Analysis Of Borehole Strain Precursory Observation Data And Research On Anomaly Extraction

China is one of the countries with strong earthquake activity and serious earthquake disaster in the world,especially in the earthquake prone areas,earthquake disaster always threatens the safety of people's life and property.Due to the complexity of earthquake preparation mechanism,earthquake prediction is still a worldwide scientific problem that needs to be solved through long-term exploration.China has established a ground precursory observation system with the subjects of gravity,crustal deformation,geomagnetism,geoelectricity and underground fluid as the observation objects,which has provided rich observation data for the study of earthquake precursory in China.It has great significance for earthquake prediction and prediction to find a suitable and effective method of anomaly identification and extraction before earthquake.Crustal deformation observation is one of the most important precursory observations.With the advantages of high precision and wide frequency band,borehole strain observation can accurately measure the crustal strain change.Therefore,borehole strain observation data analysis has become an important means of earthquake precursor research.However,there are some problems to be solved in the analysis and research of borehole strain observation data: Although the borehole strain gauge is deeply buried in the ground,the borehole strain observation data will still be affected by the factors such as strain solid tide,air temperature,borehole water level,etc.,which will interfere with the analysis and anomaly extraction of borehole strain data;the abnormal shape and distribution of borehole strain data before different earthquakes are different In the same way,conventional methods are difficult to identify and extract such anomalies effectively;In the correlation analysis between borehole strain data anomalies and earthquakes,more favorable evidence is still needed.In view of these problems,this paper first analyzes the influence of strain tide,air temperature,air pressure and water level on borehole strain observation,uses the method of Minimum Noise Fraction and Variational Mode Decomposition to remove the influencing factors,uses the method of Hilbert-Huang Transform and Principal Component calculation to extract borehole strain data anomalies,and the correlation between the extracted anomalies and earthquakes is analyzed by the criteria of earthquake precursor,random time comparison and multi-station comparison.The main contents of this paper are organized as follows:(1)Based on the principle of borehole strain observation,self-consistent equation is used to verify the validity of the data.The characteristics of strain solid tide,air temperature,air pressure and borehole water level are analyzed in detail.The influence of these factors on borehole strain observation data are analyzed through examples.(2)Using the principle of signal and noise interchange,a method based on Minimum Noise Fraction is proposed to remove the strain solid tide in the borehole strain data.By using Minimum Noise Separation,through the signal and noise interaction,the borehole strain data is decomposed into Minimum Noise Separation components arranged according to the signal-to-noise ratio,and the impact of the strain solid tide components are concentrated on the low-order components.Based on the frequency statistical characteristics of the strain solid tide,Minimum Noise Separation components corresponding to the strain solid tide are judged and removed,and the remaining Minimum Noise Separation components are reconstructed to remove the strain solid tide,through the simulation experiment and the earthquake example analysis,this method can effectively remove the strain solid tide in the borehole strain data,and can well retain the anomaly morphology.(3)A method based on Hilbert-Huang transform is proposed to extract instantaneous energy anomaly from borehole strain data.Statistical analysis of the frequency characteristics of air temperature,air pressure and borehole water level is carried out to establish the frequency(period)statistical table of the influencing factors;the borehole strain data is decomposed into multiple modes by the method of Hilbert Huang transform,and the corresponding modes are determined and removed by frequency comparison;the instantaneous energy of the remaining modes is calculated,and the instantaneous energy anomaly is extracted by selecting the appropriate threshold value,and the means of anomaly accumulation is used to analyze the correlation between instantaneous energy anomaly and earthquake.The case study of Lushan earthquake shows that this method can effectively remove the effects of strain solid tide,air temperature,air pressure and borehole water level,and extract the pre-earthquake anomalies of Lushan earthquake and Wenchuan earthquake.(4)Based on the Variational Mode Decomposition,an anomaly extraction method in the principal component domain of borehole strain data is proposed.The state space model of borehole strain observation data is established,and the number of decomposition layers of variational mode is determined.The borehole strain observation data are decomposed into five finite bandwidth modes by using Variational Mode Decomposition.The modes corresponding to strain solid tide,air temperature,air pressure and borehole water level are determined and removed through comparative analysis;the principal component eigenvalues and eigenvectors of residual modes are calculated,and the threshold analysis and similarity analysis are used to extract the anomalies of eigenvalues and eigenvectors.The case studies of Lushan earthquake and Wenchuan earthquake show that this method can effectively remove the effects of strain solid tide,air temperature,air pressure and borehole water level,and extract the pre-earthquake anomalies of Lushan earthquake and Wenchuan earthquake.