Integration of geological and seismological data for the analysis of seismic hazard: A case study of Japan

Seismic hazard analyses are associated with large uncertainties when historical data are insufficient to define secular rates of seismicity. Such uncertainties may be decreased with geological data in areas where seismicity is shallow and produced by Quaternary faulting. To illustrate, we examine intraplate Japan. Large intraplate earthquakes in Japan characteristically produce surface ruptures along mappable Quaternary faults and show a systematic relation between seismic moment, M_0 and rupture length I (log M_0 = 23.5 + 1.94 × log I). It is observed that, within the bounds placed by geologically assessed slip rates, the mean regional moment release rate M_0 resulting from slip on mapped Quaternary faults is in accord with estimates of M_0 determined with the 400-yr record of seismicity. Recent work also shows that when the repeat time T of earthquakes on Quaternary faults in southwest Japan is assumed to equal M_0/M_0^g (where M_0 is estimated for rupture extended over the entire fault length and M_0^g is the geologically assessed moment release rate of each fault), the moment frequency distribution of earthquakes predicted from the geologic record is virtually identical to that seen with the 400-yr record of seismicity. These observations indicate that the geologic record of Quaternary fault offsets contains sufficient information to predict both the spatial and size distribution of intraplate earthquakes in Japan. A contour map of the average recurrence time of ground shaking of JMA intensity ≧V is thus computed using an empirical relation between seismic moment and the areal distribution of seismic intensity and assuming that the repeat time T of earthquakes on each Quaternary fault equals M_0/M_0^g. The map demonstrates how Quaternary fault data may be used to assess long-term seismic hazard in areas of active faulting where historical records of seismicity are relatively short or absent. Another shortcoming of conventional seismic hazard analysis is that hazard is not considered a function of the time since each fault in a region last ruptured. A simple procedure is used to demonstrate how the time-dependent nature of the earthquake cycle affects the evaluation of seismic hazard. The distribution of seismic shaking characteristic of large interplate earthquakes offshore of Japan is estimated from published isoseismal maps. The observed average repeat times of ruptures along specific segments of the plate boundaries then provide the basis to make probabilistic estimates of the next expected time of seismic shaking due to plate boundary earthquakes. When data are too few to document the average repeat times of rupture, the estimates of probability are calculated with data relating to the relative coseismic slip during past earthquakes and the rate of interseismic strain accumulation, interpreted within the framework of the time predictable model of earthquake occurrence. Results are displayed as maps of instantaneous seismic hazard: the probability that seismic shaking will occur conditional to knowledge of where in time each fault in a region presently resides with respect to the earthquake cycle