Comparative analysis of underground and surface seismometer waveform records based on deep neural network denoising method
ML 3.5 earthquake occurred in Feidong,Anhui Province on February 24,2024.The waveforms recorded by underground seismograph and surface seismograph at Huai'an Earthquake Monitoring Center Station were selected for conventional filtering and noise analysis.The seismic waveforms and noise data were separated by deep neural network denoising method,and time-frequency analysis was carried out.Results shows that,① Based on the original signal,the magnitude measured by the surface seismometer is larger;② After denoising,the seismic magnitude measured by surface seismometer and underground seismometer is similar,but the waveform components are different.The horizontal component of underground seismometer has more high-frequency components,while the vertical component of surface seismometer has more high-frequency components,and the P wave is developed and the amplitude is large,which is presumed to be caused by the amplification of seismic waves by the ground layer;③ The noise recorded by the surface seismometer contains high-frequency components and is presumed to be environmental background noise,while the frequency and intensity of the secondary noise generated by the underground seismometer when recording the earthquake show a gradually decreasing trend.According to the noise pattern,it is presumed to be caused by the resonance of the elastic support of the underground seismometer under the action of seismic wave field,indicating that the noise source has changed in different spaces.
underground and surface seismometerspectral characteristicsartificial neural networkDeepDenoiser methodseismic noise
万方数据
维普
