前兆現象の経験則に基づく実用的地震予知 (2)
地震発生時期
1987 年 40 巻 4 号 p. 605-617
詳細
1987 年 40 巻 4 号 p. 605-617
Analysis of precursor time-main shock magnitude relationship and/or frequency distribution of precursor time makes it possible to evaluate the probability of earthquake occurrence within a specified period of time provided a good many data of earthquake precursor of a particular discipline are available. The approach is purely empirical basing on the nature of precursor data so far obtained in Japan.
The probabilities thus estimated are different from discipline to discipline of precursor. For a magnitude 7 earthquake, for instance, geodetic and geomagnetic signals provide a precursor of long term, the peak of probability maximum for earthquake occurrence within 1 day occurring several hundred or thousand days after the signal appearance. Meanwhile, the probability peaks for radon and other geochemical elements, underground water/hot spring and the like occurs much earlier forming a group of precursors of short term.
When a number of precursors are observed one by one, changes in the synthetic probability of earthquake occurrence can be estimated according to the existing formula [UTSU (1977)]. As an example, such probability changes with time are estimated making use of the precursor data for the 1978 Izu-Oshima Kinkai earthquake of magnitude 7.
It is interesting to see that the synthetic probability generally increases as time goes on. However, the probability discontinuously lowers when a long-term signal happens to appear after short-term signals. When the period for probability estimate is taken shorter, the probability curve fluctuates considerably making the prediction of occurrence . time difficult. Such difficulties seem to be avoided for a medium-term prediction for which a period of having the main shock within 100 days or so is assumed.
One of the most difficult problems in the probability estimate is the point that we do not know whether or not a signal really represents a precursory one. The rate of false signals against the total signals has been estimated only for land uplift and foreshock precursors. In the circumstances, the said rate for other precursor disciplines is arbitrarily assumed as 1/20 in the estimate in this paper. Such a value should be improved in the future with the accumulation of precursor data.
