Quantitative evidence for the advantage of fourier-spectrum based evaluation of site amplifications: Towards utilization of small-magnitude records

It is a fundamental question in earthquake engineering whether seismic hazard analysis for the design of infrastructure should be based on response spectra or Fourier spectra. This study focused on the evaluation of site amplification factors (SAFs) and investigated whether the choice of response/Fourier spectra affects the degree of accuracy to which SAFs can be incorporated into a hazard analysis. The focus was on nonergodic approaches in which SAFs are evaluated based on on-site recordings. While theory indicates that Fourier-spectrum based SAFs are independent of the earthquake source characteristics whereas response-spectrum based SAFs are dependent on magnitudes, this indication has not been supported by quantitative evidence in the existing literature. In this study, we focused on five carefully selected pairs of strong-motion observation sites from Japanese networks, each comprising a thick sediment site and a nearby stiff site, both with a substantial number of strong-motion records. Such station pairs are extremely rare worldwide. By calculating the spectral ratios, we did find that the Fourier spectral ratios do not exhibit magnitude dependence, whereas the response spectral ratios for large earthquakes will be underestimated if they are evaluated based on smaller earthquakes in the long period range. The cause of the underestimation is explained by a simple model. The use of Fourier-spectrum based analysis will contribute to the effective utilization of records from small-magnitude earthquakes, which will contribute to the accurate evaluation of SAFs in high seismicity countries and regions across the world based on on-site recordings.