About the efficiency of numerical 1-D and 2-D modelling of site effects inbasin structures

In the present study we compare results obtained from experimental estimate s of local site amplification effects with those from numerical modelling u sing four different techniques. We benefit from an extremely precise knowle dge of the near-surface structure and experimental estimates of the local a mplification factors which are determined from seismic weak-motion data rec orded by a dense array across a sedimentary basin at a European test-site i n Northern Greece. The possibilities and limitations of the different model ling techniques (a 1-D technique, and three 2-D techniques) to model the ef fects of local amplification effects are tested. Amplifications calculated by the numerical techniques are only qualitatively compared with observed d ata from experimental studies in the time domain and directly in the freque ncy domain. As a result we conclude that, in the case of a complex subsurface geometry, 1-D modelling underestimates the amplification patterns in terms of absolu te amplification level, and cannot correctly account for resonant frequenci es, at least for modes higher than the fundamental mode. If a more realisti c incident wave field than just a plane wave is taken into account, 2-D mod elling reveals the fundamental frequency and shows adequate amplifications not only at the fundamental frequency but also at higher frequencies. The g eneral trend of the amplifications at the lowest frequencies is well determ ined by the 2-D numerical modelling, and can therefore supply information f or seismic risk analysis.