2D equivalent linear site effect simulation: example applications to two deep valleys

In the framework of the Sismovalp European project, an equivalent linear 2D code was developed to compute the response of a valley to SH waves, using the discrete wave-number method proposed by Aki and Larner (Aki K, Larner KL (1970) J Geophys Res 75:5). To overcome the frequency upper bound limitation, the Aki and Larner’s method is combined with a one-dimensional computation using a classical multi-layer method (Aki K, Richards PG (1980) Quantitative Seismology: Theory and Methods, vols. 1 & 2. W.H. Freeman & Co, San Francisco). The so-called “Aki–Larner extended method” is associated to an iterative algorithm, as proposed by Seed and Idriss (Seed HB, Idriss IM (1969) Report No. EERC 70–10, Earthquake Research Center, University of California, Berkeley, California) which accounts for the modulus and damping degradation using a linear visco-elastic model. A comparison of the results in the linear and the equivalent linear cases, for a magnitude 6.0 earthquake, shows that the account for the equivalent linear behaviour of the soil significantly reduces the amplification level, especially at frequencies higher than the fundamental resonance frequency of the site. In the case of site effects or microzonation studies devoted to produce design spectra for engineering structures, this can have a major impact on the associated results and costs, depending on the frequency of interest for the considered structure. As a first application of the developed technique, 2D equivalent linear Aki–Larner computations are used to perform the seismic microzonation study of the upper Rhone valley, in the Visp area (Switzerland), a typical 2D alpine valley. These investigations made it possible to determine site specific spectra, associated with different zones, to be used instead of the code spectra that do not take into account the local 2D amplification.