Seismology-based hybrid ground motion prediction models of PGA in Sichuan, China

Current regional ground motion prediction models (GMPMs) of areas lacking strong earthquake records employ imperfect methods or earthquake records. A new weighting method of seismology to clarify the application scope of point-source and finite-fault models is proposed in this study. According to the local small earthquake data, five regional parameters in the Sichuan area of China are inverted, and combined with other regional tectonic parameters, point-source models are established to synthesize the GMPMs of the peak ground acceleration (PGA) of Mw 5.0–7.0. Based on the finite-fault models, the PGA of Mw 6.0–8.5 of sites in four directions is synthesized, and GMPMs of major and minor axes are summed. In two-dimensional tabular form, the GMPMs between the point-source and finite-fault models consist of the weighted average between Mw 6.0 and 7.0; point-source estimation is used for PGA less than Mw 6.0, and finite-fault estimation is used for PGA larger than Mw 7.0. Finally, hybrid GMPMs of PGA with Mw 5.0–8.5 and distances within 300 km are established. Compared with the GMPMs of the postearthquake observation data and the recent edition of seismic hazard maps in Sichuan, there is a significant improvement in the areas lacking strong earthquake data. In contrast to the American logic tree weighting method, this method shows the advantage of avoiding many repeated calculations.