Search

Broadband seismographs are used to collect seismic data over an extended period. Temperature, pressure, and humidity are field variables that can have an impact on the broadband seismograph’s performance as well as the accuracy of observational data. Variations in performance, geological structure, and noise source will cause the cross-correlation function of seismic ambient noise to alter. To evaluate the effectiveness of seismographs, we propose to use the whole cross-correlation function as well as the positive and negative time waveforms for feature extraction. The cross-correlation function is decomposed to evaluate the phase variations using Local Mean Decomposition (LMD) and Variational Mode Decomposition (VMD). This is followed by the calculation of the Multivariate Multiscale Fuzzy Entropy Partial Mean (MMFEPM). Wavelet Packet Decomposition (WPD) and MMFEPM are used to evaluate phase variations in the positive and negative time waveforms. WPD combined with Wavelet Feature Scale Entropy (WFSE) is chosen to evaluate the amplitude variations of the cross-correlation function and its positive and negative waveforms. The results show that the proposed methods can identify time offsets within 0.025-2.5s and amplitude variations of 10−16, which provide a new direction for evaluating the performance of seismographs using ambient noise.