抄録
For a prediction of magnitude of future earthquake on an active fault, it is important to estimate what portion of the fault will move in a single earthquake. Using a data set of 36 historical inland earthquakes which can be correlated to previously mapped active faults, we estimate the ratio of rupture length of historical events to the length of “seismogenic faults” proposed by MATSUDA (1990) (Matsuda model). We found that the ratio is distributed from 1/4 to 1, indicating that there are a significant number of events in which only portions of “seismogenic faults” of MATSUDA (1990) moved, although these “seismogenic faults” were thought to rupture in a single earthquake. Among thirty Matsuda model faults examined in this study, fifteen can be divided into segmented faults according to KUMAMOTO (1998) who proposed each segmented fault ruptures in a single earthquake (Kumamoto model). The ratio of rupture length to the length of Matsuda model fault for those fifteen faults is also distributed from 1/4 to 1 and this cannot be explained by Kumamoto model. We propose a new fault model which includes Matsuda model, Kumamoto model, and groups of faults composed of neighboring fault segments. We assumed equal probability for earthquake occurrence on all the modeled faults. A numerical simulation for reproducing nineteen historical earthquakes correlated with the fifteen Matsuda model faults which are divided into segments by KUMAMOTO (1998), shows that our proposed model best explains the observed distribution of fault length ratio of historical earthquake data.
