An efficient ground motion model coupling spatial correlations for large-scale field simulation

Efficient simulation of ground motion across large-scale fields is essential for emergency response and urban-scale risk assessment. Existing numerical approaches face prohibitive computational costs in large-scale simulations. Although stochastic models are more computationally efficient, current implementations focus on single-site scenarios and have yet to consider the spatial variability of ground motions. This study proposes a stochastic ground motion model coupling spatial correlations for rapid large-scale field simulation. The proposed model consists of two parts: (1) a five-parameter time-frequency-power spectral density (TFPSD) model that balances computational efficiency with sufficient time-frequency characterization, and (2) the first-of-kind developed spatial correlation models that account for the TFPSD model parameters. Predictive equations were developed to link the five parameters to three typical earthquake parameters: moment magnitude, rupture distance, and site shear wave velocity. Spatial correlation models are developed by analyzing intra-event residuals to capture the spatial variability of ground motions. The validity of the TFPSD model is established by comparing the pseudo-spectral accelerations (PSA) obtained from simulated samples with those derived from historical records. The integrated application of the TFPSD model and the spatial correlation model demonstrates that the proposed method simulates a ground motion field over a 50 km × 50 km area with a grid resolution of 0.5 km in merely 80.9 s, underscoring its high computational efficiency. Comparisons of PSA values obtained from simulations with and without spatial correlation highlight the importance and superiority of the proposed spatial correlation models. • A five-parameter time-frequency-power spectral density (TFPSD) model was proposed. • A first of kind spatial correlation model for TFPSD model is proposed. • The proposed model could rapidly simulate large scale ground motion field.