應用超低頻電磁波遙測技術於地震前兆分析—以台灣為例

Abstract

歷年來台灣曾多次遭受地震所帶來的災害，像芮氏規模7.3的1999年的921大地震造成2,415死亡及11,305受傷，進而造成土壤液化引發山崩及土流等災害，然而目前在地震預測及地震預警的工作仍有許多困難需要克服，相較於其他天然災害而言，地震預測的困難度較高，卻是急需克服的災害種類。 在學術上，目前常用地震預報的方法有電磁波訊號異常、Demeter微型衛星、動物集體行為異常現象、水氡異常以及水溫異常等，雖然監測的方法很多，然而在較少應用於實際層面之上。然而，不論是何種監測手法，目前大多數的文獻皆是做案例分析，且皆為事後分析地震的前兆特徵，進而歸納並推論地震來臨前可能的前兆特徵或是前兆訊號及頻率，在推估未來的部分較少著墨。 本研究利用現在已有進行預報的MDCB超低頻電磁波監測儀進行訊號分析工作，由於在時域空間的判釋較為困難且雜訊較多以至於難以辨別。因此本研究利用快速傅立葉轉換將資料由時域空間轉為頻率空間，並結合卷積及低通濾波器將雜訊濾除。在頻率空間的分析主要結合特徵分群的前兆訊號分析及歷史震央空間分析資料進行空間機率環域交匯分析震源深度機率分析。除此之外，利用前兆訊號數值與歷史地震資訊進行回歸分析，欲推估未來地震規模及震中距相關性，最後歸納前兆訊號特徵來推估未來地震可能發生時間。 根據以上所述的空間機率環域交匯分析、震源深度機率分析、地震規模及震中距迴歸分析及地震發生時間歸納分析進行地震前兆分析探究，希望以此分析系統增加現今地震前兆分析及地震預測的準確度，以減少災害所帶來的人才損失。

Throughout worldwide, earthquakes have deprived lots of life and property. However, earthquakes cannot be predicted precisely in terms of the epicenter, time, seismic scale, and depth with various means. Most of errors were caused by misleading signal processing. The goal of this study provide a better signal processing method to depict the potential zone of seismic epicenter. In order to realize the significant signal and frequency, this study utilizes Fast Fourier Transform (FFT) to analyze Ultra Low Frequency (ULF) signals and define a warning line for dividing normal and abnormal signals. In this study, an epicenter location can be inferred by intersection of at least three abnormal angles from different stations. In addition, epicenter estimation analysis imports probability buffer concept in spatial cross analysis, moreover, this concept also applies in depth estimation. Break time estimation concludes both lots of papers information and abnormal signal pattern, so this study define that break time of earthquake is one week after abnormal signal appearance. For magnitude regression, this study utilizes three different parameter, MAEQ, MMEQ and IAEQ, to regress the correlation with Richter magnitude scale. Up to day, this study has successfully found significant signal of earthquake precursors and also calculated the potential zone of seismic epicenter, break time, depth potential and magnitude beforehand. In conclusion, this research provides a new method for epicenter prediction by analyzing ULF electromagnetic signals.