Finite-Difference Simulation of Long-Period Ground Motion for the Sagami Trough Megathrust Earthquakes

We perform long-period ground motion simulations for Sagami Trough earthquakes by a three dimensional finite-difference method. The Sagami Trough has been the site of two well-known megathrust earthquakes, the 1923 Taisho- and the 1703 Genroku-type Kanto earthquakes. However, a lack of accumulated historical earthquake records prevents us from obtaining knowledge of the source model of the next anticipated event for long-period ground motion hazard evaluation. Therefore, it is important to consider numerous possibilities for the unknown source parameters. We compare ground motions for several scenarios with different source area, and with magnitudes ranging from Mw7.9 to 8.6. Peak ground velocity (PGV) within the Kanto basin, including the Tokyo metropolitan area, differs by several times depending on the choice of the source area. The effects of the variety in fault parameters, such as rupture starting points and asperity patterns, are also studied. They can greatly vary the ground motion within the Kanto area, especially in the direction of rupture propagation, suggesting the severe impact of directivity effects. Source models with different rupture starting points produce PGV and 5% damped velocity response (Sv) that vary from each other by as much as 10-20 times. PGV and Sv vary by up to five times depending on the asperity pattern. Our simulation results show that the predicted ground motion for the earthquake scenarios strongly depends on both the source size and other fault parameters of the source models. It is suggested that the seismic hazard assessment requires statistical evaluation of ground motions from as many source models as possible in order to overcome the uncertainties of the source.