering the plate geometry when analyzing viscoelastic deformation after a megathrust earthquake (13). We applied this model to the 1993 Kushiro-oki earthquake (M7.5) (16) and the 1946 Nankai earthquake (17), and calculated the expected postseismic deformation (see Fig. 1).
2. Interplate coupling in subduction zones in the Pacific Rim
To estimate the spatiotemporal distribution of interplate coupling in the Nankai Trough, we constructed a 3-D finite element model similar to that reported in item 1 above using interseismic leveling and triangulation survey data. Then the coupling condition was assessed with forward modeling. The geodetic data revealed for the first time that the coupling is very strong in the Nankai Trough. Additionally, other areas with strong coupling coincide well with the slip areas of the Tonankai and Nankai earthquakes (5).
In the Tokai region (8) and Kanto region (9), inversion analysis using Akaike's Bayesian information criterion for leveling and trilateration survey data clarified the coupling state at plate boundaries. We successfully estimated the stress accumulation process and the direction of the relative plate motion for the upcoming Tokai and Kanto earthquakes.
In collaboration with our students and overseas researchers, we conducted similar inversion analyses using recent GPS data in the Nankai Trough (20), the Kuril-Japan Trench (21), Mexico (31), and Cascadia in Canada (33). Because megathrust earthquakes also occur in these areas of the Pacific Rim, we clarified the coupling state between the plates and estimated the potential of megathrust earthquakes ( see Animation 1). In Mexico, large-scale slow slip events were recently observed by GPS, and we analyzed this phenomenon.
3. Numerical simulations of a deeper portion of a slab (stress field, temperature and flow field, deformation, seismic wave propagation)
- To quantitatively elucidate the physical process of slab detachment (horizontal slab tearing at depths of 100 to 300 km), we constructed a three-dimensional viscoelastic finite element model. Then the spatiotemporal variation of stress in a slab was calculated by simulating an initial slab tear from one side and considering the balance of forces acting on the slab. The results revealed two interesting findings. (1) Under certain conditions, shear stress of several hundred MPa appears in the slab at the tear tip, which is sufficient stress to propagate further tearing in the slab. (2) A lower ambient viscosity in the mantle, longer tear length, and down-dip tension in the slab increase the stress concentration (10,11) (Figs. 2 and 3).
- To clarify the mechanism of deep earthquakes, we constructed a thermo-kinetic coupling model that accounts for the kinetics of the olivine phase transition and the temperature distribution in the slab. Furthermore, the spatial distribution of the bulk modulus, rigidity, and density difference with the surrounding mantle were obtained using the physical properties of olivine acquired in high-temperature and high-pressure experiments. Then we calculated the stresses in the slab due to the density difference using the finite element method. The synergistic effects of the negative buoyancy due to the olivine to wadsleyite phase transition and the buoyancy due to the ringwoodite (Rw) to perovskite (Pv) + magnesiowüstite (Mw) phase transition generate a high stress field with down-dip compression deep within the slab. This high stress field may be the source of the seismic stress for deep earthquakes (15).
- To elucidate the formation mechanism of the slab lying at the top of the lower mantle, we performed numerical simulations of temperature and flow using a two-dimensional finite difference method. The slab lies down easier due to the buoyancy caused by the Rw→Pv+Mw phase transition and a megalith may form due to the friction from the high viscosity of the lower mantle (27) (see Animation 2, Animation 3, Animation 4).
- To evaluate the effects of the metastable phase of olivine (α) in the slab on seismic waveforms, we constructed a seismic wave velocity structure model that included temperature, pressure, and phase transition in the subduction zone based on the results of high-temperature and high-pressure experiments and temperature and flow simulations. Simulations of seismic wave propagation in the case of deep earthquakes using a two-dimensional difference method quantitatively demonstrated that the existence of a metastable phase can be detected as a coda wave of the body wave (29) (see Animation 5).