Numerical Investigation of Controls on Megathrust Earthquakes Along the Japan Trench Subduction Zone

The goal of this dissertation is to investigate physical conditions that control the generation of megathrust earthquakes, such as the 2011 Mw 9.0 Tohoku earthquake, along the Japan Trench subduction zone. Understanding the physical conditions that lead to megathrust earthquakes is of great importance to earthquake prediction and seismic hazard mitigation. In order to fulfill this goad, I establish an advanced elastodynamic modeling algorithm based on dynamic finite element method for numerical investigation of dynamic earthquake rupture and earthquake cycle behavior. I first implement nonplanar fault geometry and various forms of the laboratory-derived rate- and state-dependent friction laws in the framework of the dynamic finite element method. Using the updated method, I explore how dynamic ruptures would behave under the influence of different friction laws on a fault surface with a bump representing a subducted oceanic relief. The results show that the bump could act as a rupture barrier, and such a geometrical effect varies with the dimension of the bump and with the specific forms of friction law. I then extend the dynamic modeling method to an integrated earthquake simulator by using the adaptive dynamic relaxation technique. The new earthquake simulator is capable of capturing both long-term and short-term faulting behaviors in multiple earthquake cycles. Earthquake cycle simulations of thrust faults with various dip angles show that thrust faults tend to produce earthquake cycles with a longer recurrence interval and larger released seismic energy compared with strike-slip faults. Moreover, I find that the asymmetry in particle displacements across a thrust fault caused by earthquakes can be recovered during the interseismic phase. Finally, I conduct numerical investigation on earthquake cycle behavior in subduction fault models of the Japan Trench subduction zone. I find that a planar fault model with realistic structural heterogeneity reproduces complex faulting behavior including numerous aseismic transients in interseismic period and a megathrust earthquake that resembles the 2011 event, while a homogeneous model with complex fault geometry of a low-height, broad-base seafloor relief accumulates stress changes slowly on the geometrical irregularity over time but generates simple pattern for earthquake cycles within the limited simulation time