Repeatable process for seismic microzonation using 1-D site-specific response spectra assessment approaches. Application to the city of Nice, France

摘要

We are testing three different approaches to assess the site effects based on 1D soil column data. The first method follows the Eurocodes 8 (EC8) approach based on the determination of Vs(30) soil classes, the second one is based on an empirical relationship between Vs(30) and the fundamental resonance frequency of the site (f(0)) with a response spectra amplification function proposed by Cadet a al. (2013) and the last one is the application of a 1-D equivalent linear simulation. Then, we propose a repeatable methodology to define the zonation and the associated surface response spectrum from individual results. This methodology uses a shape recognition algorithm to build homogeneous response spectra zones and the definition of a specific EC8-like response spectrum adjusted to the distribution of the response spectra in each zone. This has been applied to downtown Nice, a city located on the French Riviera that faces the highest seismic hazard in the French mainland area. The city is built on two different sedimentary valleys that are prone to lithological site effects. The ground acceleration around 1.8 Hz can be 20 times higher in the valleys than the one recorded on the surrounding Jurassic rock outcrop. We defined a 3D geotechnical model, composed of 9 homogenous soil layers. The surface elastic response spectra in pseudo-acceleration obtained with the three methods are compared on the whole city. The results indicate that the amplification functions methodology proposed by Cadet et al. (2012b) provides results closer to the numerical simulations compared to EC8. In addition, in five sites of the French Permanent Accelerometric Network (RAP/RESIF) located in the valley of Nice, the site to reference response spectra ratio from the numerical simulations and the spectra amplification function proposed by Cadet et al. (2012b) were compared to earthquake recordings data. Although the spectra amplification functions are closer to the observations, both evaluations underestimate the amplification as they could not capture complex site geometry. Using the repeatable zonation process on the response spectra from the amplification function, three different zones are defined. The associated specific response spectra are significantly different from the recommended EC8 response spectra.