- 著者
-
林 豊
前田 憲二
- 出版者
- Japanese Society for Active Fault Studies
- 雑誌
- 活断層研究 (ISSN:09181024)
- 巻号頁・発行日
- vol.2009, no.30, pp.27-36, 2009
Six active fault zones have been selected on the basis of the reports on the long-term evaluation of active faults published until 2008 by the Earthquake Research Committee, Headquarters of Earthquake Research Promotion (ERC/HERP); the paleoseismic activity data of these zones reveal three or more earthquake recurrence intervals. Using the maximum likelihood method, seven probability density functions of a renewal process model are compared in order to determine the function that best fit the paleoseismic activity data of these active fault zones.<br>The exponential distribution model obtained by using the maximum likelihood method does not clearly reveal the earthquakes recurrence intervals. In contrast, the results obtained by using six other statistical models, i.e., Brownian passage time (BPT) distribution, lognormal distribution, gamma distribution, Weibull distribution, double-exponential distribution, and normal distribution, reveal the earthquake recurrence intervals. Thus, the new paleoseismic activity data of major active zones in Japan confirm the provisional conclusion of ERC/HERP, i.e., the exponential distribution does not clearly show the earthquake recurrence intervals. On the other hand, differences among the goodness of fit of the six models excluding the exponential distribution are small.<br>In 2001, ERC/HERP stated that when renewal process model with the BPT distribution is applied to the data of the occurrence intervals of earthquakes in the inland active fault zones in Japan, the aperiodicity parameter of the distribution should be set to 0.24 as a value common to all active faults. The aperiodicity parameter obtained by applying the same method to the data of the six active fault zones is equal to 0.44. Although the aperiodicity parameters, obtained by using the maximum likelihood method, reported in the ERC/HERP's report range between 0.17 and 0.29, those obtained in this study range between 0.09 and 0.66. Thus it is inappropriate to assume the same aperiodicity parameter for all the inland active fault zones in Japan.