Fourier amplitude spectra (FAS) offer a more direct representation of ground motion compared to acceleration response spectra, leading to significant attention on modeling of FAS in engineering seismology. Advances in ground‐motion records and computer algorithms have relaxed ergodic assumptions, enabling the development of nonergodic ground‐motion models (GMMs). An offshore ergodic GMM for the smoothed effective amplitude spectrum (EAS) is developed in this study. This model categorizes the S‐net stations into buried and unburied based on deployment method. The offshore ergodic EAS GMM is applicable for predicting various subduction earthquake scenarios in the Japan trench area, covering moment magnitudes from 4 to 7.4 and rupture distances up to 300 km. It demonstrates high amplitudes for unburied stations at low frequencies, for buried stations at high frequencies, and for intraslab events at high frequencies. There is a significant difference in frequency content between offshore and onshore ground motions as by comparing the offshore and onshore ergodic EAS GMM. Using the ergodic EAS GMM as a backbone, an offshore nonergodic EAS GMM is developed using Bayesian inference with the integrated nested Laplace approximation to reveal spatial varying path, site, and source effects. The nonergodic EAS GMM exhibits reduced aleatory variability, which is crucial for probabilistic seismic hazard analysis and seismic risk assessment. However, it also shows large epistemic uncertainty in areas with sparse ground‐motion data and smaller uncertainty in areas with abundant data. The results will provide theoretical basis for offshore seismic zoning, risk assessment, and earthquake engineering warning.

You do not have access to this content, please speak to your institutional administrator if you feel you should have access.