A hybrid approach for broadband simulations of strong ground motion: The case of the 2008 Iwate–Miyagi Nairiku earthquake

We developed a hybridization code that merges the full-wave low-fre- quency (LF) signals with stochastic high-frequency (HF) synthetics to simulate broad- band strong ground motion for engineering applications. The approach is applied to reproduce, within the known source and structure constraints, the Kiban–Kyoshin net- work (KiK-net) and Kyoshin net (K-NET) accelerometric data recorded during the 2008 Iwate–Miyagi Nairiku earthquake (Mw 7.0).The results show that our approach is able to simulate satisfactorily the observed waveforms and the related response spectra over the broadband frequency range. We estimated the effects of the hybridization in the peak and spectral parameters through the goodness-of-fit criteria: the overall fit improves (∼15%–20%) when, instead of full- wave calculated seismograms, we use the low-pass filtered recordings, showing that a good match at LFs is fundamental to obtain reliable and consistent results even in the HF band. Peak ground velocities are less affected than accelerations by the inclusion of the HF contribution and are well reproduced by our method; peak ground acceler- ations show larger discrepancies due to the intrinsic characteristics of the stochastic model and the small-scale heterogeneity that affect the seismic radiation and propaga- tion at HFs. Nevertheless, simulated signals reproduce the observed ones fairly well, and an acceptable match is also found in the comparison of the spectral responses.

We developed a hybridization code that merges the full-wave low-frequency (LF) signals with stochastic high-frequency (HF) synthetics to simulate broadband strong ground motion for engineering applications. The approach is applied to reproduce, within the known source and structure constraints, the Kiban-Kyoshin network (KiK-net) and Kyoshin net (K-NET) accelerometric data recorded during the 2008 Iwate-Miyagi Nairiku earthquake (M-w 7.0). The results show that our approach is able to simulate satisfactorily the observed waveforms and the related response spectra over the broadband frequency range. We estimated the effects of the hybridization in the peak and spectral parameters through the goodness-of-fit criteria: the overall fit improves (similar to 15%-20%) when, instead of full-wave calculated seismograms, we use the low-pass filtered recordings, showing that a good match at LFs is fundamental to obtain reliable and consistent results even in the HF band. Peak ground velocities are less affected than accelerations by the inclusion of the HF contribution and are well reproduced by our method; peak ground accelerations show larger discrepancies due to the intrinsic characteristics of the stochastic model and the small-scale heterogeneity that affect the seismic radiation and propagation at HFs. Nevertheless, simulated signals reproduce the observed ones fairly well, and an acceptable match is also found in the comparison of the spectral responses.