An Efficient Method for Probability Prediction of Peak Ground Acceleration Using Fourier Amplitude Spectral Model

Abstract

ABSTRACT The probabilistic prediction of peak ground acceleration (PGA) using the Fourier amplitude spectral (FAS) model has many advantages in regions lacking strong ground-motion records. Currently, the implementation of this approach for the calculation of annual exceedance rate of PGA relies on Monte Carlo simulations (MCSs). However, adopting MCS requires many times calculations of PGA from FAS, and each time of calculation includes complicated integrals, the computational cost is too high to be acceptable for practical applications. Therefore, this study proposes an efficient method for the probabilistic prediction of PGA using the FAS model. For this purpose, a probabilistic analysis method, referred to as the moment method, was introduced to improve computational efficiency. The probability distribution of PGA was approximated using a three-parameter distribution defined according to the first three moments. The first three moments of the PGA were obtained based on the point-estimate and dimension-reduction integration method. Numerical examples were conducted to verify the proposed method. It was found that the proposed method not only performed much more efficiently than using MCS in calculating the annual exceedance rate of PGA to obtain the hazard curve but also provides nearly the same accuracy as MCS.