[S-SS09] Strong Ground Motion and Earthquake Disaster

[SSS09-18] Quantitative assessing the variability of earthquake-source models to ground motion prediction: A case study in the Ryukyu subduction zone

*Ming-Che Hsieh¹, Yung-Ching Yang¹, Chih-Hsuan Sung² (1.National Central University, 2.University of California, Berkeley)

Keywords:Ryukyu subduction zone, Characteristic source model, Ground motion simulation, Ground motion model

Historically, an M8 earthquake took place offshore east Taiwan in 1920. Studies on this earthquake indicate that the event might be related to the existence of the Ryukyu subduction zone. To estimate the possible ground motions in eastern Taiwan if earthquakes with similar magnitudes happen again, we assume an M8 scenario earthquake in the Ryukyu subduction zone. In order to capture the possible ground motions in our target region from the M8 scenario, a series of characteristic source models (CSMs) is generated according to the procedure of Recipe. Kinematic fault-rupture parameters are modeled, including rupture directivity, rupture speed, and asperity distribution. Therefore, a 3-D traction-image finite-difference method (FDM) is utilized to perform full-waveform ground motion simulations to test the variability of ground motions from the given CSMs. Tomographic velocity models and topographic relief are accounted for in these simulations. Each simulation costs about 14 hours, using 96 CPU cores to simulate 150-s wave propagation in the source-to-station region. A total of 24,755 fictitious stations with an inter-station spacing of 500 m are deployed to output 3-component synthetic waveforms. We adopt the RotD50 approach to calculate spectral acceleration and pick 3- and 5-second PSAs for ground motion model (GMM) analysis. The comparisons show that the simulations could be a good indicator of long-period ground motions and be further merged into current ergodic or non-ergodic GMMs and even seismic hazard assessment.