论文部分内容阅读
通过研究电磁现象与大地震的联系有助于建立监测地壳运动(如地震、火山喷发)的电磁方法。在地震的紧急预报中,超低频(ULF)地磁的方法是最有前途的方法之一。本文将一种主要分量分析的新方法应用在了与2000年日本伊豆诸岛震群有关的ULF地磁资料的研究中。这次震群活动始于2000年6月26日,同年9月结束。在这次地震活动中,观测到5次震级超过6级的大地震(分别发生在7月1日、8日、15日、30日和8月18日)。在此期间,我们的超低频地磁观测站幸运地记录下了数据,其中有三个观测站相距很近(大约5 km的距离),距地震中心大约80~100 km。因为我们可以在正交分解的基础上,用主要分量分析法将一些噪声分辨出来,所以我们把主要分量分析法用在了超低频地磁水平方向的NS分量中。我们研究了主要分量分析法结果中本征值和本征矢量的临时变化,这些结果总结如下。第一主要分量是源于日地交互作用影响的信号,如:地磁脉动。第一主要分量的本征矢量的变化说明这种信号在整个分析过程中非常稳定。第二主要分量中混有当地的人为的噪音。而最小的第三分量出现在伊豆半岛当地的子夜,在大地震前几天,第三本征值明显地增加。大约在震群活动开始前三个月,第三本征值的水平有微小增强。相应地,在信号的子空间中本征矢量方向的模式同时改变,在震群活动后又恢复到了其原来的水平。这些特征可能与大地震有关。最后我们强调主要分量分析的方法是目前监测地壳活动最有前途的方法。
By studying the connection between electromagnetic phenomena and major earthquakes, it is helpful to establish an electromagnetic method to monitor crustal movement (such as earthquakes and volcanic eruptions). In the emergency prediction of earthquakes, ultra-low frequency (ULF) geomagnetic methods are one of the most promising methods. In this paper, a new method of principal component analysis is applied to the study of ULF geomagnetic data related to the earthquake swarm in Izu Islands, Japan in 2000. The earthquake swarm activity started on June 26, 2000, and ended in September the same year. During this earthquake, five major earthquakes of magnitude 6 were observed (July 1, 8, 15, 30 and August 18, respectively). During this period, our ultra-low frequency geomagnetic observatories fortunately recorded the data, of which three were close (about 5 km) and about 80-100 km from the seismic center. Because we can distinguish some noises by principal component analysis on the basis of orthogonal decomposition, we use the principal component analysis in the NS component of the ultra-low frequency geomagnetic horizontal direction. We study the temporary changes in eigenvalues and eigenvectors in the results of the principal component analysis. These results are summarized below. The first major component is the signal that is derived from the interaction of the earth and the earth, such as geomagnetic pulsations. The variation of the eigenvector of the first principal component shows that this signal is very stable throughout the analysis. The second main component is mixed with local artifacts. The third smallest component appeared on the midnight of the Izu Peninsula. A few days before the earthquake, the third eigenvalue increased significantly. Approximately three months before the start of the earthquake swarm activity, there was a slight increase in the level of the third eigenvalue. Correspondingly, the patterns of the eigenvectors in the subspace of the signal change simultaneously and return to their original levels after the swarm was activated. These features may be related to large earthquakes. Finally, we emphasize that the method of principal component analysis is the most promising method of monitoring crustal activity.