Site effects are one of the most predictable factors of destructive earthquake ground motion but results depend on the type of model chosen. We compare simulations of ground motion for a 3D model of the Mygdonian basin in northern Greece (Euroseistest) using different approximation for this basin. Site effects predicted using simple 1D models at many points inside the basin are compared to site effects predicted using four different 2D cross sections across the basin and with results for a full 3D simulation. Surface topography was neglected but anelastic attenuation was included in the simulations. We show that lateral heterogeneity may increase ground motion amplification by 100 %. Larger amplification is distributed in a wide frequency range, and amplification may occur at frequencies different from the expected resonant frequencies for the soil column. In contrast, on a different cross section, smaller conversion of incident energy into surface waves and larger dispersion leads to similar amplitudes of ground motion for 2D and 1D models. In general, results from 2D simulations are similar to those from a complete 3D model. 2D models may overestimate local surface wave amplitudes, especially when the boundaries of the basin are oblique to the selected cross section. However, the differences between 2D and 3D site effects are small, especially in regard of the difficulties and uncertainties associated to building a reliable 3D model for a large basin.
Read full abstract