Large-Scale Physics-Based Broadband (up to 8 Hz) Ground-Motion Simulations of Mw 6.4+ Earthquakes for Hualien, Taiwan, Area

Abstract Surrounded by multiple devastating seismic faults, complex surface topography and ocean bathymetry, and with underlying soft alluvial deposits, all these detrimental factors make Hualien city and nearby areas in Taiwan one of the most seismically active and vulnerable regions in the world. In this study, we aimed to leverage the physics-based ground-motion simulation method to re-evaluate the ground-shaking levels induced by four catastrophic historical earthquakes that happened near Hualien city. Specifically, we investigated the 1951 Mw 7.3 Hualien earthquake––which is generally considered the first well-documented Mw 7+ earthquake event in the Hualien area that caused severe local destruction. We also revisited the 2013 Mw 6.4 Ruisui, 2018 Mw 6.4 Hualien, and the most recent 2024 Mw 7.4 Hualien earthquakes that all caused severe structural and geohazards. These historical earthquakes were simulated on a supercomputer with very large scale and achieved a maximum frequency (fmax) of 8 Hz. The simulation incorporated realistic 3D velocity, topography, and bathymetry, and calibrated finite-fault slip models. The recorded and simulated waveforms were compared, and good agreement demonstrated the validity of our simulations. Furthermore, we tweaked the hypocenter of each earthquake and placed it in three different locations along the strike direction of the fault to explore the directivity effects. For each hypocenter, four finite-fault models were generated to consider the uncertainty in source modeling. We recorded the waveforms of the entire surface of the simulated area, extracted the ground-motion intensity measures such as peak ground acceleration and peak ground velocity, and compared the spectral acceleration against an empirical ground-motion model (GMM). The results indicate that the physics-based simulations can exceed GMM’s mean predictions, where significant site amplifications are observed. The enormous synthetic ground-motion database provided by this study will also benefit the seismic hazard and risk analysis for the Hualien area.