Probabilistic models of seismic ground acceleration are reviewed. The
review includes uniformly modulated, oscillatory, amplitude and phase
modulated, ARMA, and other processes. Moreover, a novel and general no
nstationary Gaussian model is presented. The model consists of a super
position of harmonics with correlated Gaussian amplitudes. The harmoni
c frequencies and the amplitude probabilistic characteristics depend o
n the duration and the correlation of the target seismic ground motion
, respectively. The model can be generalized to represent the seismic
ground motion at two or more sites. Methods are also presented for gen
erating realizations of the probabilistic models in the paper. The met
hods are applied to demonstrate the generation of artificial seismic g
round motions and to evaluate the usefulness of various probabilistic
models in earthquake engineering.