hide
Free keywords:
Ground-motion model, Fourier spectra, Robust mixed-effects regression,Regionally adaptable,Seismic hazard and risk,Europe
Abstract:
Typical seismic ground-motion models predict the response spectral ordinates (GMM-SA), which are the damped responses of a suite of single-degree-of-freedom oscillators. Response spectra represent the response of an idealized structure to input ground-motion, but not the physics of the actual ground-motion. To complement the regionally adaptable GMM-SA of Kotha et al. (2020), we introduce a model capable of predicting Fourier amplitudes (GMM-FA); developed from the Engineering Strong Motion (ESM) dataset for pan-Europe. This GMM-FA reveals the very high variability of high frequency ground-motions, which are completely masked in a GMM-SA. By maintaining the development strategies of GMM-FA identical to that of the GMM-SA, we are able to evaluate the physical meaning of the spatial variability of anelastic attenuation and source characteristics. We find that a fully data-driven geospatial index, Activity Index (AIx), correlates well with the spatial variability of these physical effects. AIx is a fuzzy combination of seismicity and crustal parameters, and can be used to adapt the attenuation and source non-ergodicity of the GMM-FA to regions and tectonic localities sparsely sampled in ESM. While AIx, and a few other parameters we touch upon, may help understand the spatial variability of high frequency attenuation and source effects, the high frequency site-response variability—dominating the overall aleatory variance—is yet unresolvable. With the rapid increase in quantity and quality of ground-motion datasets, our work demonstrates the need to upgrade regionalization techniques, site-characterisation, and a paradigm shift towards Fourier ground-motion models to complement the traditional response spectra prediction models.
