SummaryA predictive stochastic model is developed based on regression relations that inputs a given earthquake scenario description and outputs seismic ground acceleration time histories at a site of interest. A bimodal parametric nonstationary Kanai-Tajimi (K-T) ground motion model lies at the core of the proposed predictive model. The functional forms that describe the temporal evolution of the K-T model parameters can effectively represent strong non-stationarities of the ground motion. Fully non-stationary ground motion time histories can be generated through the powerful Spectral Representation Method. A Californian subset of the available NGA-West2 database is used to develop and calibrate the predictive model. Samples of the model parameters are obtained by fitting the K-T model to the database records, and the resulting marginal distributions of the model parameters are efficiently described by standard probability models. The samples are translated to the standard normal space and linear random-effect regression models are established relating the transformed normal parameters to the commonly used earthquake scenario defining predictors: moment magnitude M w , closest-to-site distance R rup , and average shear-wave velocity V S 30 at a site of interest. The random-effect terms in the developed regression models can effectively model the correlation among ground motions of the same earthquake event, in parallel to taking into account the location-dependent effects of each site. For validation purposes, simulated acceleration time histories based on the proposed predictive model are compared with recorded ground motions. In addition, the median and median plus/minus one standard deviation elastic response spectra of synthetic ground motions, pertaining to a variety of different earthquake scenarios, are compared to the associated response spectra computed by the NGA-West2 ground motion prediction equations and found to be in excellent agreement. KEYWORDSpredictive stochastic ground motion model, analytical evolutionary power spectrum, ground motion parametrization, random-effect regression, NGA-West2 database

No abstract

No abstract

No abstract

No abstract