Publikationsdatum:
2023-04-20
Beschreibung:
In current approaches for developing non-ergodic ground-motion models (GMMs), the non-ergodic path effect is through the cell-specific linear-distance scaling (Dawood & Rodriguez-Marek, 2010) that mimics the effects from a 2-D Q structure, but the significant Q effects on the ground-motion amplitudes are mainly limited to short-period ground motion at large distances. The path effects due to 3-D velocity structures are not captured by the cell-specific distance scaling approach. In this study, we adopt the new methodology from Sung et al. (2023) and Lacour et al. (2023) to estimate the spatial distribution of non-ergodic path effects using residuals from the Morikawa and Fujiwara (2013) GMM to model path effects related to 3-D velocity structures using the varying coefficient model (VCM). The epistemic uncertainty of the non-ergodic term is given by the posterior distribution in the VCM, which is small in regions with data and large in regions with spare data. The dataset is from the dense strong-motion observation networks in Japan (e.g., K-NET and KiK-net). The standard deviation of non-ergodic path terms is approximately 0.15 LN units when only considering 2-D Q path terms, whereas the standard deviation is much larger (about 0.30-0.35 LN units) if both Q and 3-D velocity structure non-ergodic terms are included. Fully non-ergodic models lead to an aleatory variance of residuals for GMMs that is reduced by 40-70% compared to ergodic GMMs. The combined effect of the shift in the median with the reduced aleatory variability can significantly affect seismic hazard calculations for the Japan region.
Sprache:
Englisch
Materialart:
info:eu-repo/semantics/conferenceObject
Permalink