Abstract

A slow slip event has been progressing on the plate interface in the area near Lake Hamana, western boundary of the estimated focal region of the Tokai earthquake, since October 2000 (Ozawa et al., 2002). Occurrence of the slow slip was revealed by the gps data of the Geographical Survey Institute. The aim of this article is to reproduce the slow slip event in a simulation of plate subduction with a rate- and state-dependent friction law in which 3D configuration of plate interface is taken into consideration. By assigning a zone of large L value within the seismic region where ab is negative, we succeeded in producing slow slip events. Here, the ab value expresses the rate dependence of steady-state frictional strength, and the L value represents the characteristic slip distance over which the state evolves. They were found to happen five times during one cycle of interplate earthquakes. Fifth slow slip event finally brings about the catastrophic rupture. The moment release rate of the preslip is 10 to 100 times as large as that of the preceding slow slip events. On average a seismic moment equivalent to Mw 6.7 to 7.0 is released by slow slip events each year, while the moment magnitude of the earthquake is Mw 8.3. Duration of slow slip events is 13 to 15 years except for the last one that starts several years before the earthquake. The temporal change of volumetric strain produced by the preslip is significantly different from that of the preceding slow slip events. That is, the moment release rate accelerates in the preslip, while it is almost constant in the preceding slow slip events.

In the present model, slow slips occur in rather large areas. This may be caused by an assignment of a large L value in the seismic region. Both spatial inhomogeneity in L and curvature of the plate interface are considered indispensable to create appropriate inhomogeneous stress fields that generate slow slip events and, then, eventually a preslip and a great earthquake.

You do not currently have access to this article.