论文部分内容阅读
以希腊地震电信号观测为基础的一个连续三年地震预报实例已由Varotsos和Lazaridou(1991)发表。分别从4个方面研究分析了这个实例,得出地震预报的成功率远比偶然概率要高的结论。另一方面,Mulargia和Gasperini(1992)(以后记为MG)宣称这些预报可以归因于偶然性。在这篇文章中,我们考查了这一不同意见的起源。应当指出MG研究中有一些严重问题,如:(1)地震预报偶然成功的概率应当近似地看作3个概率P_T、P_E、P_M的乘积,它们分别是关于时间、震中、震级的概率。尽管P_E、P_M很重要,它们还是被MG忽略了。P_E的加入会降低偶然成功的概率多于10倍(当记入P_E时,可以看出VAN预报不能归因于偶然)。(2)MG通过对地震和预报选取不同的下限,过高估计了应当预报的地震数目。由于这种过高估计,MG甚至可能会否认一个理想的完美地震预报方法。(3)MG的程序方法没有考虑预报是以3种具有不同超前时间不同的地电前兆为基础的。(4)MG地震时间段序列采用的泊松分布,包括了大量的余震。(5)MG所说的预报和地震反向时间相关是由于预报内容的错误解释,同时不正确地使用了余震的结果。虽然只对第一个问题的讨论就足以使MG的说法无效,我们还是要讨论其他4个问题,因为单是在时间域内MG就违背了一些基本原理。本文得出的结果在考察地震与各种地球物理现象的相关关系是否超越随机性方面有着普遍的应用意义。
An example of a continuous three-year earthquake prediction based on the Greek earthquake electrical signal observations has been published by Varotsos and Lazaridou (1991). This example is studied from four aspects and the conclusion is drawn that the success rate of earthquake prediction is far more than the chance probability. On the other hand, Mulargia and Gasperini (1992) (hereinafter MG) claim that these forecasts can be attributed to contingencies. In this article, we examine the origins of this dissenting opinion. It should be pointed out that there are some serious problems in the study of MG, such as: (1) Probability of accidental success of earthquake prediction should be approximated as the product of three probabilities P_T, P_E and P_M, which are the probabilities of time, epicenter and magnitude, respectively. Although P_E, P_M are important, they are ignored by MG. The addition of P_E reduces the probability of accidental success by more than 10 times (when P_E is credited, it can be seen that VAN prediction can not be attributed to chance). (2) The MG overestimated the number of earthquakes that should be predicted by taking different lower bounds on earthquakes and forecasts. Due to this overestimation, MG may even deny an ideal perfect earthquake prediction method. (3) The method of MG does not consider the forecasting method based on three different precursor precursors with different lead times. (4) The Poisson distribution used in the sequence of MG earthquake time series includes a large number of aftershocks. (5) The correlation between the forecast by MG and the reverse time of the earthquake is due to the wrong interpretation of the forecast content, and incorrectly uses the aftershock result. Although the discussion of the first question alone is enough to invalidate MG’s assertion, we still have to discuss the other four issues, since MG runs counter to some of the rationale alone in the time domain. The results obtained in this paper have universal significance in examining whether the correlation between earthquakes and various geophysical phenomena transcends the randomness.