How topography-dependent are topographic effects? Complementary numerical modeling of centrifuge experiments

Abstract

Topographic effects, the modification of seismic shaking by irregular topographies compared to flat ground, have been extensively studied. Very few studies, however, have investigated the effects of the stratigraphy and nonlinear response of the underlying geology on topographic amplification. Furthermore, most experimental studies have been performed in the field, where it is often difficult to establish an ideal flat-ground reference station, as well as to characterize the soil properties and their spatial variability in sufficient detail. Dafni [1] recently tested the seismic response of step-like slopes in a series of centrifuge experiments, where the incident motion, reference station and material properties were characterized in detail. In this study, we investigated the influence of the container boundary on topographic effects observed in the centrifuge experiments by performing numerical simulations with and without the container boundary. Our analysis suggested that the rigid-body rocking motion of the centrifuge container likely increased the experimental topographic spectral ratios, contributing to the discrepancy between the simulated and observed spectral ratios. We also found that although the laminar box lateral boundaries caused spurious reflections, they didn’t qualitatively affect the ground surface amplification pattern compared to numerical predictions of the same configuration without boundaries. At the same time, and most importantly, however, we found that the baseplate –by trapping waves scattered and distracted by the slope– amplified the ground motion at the crest up to one order of magnitude compared to numerical predictions of the response in absence of the baseplate. Our results show that topographic effects can be significantly affected by the underlying soil stratigraphy; and allude to the potentially significant role of this phenomenon in elevating seismic risk in regions with strong topographic relief. The findings of this study also suggest that future studies will benefit from clear understanding and careful considerations of capabilities and limitations of different investigation methods and that the numerical modelling and the lab testing (or the field testing) methods should complement each other.