Polarization characteristics of Rayleigh waves to improve seismic site effects analysis by HVSR method

The horizontal-to-vertical spectral ratio (HVSR) method based on ambient noise recordings is a popular and efficient approach to estimate the seismic site effect parameters. This method is sufficiently robust in determining the predominant frequency (f0) of the site, whereas its estimate of the amplification factor lacks consistency because generation mechanisms, source locations and proportions of the wave types making up ambient noise vary locally and are not well understood, in addition to lack of clarity in the theory of the HVSR method. As the generation mechanisms, source locations and wave compositions of the single- and double-frequency (SF and DF respectively) microseisms (0.05–0.5 Hz) are better studied compared to the noise in other frequency bands, this study focuses on investigating the site effects on the fundamental Rayleigh (Rg) waves in the SF and DF microseisms recorded at 30 Transportable Array stations in the east coast of the United States. HVSRs are calculated first to obtain f0s as well as the unconsolidated sediment thicknesses (USTs). Then the polarization analyses are carried out to select the ellipsoids approximating the particle motions of the Rg waves and to estimate the HVSR of these Rg ellipsoids (HVSRR). The polarization parameters and HVSRR indicate that the Rg waves' particle motions form oblate ellipsoids in the UST range whose f0 values fall within the DF band. Comparison of the observed HVSR and HVSRR and the computed HVSR values based on the body wave and the Rayleigh wave theories shows that the observed HVSR in the DF band on bedrock and/or thin unconsolidated sediments are mainly due to the ellipticity of Rg waves, while those on thick unconsolidated sediments whose f0 values fall within the DF band result from the combined effects of the ellipticity of Rg waves and multiple reflection of S-waves. This outcome will help develop procedures to improve the consistency of HVSR's amplification estimates and hence its reliability in seismic site effects analysis.