1995 年 48 巻 3 号 p. 331-340
Records observed at Gifu observatory by an Imamura's type strong motion seismograph are one of the most useful records in Japan to investigate a source process of the 1923 Kanto earthquake (M=7.9). It is because amplitudes of the records are not saturated in EW and UD components, instrumental response of the seismograph has been clarified, and many records due to recent events occurred near the focal region of the Kanto earthquake have been obtained by more accurate seismographs at the same site. In the present study, a source process of the 1923 Kanto earthquake is elucidated through a simulation of the records using the normal mode theory in the period range from 2 to 20s. First, a crustal structure from the source to the station is estimated so as to explain dispersive characteristics of Love waves observed at Gifu observatory for the recent events, and their records are simulated to confirm a validity of the estimated crustal structure. Secondly, the records from the Kanto earthquake are simulated using the obtained crustal structure to deduce the source process of this event. According to KANAMORI (1971), a macroscopic faulting is a reverse right-lateral fault on a plane dipping 34° towards N20°E, whose slip has much strike component. If two big subevents with the same focal mechanism obtained by KANAMORI (1971) and with a time interval of about 12s are assumed on the fault plane, the observed records can be well explained. The first subevent is located under the Odawara city and the second one under the Miura Peninsula. The focal depth of the second event is 15 to 35km being deeper than that of the first event, which is 5 to 25km in depth. The seismic moments and the rise time are assumed 2.5×1027 dyne-cm and 5s for both the events respectively. On the other hand, if the focal mechanism of the second event is dip slip type, the observed records can be also explained well, even though the focal depth of the second subevent is the same as that of the first one. This model is consistent with a slip distribution on the fault plane obtained from geodetic data.