Identification and extraction of Rayleigh waves from three-component accelerometers based on polarization characteristics in time-frequency domain

Usually surface waves, compared to other types of waves, cause more displacements on ground surface. In addition, Rayleigh waves create elliptical motions. Consequently, Rayleigh waves have the most destructive effects on structures located on ground surface. Therefore, identification and extraction of surface waves, especially Rayleigh waves, are among the most important tasks in seismology and applied geophysics. Many studies have been carried out on this topic. In most cases, identification and extraction of surface waves are carried out entirely in time domain or entirely in frequency domain. Since these kinds of studies have some limitations, recent studies on surface waves have been carried out in time-frequency domain. Recently, new time-frequency transforms have been presented to transfer signals from time domain to time-frequency domain. Earthquake signals are nonstationary, thus precision of extracting surface waves from signals depends noticeably on time-frequency transform. In this study, synchrosqueesed continuous wavelet transform has been used to transfer signals to time-frequency domain. This transform has higher resolution than other transforms, and also, it is less dispersive than other transforms in timefrequency domain. In this study, the main polarization characteristic of Rayleigh waves, which is elliptical motion, has been used to filter and extract Rayleigh waves from three-component accelerometers. Reciprocal ellipticity parameter has been used as the polarization characteristic parameter. MATLAB software package has been used for programming and analysis in this study. First, significance of the mentioned transform compared to other transforms has been investigated by a synthetic signal. Then, a three-component signal has been created and the presented algorithm in this study has been verified by it. Finally, Rayleigh waves from several real signals have been extracted.