Uncertainties and variabilities in seismic ground response analyses

Ground response analyses (GRAs) represent a key element for the nonergodic (site-specific) evaluation of the seismic hazard. In this respect, epistemic uncertainties and aleatory variabilities need to be properly identified, quantified, and managed to obtain consistent estimates. Several collaborative efforts have been carried out in recent years to assess the influence of uncertainties and variabilities in the GRA parameters through benchmark studies. Specifically for site characterization, efforts are required to guarantee a sufficient quality of in situ and laboratory tests. However, benchmark tests have shown the existence of a certain level of “uncompressible uncertainty”. Stochastic models implemented in GRAs are therefore required to evaluate the impact of uncertainties and variabilities on the computed seismic hazard. Such models are to be based on large databases of experimental data to produce consistent estimates. Examples will be provided with a specific focus on geophysical tests for the evaluation of shear wave velocity models, which are likely the most influent parameters in GRAs.