UC Berkeley

ABSTRACT Using a numerical modeling scheme, we evaluate the azimuthal source-to-site dependence of the horizontal-to-vertical spectral ratio (HVSR) by comparing wavefields propagated through a horizontally layered velocity structure and more complex velocity structures. We choose the best combination of the radial, tangential, and vertical source components to model the ambient noise’s seismic source that better explains the observations on a test site. We adopt this derived source configuration to generate simulated wavefields propagated in a horizontally layered velocity model (1D) and a canonical 3D velocity structure and then compute their HVSR spectra. Our simulations use four distinct source locations, each providing a different source–receiver azimuth, to evaluate how variations in wave propagation from these configurations influence HVSR spectra. We extend the analysis to two additional velocity models by introducing stochastic perturbations to the deterministic 1D and 3D velocity structures. We find that the source configuration strongly affects the HVSR amplitude across all analyzed velocity structures. The results suggest that the spectral shape of HVSR tends to be stable along all azimuths in sites placed over velocity structures in which a horizontally layered velocity model can be assumed. Conversely, sites located over 3D velocity structures may exhibit strong azimuth-dependent spectral shapes in HVSR if the wavefield is governed by unidirectional wave fronts, even in some cases making the predominant frequency dependent on the source-to-site azimuth. We propose a methodology based on 2D seismic array analysis and wavenumber-domain filtering to identify HVSR spectral-shape fluctuations as a function of azimuth. We discuss our findings and implications in constraining site terms in ground-motion modeling.