Abstract: In recent years, more and more attention has been paid to the anomaly of annual well level variation and its implication to mid-term and short-term earthquake prediction. However, the qualitative comparison of dynamic pattern of the anomaly of annual well level variation displays an obvious uncertainty. To solve this problem, the random process and time series analysis techniques from probability theory and mathematical statistics are used to establish "the method of dynamic annual period" and "the method of comparative rate at comparative periods" for quantitative analysis of the anomalies of annual dynamic variation of well water levels. These methods are applied to the analysis of the observation data since 1985 till now from 21 wells of Tianjin well networks. The result has proved the feasibility and effectiveness of these methods for earthquake prediction. The analysis results demonstrate that the anomalies of well level occurred 17 times in 7 wells of Tianjin network one months and a half to six months before the occurrence of 4 moderate-strong (M_S≥5.8) earthquakes in this region. This fact may indicate again the applicability of the anomaly of annual well level variation to mid-term and short-term earthquake prediction.

Key words: dynamic annual period, comparative period velocity, anomaly of annual water level variation, mid-term and short-term prediction

摘要： 近年来,中国地震预报学者开始关注井水位的年变异常及其中短期预测意义问题,但井水位年变异常判别采用动态图像的定性对比方法,表现出一定的随意性。针对这种现状,文中引进概率论与数理统计中的随机过程理论与时间序列分析技术(盛骤等,1989),提出了“井水位动态年周期法”与“相对时段速率比较法”,解决了井水位年变异常的定量识别方法,并应用到天津井网21口井的观测数据(1985年以来)分析中,证明了该方法的可行性与有效性。分析结果表明,天津井网中有7口井在首都圈邻区4次中强(M_S≥5.8)地震之前,表现出17井次的井水位年变异常,且多在震前1.5～6个月内出现,从而再次证明了井水位年变异常具有一定的中短期预测意义。

关键词: 动态年周期, 相对时段速率, 水位年变异常, 中短期预测