Contrasting offshore and onshore strong ground motions from the 2025 Mw 7.6 Aomori subduction interface earthquake: Characteristics and model evaluation

Abstract

On December 8, 2025, an Mw 7.6 subduction interface earthquake occurred off the coast of Aomori Prefecture, Japan, representing the largest-magnitude event recorded to date by the Seafloor Observation Network for Earthquakes and Tsunamis (S-net). Using a dataset comprising 84 offshore and 418 onshore strong-motion records, this study systematically compares the spatial distributions and attenuation characteristics of peak ground acceleration (PGA), spectral accelerations across multiple periods, and significant duration metrics (D70 and D90). The predictive performance of existing onshore and offshore ground-motion models (GMMs) is quantitatively evaluated. Our key findings are as follows: (1) Offshore ground motions exhibit markedly slower attenuation and systematically longer durations than their onshore counterparts. The region of maximum PGA is displaced from the epicenter and concentrated in areas with unburied seafloor stations, whereas the spatial centroid of long-period (T = 5.0 s) spectral acceleration shifts back toward the source region. This pattern indicates that short-to-intermediate-period components are predominantly controlled by local sedimentary layers, while long-period components are governed by source processes. (2) Observed ground motions from this event are generally lower than median predictions from existing GMMs, although most observations fall within ±1 standard deviation of the predictions. Offshore-specific models outperform onshore models in predicting offshore PGA and short-to-intermediate-period spectral accelerations, while onshore duration models systematically underestimate offshore significant duration. (3) Intra-event residual analysis reveals that sediment thickness (ST) estimated from seismic reflection surveys serves as a more spatially stable site-characterization parameter than D1400 (depth to shear-wave velocity of 1.4 km/s). Moreover, water depth exhibits a nonlinear influence on prediction residuals, suggesting that a piecewise functional representation may be more appropriate. Collectively, these results underscore the necessity of employing region-specific ground-motion input models that distinguish between offshore and onshore conditions in marine seismic design and offshore hazard assessments, and highlight the importance of developing independent offshore duration prediction models.