Influence of 2D heterogeneous elastic soil properties on surface ground motion spatial variability: Fid-Bau Portal

Influence of 2D heterogeneous elastic soil properties on surface ground motion spatial variability

El Haber, E. / Cornou, C. / Jongmans, D. / Youssef Abdelmassih, D. / Lopez-Caballero, F. / AL-Bittar, T.

Abstract Spatial variability of earthquake ground motion (SVEGM) refers to the differences in amplitude and phase between recordings of the same earthquake at different locations. In the near-surface, geological processes (sedimentation, erosion) and anthropogenic activities can lead to small scale spatial heterogeneities of soil mechanical properties, which may affect SVEGM. In this paper, the effect of shallow 2D spatial variability of the shear wave velocity (Vs) on the surface ground motion is assessed through a set of numerical experiments, using a simple 2D velocity structure (a sedimentary layer over a half-space). Non-linearity or damping are not considered in the wave propagation calculation in order to solely focus on the effects of soil elastic property variability. Vs is modeled as a random field using the EOLE method (Expansion Optimal Linear Estimation) and considering three statistical parameters: the coefficient of variation, and the horizontal and vertical autocorrelation distances. Seismic ground motions are numerically simulated for a plane wave excitation with SV polarization. Modeling results clearly highlight the scattering of surface waves by ground heterogeneities, leading to large spatial variations in surface ground motion. We computed surface ground motion indicators (resonance frequency, spectral amplification, Arias intensity and duration) and we showed that their spatial variability is mainly controlled by the Vs coefficient of variation. A comparison between 2D and 1D ground motion probabilistic modeling shows that the 1D probabilistic approach may correctly reproduce average fundamental resonance frequencies and corresponding amplification. However, the 1D approach significantly under-predicts both ground motion amplification at higher frequencies and related variabilities, as well as Arias intensities and inferred durations, which are all controlled by the generation of locally diffracted surface waves.

Highlights • The shear-wave velocity coefficient of variation is the statistical parameter controlling the variability of the surface ground motion. • The effect of the horizontal autocorrelation distance is observed for small inter-receivers distances and for small wave length. • Locally diffracted surface waves generate ground motion with larger duration and energy. • 1D probabilistic approaches cannot replace 2D ones. They underestimate amplification at high frequencies and both duration and energy.