PHYSICS-BASED SITE-AMPLIFICATION PREDICTION EQUATIONS: A DREAM AT REACH ?
Résumé
Previous ESG blind tests and symposia, together with the advances in computation capabilities, allowed to significantly improve the reliability of numerical simulation for site response estimates. In parallel, the considerable progress in instrumentation has made available numerous high quality data sets, which allowed impressive developments in empirical ground motion prediction tools (GMPEs). Yet a number of important site amplification components are not, or incompletely, accounted for in such GMPEs, because of too poor site metadata or missing recordings, given the complexity of site amplification physics. One might dream that present-day numerical simulation tools, large data sets, and machine learning could be merged to establish more physics-based, statistically meaningful site terms providing a better (i.e., more accurate and less scattered) prediction of site amplification. This paper aims at discussing the maturity of such dreams for different components of site amplification, i.e., 1D linear site response, non-linear modifications, effects associated with surface topography, and aggravation factors in valleys and basins.
Origine : Fichiers éditeurs autorisés sur une archive ouverte