Shifting Correlation Analysis Of Earthquakes And Electromagnetic Signals

As a potential precursor of impending earthquake, the seismic electromagnetic signals (SEMS) have been reported by many researchers in China and elsewhere in the world. But there also have a lot of controversies on this subject. On one hand, because physical mechanism of the electromagnetic (EM) emissions associated with seismic activity is not clear. On the other hand, it is a fact that no precursors occur prior to some earthquakes, while some abnormal signals have no subsequent earthquakes. Although many methods have been developed to extract or recognize the SEMS from background EM fields, whether SEMS exist or not and the spatial distribution of SEMS are not clear. In other words, it is not confirmed whether the observed EM signals (including that before, during, and after earthquakes) are correlated to the earthquakes. Thus it needs to develop new methods to analyze such correlation if any.For a specific earthquake, SEMS may arise during the seismogenic process, rock rupturing in co-seismic moment, and seismogenic structure recovering in post-seismic time. That is, SEMS may or may not exist in pre-, co-, and/or post-seismic stages. Moreover, the SEMS of the above three stages may overlap each other if several earthquakes happen within a small time interval. And the SEMS possibly submerge into the background EM or noise fields. So, the relationship between earthquakes and EM signals and the spatial-temporal distribution of SEMS cannot be described exactly by the conventional methods based on single-earthquake cases.Earthquakes are caused by tectonic movement with similar dynamic processes. For all earthquakes or all earthquakes of a specific type, the corresponding SEMS may share similar temporal distribution characteristics. This means it is possible to choose multiple earthquakes for statistical analysis and extract the common features of SEMS. In view of the asynchronous related between earthquakes and EM signals, this work suggests a shifting correlation method (SCM) for recognizing the SEMS in pre, co-, and post-seismic stages simultaneously. The SCM for SEMS is to fix the earthquake or EM signal time-series on the time axis and then shifting one sequence relative the other, meanwhile, calculating correlation coefficients between the two sequences. The distribution of the maximum correlation coefficients (MCC) permits to analyze the time property of the SEMS. Using the MCC of multiple stations makes it possible to obtain the spatial distribution characteristics of SEMS.In this thesis, the basic principle of SCM is first introduced, and the validity of SCM has been verified by different kinds of synthetic data. The calculation results suggest that this method could not only recognize the single-related signals associate with the long sequence, but also recognize the asynchronous related signals which appear at different times and overlapped. In addition, this work also designs a synthetic data model according to the actual condition of SEMS that occurrence times of SEMS corresponding to different earthquakes are changed randomly within a certain time range. And the results show the related signals could be recognized clearly with enough samples. The calculation results of the synthetic data model indicate SCM has powerful recognition capability for the complex, hidden, and weak related signals.After the above verified, SCM has been applied to analysis of the EM monitoring data of the Minxian-Zhangxian earthquake and the continuous geoelectric monitoring data of Taiwan. This work attempts to investigate the characteristics of SEMS of different earthquake types and regions.Four magnetotelluric instruments were installed within 15km from the main shock epicenter after the Minxian-Zhangxian earthquake of 22 July 2013. Firstly, this work analyzed the changes of EM in pre-, co-, and post-seismic times by the conventional method preliminary based on the EM monitoring data. The results suggest that there have some abnormal phenomena that may relate to the earthquake in the EM spectrum variation series and apparent resistivity or phase variation diagrams. Then this work applied the SCM to the monitoring EM data and the main-after shocks. Firstly, the earthquake sequence is converted into equivalent magnitude (Meq), and EM sequence is processed as diurnal variations of power spectra at different frequencies. From the exploratory standpoint, this work supposed it is possible that SEMS could be radiated in pre-, co-, and post-seismic stages of a shock. Then, the shifting correlation coefficients are calculated by shifting the spectrum sequences of different frequencies left or right in time relative to the Meq starting from the co-seismic time, and examines the frequencies and shifting times of the strong correlation. The results show the SEMS may appear 23 days before the earthquake and disappear 5 days before the shock at the involved frequencies. There is remarkable correlation at relative low frequency in the stage of co-and post-seismic times on the ZJW station, and lasting for 10 days. It does not record the co-seismic signals at relatively high frequency at the two monitoring stations. In addition, comparing with the two stations, the results suggest the occurrence times of SEMS are different and the frequency corresponding to the remarkable correlation is different at different stages, even though the distance of two stations is only about 12km. It is speculated that it may be related to the discrepancy of deep electrical structures below the two stations.For further studying the temporal and possible spatial distribution of the SEMS, this work preforms SCM on the observation data of 20 geo-electric stations which are approximately of equal interval and cover the whole area of Taiwan Island. The shifting correlation coefficients are calculated between the diurnal variations of geoelectric signals of each station and the earthquakes within 50km from the corresponding station. Firstly, this work statistically investigates the occurrence times of the higher correlation coefficients in pre-seismic time of 20 stations, and finds that the most frequent higher correlation coefficients appear 23 days before earthquake. The result is in line with the case of the Minxian-Zhangxian earthquake. Then, the spatial correlation coefficient contour maps of pre-, co-, and post-seismic times are prepared respectively by taking the maximum values of shifting correlation coefficients of corresponding stages of each station. It is found that the contour maps have good consistence in different directions of geo-electric fields generally, but still some discrepancies existing between the X and Y directions. This may suggest the variation of geo-electric signals is directivity at local regions. Finally, the maximum values of the two directions of each station are taken again as the final correlation coefficients of the station, and the contours of maximum correlation coefficients (MCC) in pre-, co-, and post-seismic times are obtained, respectively. And the relationship between MCC and earthquakes or neotectonics distribution is analyzed. Three points have been shown in the results:(1) The biggest correlation coefficients are along the axis of Taiwan and the values on west and east are relatively smaller in pre-, and post-seismic times. The contours are in general parallel to the strike of neotectonics. In co-seismic stage, the coefficients are very small and the contours have no clear relation with the neotectonics. (2) The results suggest stations near the thrust fault have the bigger correlation coefficient values than the stations near the normal or strike-slip faults. (3) The correlation coefficients of stations on east flank of Taiwan which have more earthquakes are higher than the stations on west flank, but there are few sizable earthquakes close to the stations at which coefficients are the biggest among the 20 stations. According to the above phenomena and dynamic background of the Taiwan orogenic, it is speculated that thust temblors are likely more capable stimulate by SEMS than strike-slip and normal faulting events.In summary, this thesis presents some prevail research methods for the relation between SEMS and earthquakes firstly. The changes of EM fields and deep electrical parameters during the 2013 Minxian-Zhangxian earthquake are studied. Especially, this work proposes a new method called shifting correlation method (SCM) which can investigate the temporal and spatial characteristics of SEMS simultaneously. And then SCM is applied to monitoring data of the Minxian-Zhangxian earthquake and geoelectric observation data in Taiwan Island. The new method proposed shows its powerful capacity for studying the features of SEMS. If it is combined with the available methods extracting or recognizing of abnormal signals, it is possible to recognize the characteristics of spatial and temporal distribution of SEMS more clearly in future.