Separation of rayleigh waves in ambient noise using polarisation-based method

Abstract

Noise is a general word that is used in seismology for minor earth vibration with small amplitude known as ambient noise or microtremor. Microtremors are composed of waves that are propagating in different directions. Separation of the noise wavefield into body waves (P, SV, SH) and surface waves (Rayleigh and Love waves), and quantify the proportion of each type of waves would be the desired purpose. Exchanging data from frequency domain to time domain or vice versa is useful in data analysis. Complex trace analysis, which is a transform technique, analyzes seismic data similar to analytic signal. However, this technique which relies on conventional Hilbert transformation is susceptible to random noise and abrupt frequency variations in seismic signals. Therefore, the Sparsity-based, adaptive S-transform accompanied with a sparsity-aware weighting function, inspired by the Geman-McClure and Laplace functions is used to address this problem. Obtaining the geological information from the seismic data motivates the researchers to use polarisation-based methods to discriminate between different phases of the seismic wave based on their polarities. Combining the intuitive and straightforward definition of the polarity in the frequency domain by the non-stationary properties of the time domain methods attracts the attention towards using time-frequency (TF) approaches owing to their potential in localizing the extracted information of the seismic waves in both time and frequency domains. Polarisation analysis also helps in detecting the site directional resonance more reliably than the previous known methods via separating the Rayleigh wave ellipticity.