Detecting Earthquake-Related Borehole Strain Data Anomalies With Variational Mode Decomposition and Principal Component Analysis: A Case Study of the Wenchuan Earthquake

Borehole strain monitoring has high sensitivity and is therefore widely used to study slow earthquakes, volcanic activity, earthquake precursors, and other nature phenomena. However, environmental factors seriously affect the identification of strain changes caused by crustal deformation. This paper proposes a method of anomaly detection based on variational mode decomposition (VMD) and principal component analysis (PCA). The borehole strain signal is decomposed into a number of modes simultaneously using VMD, and a new state-space model used to determine the number of the modes those are decomposed by the VMD algorithm. The influencing factors of each component are determined by spectrum analysis and comparative analysis. An example of the separation process of borehole strain data by the VMD method is presented. Then, we use PCA to calculate eigenvalues, which are used to detect anomalies associated with an earthquake, and eigenvectors, which are applied to show the spatial distribution characteristics of the data. Our method has been applied to detect borehole strain data anomalies associated with the Wenchuan earthquake; the VMD demonstrates excellent separation performance for borehole strain signals, and eigenvalues and eigenvectors together reflect the accelerated deformation of focal faults and adjacent areas before earthquake in time and space.