Alternative interpretations of horizontal to vertical spectral ratios of ambient vibrations: new insights from theoretical modeling

Different positions exist about the physical interpretation of horizontal to vertical spectral ratios (HVSR) deduced from ambient vibrations. Two of them are considered here: one is based on the hypothesis that HVSR are mainly conditioned by body waves approaching vertically the free surface, the other one assumes that they are determined by surface waves (Rayleigh and Love, with relevant upper modes) only. These interpretations can be seen as useful approximations of the actual physical process, whose reliability should be checked case-by-case. To this purpose, a general model has been here developed where ambient vibrations are assumed to be the complete wave field generated by a random distribution of independent harmonic point sources acting at the surface of a flat stratified visco-elastic Earth. Performances of the approximate interpretations and complete wave field models have been evaluated by considering a simple theoretical subsoil configuration and an experimental setting where measured HVSR values were available. These analyses indicate that, at least as concerns the subsoil configurations here considered, the surface-waves approximation seems to produce reliable results for frequencies larger than the fundamental resonance frequency of the sedimentary layer. On the other hand, the body waves interpretation provides better results around the resonance frequency. It has been also demonstrated that the HVSR curve is sensitive to the presence of a source-free area around the receiver and that most energetic contribution of the body waves component comes from such local sources. This dependence from the sources distribution implies that, due to possible variations in human activities in the area where ambient vibrations are carried on, significant variations are expected to affect the experimental HVSR curve. Such variations, anyway, only weakly affect the location of HVSR maximum that confirms to be a robust indicator (in the range of 10%) of the local fundamental resonance frequency.