Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article


Master of Science




Goda, Katsu

2nd Supervisor

Mora-Stock, Cindy



In this thesis, a new segmented earthquake occurrence simulation method is developed and is applied to the Nankai-Tonankai megathrust earthquakes, which have historically caused significant damage to the coastal areas of Japan and have a large potential to occur in the next 50 years. For targeting the segmented rupture pattern in Nankai-Tonankai Trough, the simulation method incorporates the multivariate Bernoulli method, which models the rupture pattern by considering both spatial and temporal correlations of individual segments, and the stochastic slip modelling method that produces heterogeneous slip models corresponding to different rupture patterns. Furthermore, possible predominant factors that could affect the earthquake simulation, including the earthquake occurrence catalogs, interarrival time distributions and segmentation pattern, are examined through varying simulation settings. According to its application in seismic hazard assessment, the multivariate-stochastic hybrid method successfully captures the historical seismicity of the Nankai-Tonankai Trough, while emphasizing additional seismic features depending on the predominant factors.

Summary for Lay Audience

The Nankai-Tonankai megathrust in Japan hosted many large earthquakes with magnitudes greater than 8. Historically, the Nankai-Tonankai megathrust earthquakes have a recurrence interval of 100-150 years. Considering the 75-year elapsed time since the last earthquakes in 1944 and 1946, the Nankai-Tonankai megathrust has an immense potential to host a giant earthquake in the next 50 years. Therefore, the development of credible earthquake occurrence simulation methods is essential for seismic hazard assessment and disaster preparedness.

The rupture pattern of the Nankai-Tonankai Trough varies from individual-segment ruptures to synchronized multiple-segment ruptures. In this thesis, a multivariate Bernoulli modelling method is adopted for earthquake occurrence simulation. The rupture conditions in individual segments are simulated by assuming that the interarrival time follows a probabilistic model, such as Brownian Time Passage distribution, and the multiple-segment rupture scenarios are characterized by the spatial correlation between adjacent segments. Compared with existing probabilistic models, the adopted method captures both spatial and temporal dependency of earthquake ruptures along the segmented fault. Moreover, to generate realistic earthquake slip distributions, the stochastic slip modelling method is combined with the multivariate Bernoulli method. The multivariate-stochastic hybrid method thus can simulate earthquake ruptures by considering their space- and time-interaction, and link the rupture patterns of each simulation with the heterogeneous slip model.

For the Nankai-Tonankai Trough, the integrated model simulates earthquake occurrence with interarrival times and rupture patterns that are consistent with the historical seismicity of the fault segments. Additionally, all simulated rupture histories have realistic earthquake magnitudes, asperity areas, hypocenter locations, and slip values similar to the historical megathrust earthquakes. By applying this new earthquake rupture model to seismic hazard assessment, the anticipated regions with intense ground motions can be identified. The high-risk regions correspond to coastal areas near the middle segments of the Nankai-Tonankai Trough, where the effects due to the synchronized ruptures are more likely. Overall, the hybrid earthquake simulation method provides a consistent framework for carrying out earthquake occurrence simulation. Its application in seismic hazard assessment also allows carrying the ground motion estimation in a probabilistically time- and space-dependent way.