Use of Alternative Site Amplification Models in Simulations and Uncertainty Analysis

Ground motion simulations are used for earthquake engineering purposes as alternative to real records. The accuracy of the simulations is evaluated using misfit functions in terms of seismological and structural parameters computed for real and observed ground motions. Among all input parameters used in the simulations, site amplification factors carry significant uncertainty. Alternative approaches to estimate site amplifications include generic site amplifications based on quarter-wave-length technique, empirical Horizontal-to-Vertical-Spectral ratios and theoretical transfer functions computed from one-dimensional velocity models. In this study, ground motion simulations of past earthquakes recorded at select stations in the Erzincan and Düzce regions in Turkey and the Anza and Imperial Valley regions in Southern California, U.S.A., are performed with alternative site amplification estimates while keeping the source and path parameters the same. The misfits are computed for the peak and selected spectral ground motion parameters in time, frequency and period domains. The variability in the misfits according to alternative site amplification models is evaluated. The results reveal that use of alternative methods causes a large variation in the resulting ground motions and whenever possible, several techniques should be employed