地震动
地质学
运动(物理)
大地测量学
物理
经典力学
地震学
作者
Chenying Liu,Jorge Macedo,Norman A. Abrahamson,Albert Kottke
摘要
ABSTRACT Existing ground-motion models (GMMs) for Türkiye have primarily relied on ergodic or partially nonergodic frameworks, which are limited in capturing systematic ground-motion characteristics and the trade-off between aleatory and epistemic uncertainties. This study uses a comprehensive Turkish database of over 35,000 shallow crustal ground-motion recordings spanning 1983–2023 to develop a fully nonergodic GMM that accounts for systematic source, site, and path effects. The backbone of the nonergodic model is a regionalized ergodic GMM developed as part of this study, which incorporates an event-specific attenuation term. The nonergodic systematic effects are modeled using Gaussian processes (GPs) based on the residuals of the ergodic backbone GMM. GP parameters are constrained through a variogram-based approach, which is shown to provide advantages compared to a direct GP fitting. Path effects are evaluated using cell-specific attenuation and spatially correlated GPs, accounting for variations in the quality factor and velocity structure. The evaluated source, site, and path terms exhibit different spatial patterns and correlation lengths across spectral periods. Potential factors influencing the patterns are discussed. In particular, we find that the correlation lengths of kappa and the site term at short periods are comparable, suggesting dominating kappa effects at short periods. In addition, we observe parallels between the estimated path effects and available quality factor maps for Türkiye. The spatial distribution of short-period source effects and regional stress drop in western Türkiye also share similarities. The developed nonergodic model reduces aleatory variability by 35%–50% depending on the spectral period. This reduction, once considered in the context of epistemic uncertainties, is key for seismic hazard quantification and performance-based earthquake engineering applications in Türkiye.
科研通智能强力驱动
Strongly Powered by AbleSci AI