1999 年 52 巻 2 号 p. 317-333
Nozawa et al. (1995) proposed a source model with two big subevents of the same seismic moment for the 1923 Kanto earthquake (M=7.9), through the simulation of the records by the Imamura-type strong motion seismograph (displacementmeter) at Gifu observatory. This model was named Model I in the present study. The first subevent of Model I is located under the Odawara city, having a fault plane with the strike of N290°E and the rake angle of 162°. This fault has much strike slip component, which is consistent with the focal mechanism solution by KANAMORI (1971). However, the direction of the strike is not compatible with the trench axis of the Sagami trough. The second subevent occurring 12s after the first subevent is located under the Miura Peninsula. The fault of the second subevent, having much dip slip component, well explains the geodetic data. Recently, the seismograms by the Imamura-type strong motion seismographs at Sendai (Mukaiyama) observatory and Yamagata observatory were examined and the instrumental responses of the seismographs were revealed. Crustal structure from source to stations was estimated in the present study so as to explain the observed Love and Rayleigh waves at Sendai (JMA) and Yamagata observatories from the recent events occurring near the focal region of the 1923 Kanto earthquake. However, Model I failed to explain the records of the 1923 Kanto earthquake at Sendai (Mukaiyama) and Yamagata observatories, using the obtained crustal structure. Then, we revised Model I to explain these records, in consideration of the newly determined focal mechanism solution by Lallemant et al.. (1996) and iso-depth contour of the upper boundary of the Philippine Sea plate by Ishida (1992). The first subevent of the revised model (Model R) has a fault plane with the strike of N321°E and the rake angle of 128°, and the twice of seismic moment of the second subevent. The direction of the fault strike of the first subevent is parallel to the trench axis of the Sagami trough, while the fault plane of the second subevent is the same as Model I. Model R succeeded in explaining not only the records at Sendai (Mukaiyama) and Yamagata observatories but also those at Gifu observatory in the period range from 2 to 20s. This shows the fault model, being in agreement with the geometry of subduction zone along the Sagami trough, is better to explain the seismic records observed in Japan.