Evaluation of the Maximum-Likelihood Estimates of Site Fundamental Frequencies for a Subset of KiK-Net Strong Motion Data

In ground motion prediction equations (GMPEs), site characteristics are usually addressed by the shear-wave velocity of top 30 m (V_S30) or depth to 1.0 km/s shear-wave velocity isosurface (Z_1.0); however, these parameters are either inferred or limited to the shallow depths. Therefore, using other site proxies such as site fundamental frequency is useful. In this study, a data set from the KiK-net strong ground motion network is used to estimate the maximum-likelihood estimate of site fundamental frequency (f_ml) and its corresponding amplitude (A_ml) using the horizontal-to-vertical spectral ratio (HVSR) of surface ground-motions. First, the recorded ground motions with PGA ≥ 0.1% g and Mw > 3 are selected and processed for 100 stations. Second, the HVSR is calculated for each recorded event at each site. Third, f_ml and A_ml are estimated for each site using previously developed automated methodologies by the authors. These automated methodologies provide the uncertainties associated with f_ml and A_ml estimates, whereas the conventional HVSR methods yield one value. Lastly, the results obtained in this study are compared with a recent independent study on a similar data set. The results of automated methods are very consistent with those of the independent study. According to the results, using f_ml can be referred to using the first peak frequency (f₀) and highest peak frequency (f_peak) in the HVSR curve. In addition, the findings show that the Fourier amplitude spectrum and 5%-damped pseudo-spectral acceleration have fairly comparable performance in estimating site fundamental frequency and its amplitude.