S43B-2781
Validation of Broadband Ground Motion Simulations for Japanese Crustal Earthquakes by the Recipe

Thursday, 17 December 2015
Poster Hall (Moscone South)
Asako Iwaki1, Takahiro Maeda1, Nobuyuki Morikawa2, Hiroe Miyake3 and Hiroyuki Fujiwara1, (1)National Research Institute for Earth Science and Disaster Prevention, Tsukuba, Japan, (2)NIED National Research Institute for Earth Science and Disaster Prevention, Tsukuba, Japan, (3)University of Tokyo, Bunkyo-ku, Japan
Abstract:
The Headquarters for Earthquake Research Promotion (HERP) of Japan has organized the broadband ground motion simulation method into a standard procedure called the “recipe” (HERP, 2009). In the recipe, the source rupture is represented by the characterized source model (Irikura and Miyake, 2011). The broadband ground motion time histories are computed by a hybrid approach: the 3-D finite-difference method (Aoi et al. 2004) and the stochastic Green’s function method (Dan and Sato, 1998; Dan et al. 2000) for the long- (> 1 s) and short-period (< 1 s) components, respectively, using the 3-D velocity structure model.

As the engineering significance of scenario earthquake ground motion prediction is increasing, thorough verification and validation are required for the simulation methods. This study presents the self-validation of the recipe for two MW6.6 crustal events in Japan, the 2000 Tottori and 2004 Chuetsu (Niigata) earthquakes. We first compare the simulated velocity time series with the observation. Main features of the velocity waveforms, such as the near-fault pulses and the large later phases on deep sediment sites are well reproduced by the simulations. Then we evaluate 5% damped pseudo acceleration spectra (PSA) in the framework of the SCEC Broadband Platform (BBP) validation (Dreger et al. 2015). The validation results are generally acceptable in the period range 0.1 – 10 s, whereas those in the shortest period range (0.01-0.1 s) are less satisfactory. We also evaluate the simulations with the 1-D velocity structure models used in the SCEC BBP validation exercise. Although the goodness-of-fit parameters for PSA do not significantly differ from those for the 3-D velocity structure model, noticeable differences in velocity waveforms are observed. Our results suggest the importance of 1) well-constrained 3-D velocity structure model for broadband ground motion simulations and 2) evaluation of time series of ground motion as well as response spectra.