Simulation of long-period ground motion in the Osaka sedimentary basin: performance estimation and the basin structure effects

Abstract

We assessed the applicability of the 3-D subsurface velocity structure model of the Osaka sedimentary basin in western Japan for simulations of long-period ground motion (3–20 s) based on two types of long-period ground motion characteristics: (1) strong ground motion from a specific earthquake and (2) site-specific, event-independent horizontal-to-vertical spectral ratio (HV) characteristics obtained from earthquakes with various azimuths. For the first type, we compared the observed and synthetic ground motions in terms of velocity waveforms and pseudo-velocity response spectra; we then proposed an objective index to evaluate the goodness-of-fit from one site to the next. The simulation satisfactorily reproduced the observed ground motion at many stations; however, the model needs improvement for reproducing the ground motion at some stations. For the second type, we first demonstrated that site-specific HV characteristics can be obtained from the coda portion of strong motion observation records for earthquakes with various azimuths. We then demonstrated that the HVs obtained from 3-D ground motion simulations for earthquakes with various azimuths (3DHV) also have site-specific stable values. The synthetic 3DHV was compared with the observed HV, and the 3-D basin velocity structure model was demonstrated as able to reproduce the HV peak periods within ±1 s at many stations. However, at some stations, the peak periods of the 3DHV were systematically longer than those of the observed HV by more than 1.5 s, which corresponded to the stations of poor goodness-of-fit scores in the ground motion simulation. 3DHV was also compared with the theoretical ellipticity of the fundamental Rayleigh wave (1DHV). The peak periods of 3DHV and 1DHV were inconsistent with each other at some sites, which implies that 3DHV is affected by the lateral heterogeneity of the structure around the site. We found the systematically biased characteristics of the discrepancy in 3DHV and 1DHV peaks in specific regions. The spatial extent of the area that affects 3DHV is related to the complexity in the bedrock structure and the dominant period of 3DHV.