Simulating site-specific nonlinear site response in low-to-moderate seismicity areas – insight from Switzerland

The significance of local site effects in seismic hazard assessment is well-recognised. However, nonlinear soil behaviour during high-strain conditions is often neglected or oversimplified, even for large return periods. This study aims to estimate the impact of nonlinearity and liquefaction on the local seismic hazard using the multistep approach by Janusz et al. (2024), in particular in low-to-moderate seismicity areas, where nonlinearity is not instrumentally observed and hence particularly difficult to assess. We focus on sites in Switzerland, where strong earthquakes are relatively rare, although documented in the past. An exemplary area is the subalpine sedimentary basin of Lucerne, which is vulnerable to nonlinearity and liquefaction because of soft alluvial deposits (V_S30<300 m/s) and a shallow water table depth (~1–4 m). 

Typically, simplified linear equivalent models are used for modelling nonlinearity. Here, we use fully nonlinear numerical estimators with a constitutive model that accounts for pore pressure excess development, allowing for simulating the dilatant behaviour of the soil and indicating the onset of liquefaction. Moreover, the soil models are often characterized either using generalised values from literature or expensive laboratory measurements, which may not reflect in-situ conditions. We use soil models calibrated using cone penetration tests (CPT), which are in-situ geotechnical surveys, allowing for site-specific assessment.

Our findings show that the impact of the nonlinear soil behaviour cannot be neglected even in low-to-moderate seismicity areas like Lucerne. In the case of a strong shaking consistent with the local seismic hazard for 475 and 975 years return periods, we observe the high impact of the nonlinearity such as increased damping leading to a decrease of the site amplification. Moreover, due to nonlinearity, the soil resonance frequencies shift towards lower values, which may affect the risk estimation for some buildings. Additionally, in our simulations, strong deformation is induced in some sandy layers due to the rapid build-up of the pore pressure, with a high possibility of liquefaction. However, the variability between tested sites is significant, indicating that nonlinear site response is highly site-specific, and hence, a reliable characterization of the soil profile is crucial. Even though we observe some similarities between sites of the same soil class and characterized by similar properties e.g. water table depth, the correlations are highly dispersed. Furthermore, we explore the uncertainty and sensitivity due to the model parameters and the input ground motions.

The current work concentrates on verifying the results using strong-motion recordings with nonlinear observations, which are currently lacking in Switzerland. The indirect comparisons with empirical data from Japanese sites show similar trends and values. For direct validation, we aim to apply the procedure using the CPT-calibrated soil models from Wellington (New Zealand), where nonlinearity was observed during the 2016 M_w 7.8 Kaikōura earthquake.

This study was part of the Horizon 2020 ITN-funded URBASIS-EU and the ENSI-“Seismological research for Swiss nuclear facilities” project.

Janusz P, Bergamo P, Bonilla LF, et al (2024) Multistep procedure for estimating non-linear soil response in low seismicity areas—a case study of Lucerne, Switzerland. GJI, 239:1133–1154. https://doi.org/10.1093/gji/ggae324