欢迎访问《地震》，
地震

地震 ›› 1999, Vol. 19 ›› Issue (2): 135-141.

• • 上一篇    下一篇

概率方法应用于地震短期预测的探索①

张天中, 王林瑛, 刘庆芳, 丁秋琴   

  1. 中国北京 100081 中国地震局地球物理研究所
  • 收稿日期:1998-05-25 修回日期:1998-07-20 出版日期:1999-04-30 发布日期:2022-09-19
  • 作者简介:张天中, 男, 1945年 12月出生, 副研究员,主要从事地震波衰减、工程地震和地震预测等研究工作。
  • 基金资助:
    ①中国地震局“九五”科研攻关项目( 95-04-05-02-05)部分成果; 中国地震局地球物理研究所论著( 98 A02070)

AN APPLICATION OF PROBABILITY METHOD TO THE SHORT-TERM EARTHQUAKE PREDICTION

Zhang Tianzhong, Wang Linying, Liu Qingfang, Ding Qiuqin   

  1. Institute of Geophysics, CSB, Beijing 100081, China
  • Received:1998-05-25 Revised:1998-07-20 Online:1999-04-30 Published:2022-09-19

PDF (PC)

摘要： 根据爆发地震和地震平静两项活动性前兆的统计结果,对地震发生的背景概率、条件概率和概率增益进行了估计,给出了张北地震发生前发震概率逐步增加的时间过程,由背景概率P ( E) 增至中期前兆 A(爆发地震 ) 出现后的条件概率 P ( E|A),再增至短期前兆 B (平静 ) 出现后的联合条件概率 P ( E|A, B )。结果表明,在 1997年 12月 17日后的一个月内,华北地区发生 6级以上的条件概率达 38%,概率增益超过 20。对概率预测结果的使用进行了探讨,提出应充分注意发震概率的大小与地震发生的时空强范围之间的关系。对于短期预测,要求估算出很高的发震概率目前是很困难的,重要的是要有足够的概率增益。对于各种前兆和预测方法进行较为严格的统计检验是概率预测的基础,也是使地震预测健康发展的基础。对于各种实用的地震预测方法宜采取跟踪检验的方法,在提供预测意见同时,有条件的方法均应提供最新的统计检验结果。

关键词: 概率预测, 条件概率, 概率增益, 贝叶斯方法

Abstract: Based on the statistical result of two seismicity precursors of blast earthquake and seismic quiescence, the background probability, conditional probability and probability gain of earthquake occurrence are estimated. The time history of probability of earthquake occurrence before Zhangbei earthquake is introduced. The background probability P (E) increased a t first to the conditional probability P (E|A) after appearance of intermediate-term precursor A (blast earthquake ), then to the combined conditional probability P ( E|A, B ) after appearance of the short-term precursor B ( Seismic quiescence). Within a month after December 17, 1997, the combined conditional probability of earthquake occur-rence with magnitude greater than 6 in Northchina was 38% with probability gain over 20. How to use the result of probability prediction is discussed in this pa per. It is noticed that the probability of earthquake occurrence strongly depends on the time, space and magnitude spa n to be predicted. A high probability for a short-term prediction is hard to beobtained at present, and a high probability gain is important. The hypotheses test for various precursors and prediction methods is essential for a sound development of earth-quake prediction research. For the practical prediction methods the trace test is suggested. While a prediction is made by a certain metho d, the latest result of hypo theses test for the method is suggested to be provided if possible.

Key words: Probability prediction, Conditional probability, Probability gain, Bayes method