Nonstationary Shear-Wave Velocity Randomization Approach to Propagate Small-Scale Spatial Shear-Wave Velocity Heterogeneities into Seismic Response

Recent studies in earthquake engineering have outlined the difficulty of ground response analyses (GRAs) to replicate the observed ground motion and related variability at borehole array sites. Improvement of the seismic site response estimation requires accounting for and propagating the uncertainties in local soil conditions into surface ground motion. Uncertainties in site conditions arise from a number of factors, which include the uncertainties in the shear-wave velocity (VS) that are mainly caused by the natural spatial variability of soils and rocks. In this paper, a novel VS randomization approach is proposed to propagate the small-scale spatial VS heterogeneities into samples of VS profiles within a nonstationary probabilistic framework, to be further used in one-dimensional (1D) GRAs. The nonstationary approach is based on partitioning a borehole base-case VS profile into several locally stationary layers. The proposed approach was applied at three European sites exhibiting different subsurface soil conditions. Compared with both the classical stationary and an approach from the literature for VS randomization, the proposed approach provides a set of VS profiles fully consistent with the pseudoexperimental site signatures in terms of surface-wave dispersion curves, fundamental and higher-mode resonance frequencies, and site amplification. This paper also outlines the importance of the method used to measure VS profile in both the estimation of depth-dependent variability of VS at a given site and the prediction of site response variability.