Stochastic model for the simulation of synthetic main shock-aftershock ground motion sequences

According to the current seismic codes, structures are designed to resist the first damaging earthquake during their service life. However, after a strong main shock, a structure may still face damaging aftershocks. The main shock-aftershock sequence may result in major damage and eventually the collapse of a structure. Current studies on seismic hazard mainly focus on the modeling and simulation of main shocks. This paper proposes a stochastic model to generate main shock-aftershock sequences. The model takes into consideration the statistical properties of the aftershocks’ occurrence times, locations and magnitudes by using a branching aftershock sequence (BASS) model. The stochastic model generates pairs of orthogonal horizontal ground motions and has separable non-stationarity in time and spectral domains. Prediction equations are developed for the controlling parameters, where the predictors are the site conditions and the aftershock characteristics from the BASS model. The coefficients of the prediction equations and the correlation between the model parameters (of two horizontal components of one record or of several records in one sequence) are estimated using a database of aftershock accelerograms. A backward stepwise deletion method is used to simplify the initial candidate model and avoid overfitting the data. The most relevant features of the recorded ground motions such as intensity, spectral content and duration are well captured by the synthetic realizations. The spectral shapes of the synthetic aftershock ground motions show good agreement with the spectral shapes of the recorded data and capture well the underlying uncertainties. The model, based on easily identifiable engineering parameters, is a useful tool to incorporate effects of aftershocks into seismic analysis and design