Prediction models for significant duration of offshore ground motions in subduction zone using the S-Net database

Abstract

There have been relatively few studies on significant duration prediction models for offshore ground motions in subduction zones, and the differences in significant durations between offshore and onshore ground motions remain insufficiently understood. In this study, 19,058 offshore strong-motion records from 150 ocean-bottom sites within the S-net in the Japan Trench were analyzed to develop prediction models for offshore ground motion significant durations of interface and slab earthquakes. The proposed model incorporates the effects of moment magnitude, rupture distance, focal depth, site conditions, and station installation methods across different earthquake types in subduction zones. Furthermore, a comprehensive comparison was conducted between the significant duration prediction models for offshore and onshore ground motions. Based on within-event residuals and single-station residuals of offshore significant durations, a spatial correlation model for offshore significant durations was developed using a combination of semi-variogram functions and exponential regression. This approach enabled a detailed analysis of spatial correlation differences between interface and slab earthquakes. Results indicate that the offshore significant duration models for interface and slab earthquakes exhibit high predictive accuracy. Notable differences were observed in the significant duration between offshore subduction slab and interface ground motions. Site classification of ocean-bottom stations based on the HVSR technique more effectively captures the effect of the predominant frequency of ocean-bottom sites on offshore significant durations. Comparisons of significant durations between offshore and onshore ground motions across different earthquake types reveal that offshore ground motions generally have longer significant durations than onshore motions. Moreover, the spatial correlation coefficient of offshore significant durations decreases with increasing separation distance. The proposed prediction models for offshore significant durations of interface and slab events, as well as the corresponding spatial correlation model, can serve as valuable references for seismic hazard assessment and seismic design in offshore regions.