ABSTRACT Two types of data commonly used for microtremor exploration are phase-velocity dispersion curves obtained through an array measurement and horizontal-to-vertical spectral ratios (HVSRs) obtainable by a single-station measurement. Phase-velocity dispersion curves obtained by applying the spatial autocorrelation method to the array waveforms have a characteristic peaked shape in some cases. This dispersion curve shape has traditionally been explained as a consequence of the predominance of higher modes over fundamental mode in the Rayleigh waves. In this study, the effects of body waves on phase velocities and HVSRs were investigated based on both field measurements and theoretical calculations of microtremors. We used vertical-component array waveforms and single-station three-component waveforms of microtremors, obtained at and around a site where combined P-wave–S-wave (PS) and density loggings were conducted in the Kyoto basin, Japan (site KD-1), to identify phase velocities and HVSRs at frequencies in the range 0.2–2 Hz. The corresponding theoretical phase velocities and HVSRs were identified using full-wavefield synthetic data, which were generated assuming excitation points randomly distributed over the surface of a horizontally stratified velocity structure model created based on the logging data. The following key results were obtained. The measured phase-velocity dispersion curve exhibits a peaked shape with the value exceeding the S-wave velocity of the Tamba Group (Tb-Group), which is the bedrock (half-space) of the velocity structure model. Theoretical calculations based on the surface-wavefield theory were unable to reproduce this peaked shape; however, theoretical calculations based on the full-wavefield theory reproduced it with extraordinary accuracy. To reproduce the peaked shape based on the surface-wavefield theory, it was necessary to construct a model containing a cap (i.e., high-velocity layer) connected under the Tb-Group. The theoretical calculation based on the full wavefield also accurately reproduced the peak value and peak frequency of the measured HVSRs.
Read full abstract