- 公益社団法人 日本地震学会
- 地震 第2輯 (ISSN:00371114)
- vol.49, no.2, pp.141-168, 1996-08-23 (Released:2010-03-11)
96 photographs of seismograms from the 1923 Kanto earthquake (M=7.9) have been preserved at JMA (the Japan Meteorological Agency). These seismograms were obtained at meteorological observatories in and around Japan and Tohoku Imperial University in Sendai, whose epicentral distances are within about 2000km. P-wave forms of the main shock are recorded at 37 observation stations, though larger shaking parts of the ground motions are saturated on the records at almost all the stations. First, a type of seismograph and its instrumental characteristics are examined for each record from old materials and old records preserved at JMA. It is found that the seismograms at 29 stations were recorded by Portable Seismographs with natural period T of about 5sec and magnification V of 10 to 100, and most of the others by Omori's Seismographs with T of 10 to 30sec and V of 10 to 20 and by Wiechert Seismographs with T of about 5sec and V of about 100. Secondly, P-wave characteristics of each record are examined. We reexamined polarities of initial P-wave motions and reviewed focal mechanisms obtained for the Kanto earthquake. We also found that P-wave forms at many stations have some characteristics in common. The P-waves begin with small amplitude and the first remarkable phase with a large amplitude appears after the duration of the initial small phases. The second remarkable phase is also identified several seconds after the first remarkable phase, irrespective of the epicentral distances of the stations. The duration time τ01 of the initial small amplitudes correlates with the epicentral distance at each station. It can be explained by the difference of calculated arrival times between Pn and Pg phases or between Pn and P* phases in addition to a duration of the initial small rupture near the hypocenter. The duration of the initial small rupture is evaluated about 4.5sec from τ01's at observation stations within the epicentral distance of 170km. In the meantime, the difference τ12 of arrival times between the first and the second remarkable phases correlates with the azimuth angle of each station from the epicenter. Geometry and difference of occurrence times of big subevents are evaluated on the fault plane of the main shock from the data of τ12's, if we assume the two remarkable phases are due to two big subevents, respectively. The second subevent is located about 40km distant from the first subevent in the direction of about N 100°E, and the time interval of the occurrences of the two subevents is about 14sec.