Stochastic Ground Motion Model with Time-varying Intensity, Frequency and Bandwidth Characteristics

A recently developed nonstationary stochastic model of strong earthquake ground motion is described. The model employs filtering of a white-noise process, where nonstationarity is achieved by modulating the intensity and varying the filter parameters in time. The model is fitted to a target accelerogram by matching a set of statistical measures that characterize the evolution of the intensity, predominant frequency and bandwidth of the record. For performance-based earthquake engineering (PBEE) analysis, the model parameters are identified for a dataset of strong ground motion records, where each accelerogram is treated as a target ground motion with known earthquake and site characteristics. Statistical data analysis is then performed on the identified model parameters to investigate relations that will help predict the model parameters for a given set of earthquake and site characteristics. Preliminary results from this analysis, including fitted distributions of the model parameters, are the focus of this paper. Once these relations are fully developed, simulation algorithms will be used to generate an ensemble of artificial ground motion records for a given set of earthquake and site characteristics. Such simulated motions can be used in place of actually recorded accelerograms in PBEE analysis, thus avoiding the difficult and often questionable tasks of selecting and scaling ground motions recorded at different sites for a single project site.