A Spectral-Based Stochastic Ground Motion Model with a Non-Parametric Time-Modulating Function

This study presents a novel fully non-stationary stochastic model for simulation of synthetic arrays of ground motion. The model employs a modulated and filtered white-noise process defined via spectral represen- tation. This study is an extension of and improvement upon the previ- ous work of Rezaeian and Der Kiureghian. The proposed model preserves the key advantage of separating the temporal and spectral non-stationary characteristics of the process, thus simplifying the modeling and parameter estimation. Moreover, the spectral representation used here allows added flexibility in the choice of the filter. Additionally, we propose a novel non- parametric time-modulating function based on a monotonic cubic interpo- lation and that has the following advantages: (i) it is directly defined by physically meaningful parameters related to time intervals of the buildup of Arias intensity of the recorded ground motion, thus simplifying the fitting procedure; (ii) it allows for a more flexible range of shapes, thereby accom- modating multiple peaks. To ensure zero residual velocity and displacement, a high-pass filter is applied according to the evolutionary theory of Priest- ley. This, together with an energy correction factor, eliminates a bias on the cumulative energy of the post-processed simulated motions, which is implicitly present in most stochastic simulations of ground motion.