Evidence of complex site effects and soil non-linearity numerically estimated by 2D vs 1D seismic response analyses in the city of Xanthi

• Equivalent-linear (1D & 2D) and non-linear (2D) site response analysis of Xanthi.
• Site response seems to be dominated primarily by 1D resonance phenomena.
• Non-linear approach produces lower amplification levels than 1D & 2D equivalent-linear.
• Theoretically calculated response spectra exceed EC8 seismic code provisions.

This study discusses the effects of local site conditions on earthquake ground motion for city of Xanthi, North-Eastern Greece, focusing on the influence of complex site effects and soil non-linearity. Although city of Xanthi is characterized as a low seismicity area, documented strong earthquake events of past centuries indicate the necessity of an appropriate estimation of ground surface motion of the region. Therefore, a typical 2D cross-section along the city following an E-W direction is constructed, based on the synthesis of available geological, geotechnical and geophysical data of the broader area. 1D and 2D numerical models are generated utilizing the Finite Difference Method and site response is investigated for different seismic scenarios. Wave amplification is captured as a result of 1D resonance phenomena integrated with potential 2D complex wave effects, due to lateral propagation of the locally generated at the discontinuities surface waves. The additional, relatively to 1D, 2D amplification or de-amplification of ground shaking, in terms of its amplitude and its spectral content is quantified. The impact of soil deposits non-linearity on amplification level, as well as, on generated 2D wave fields is considered by implementing both equivalent-linear and non-linear approaches. The results reveal that, across the city, spatial variation of ground surface motion is obvious, attributed primarily to the variation of subsurface soil profiles. 2D wave effects have been generated signifying that, at particular site locations, ground motion is additionally amplified or de-amplified with respect to 1D response. However, 2D versus 1D differentiation has been calculated of ±20% in terms of PGA and spectral accelerations up to 1.0 s, in the majority of the examined sites, implying that site response even though, influenced by 2D resonance phenomena, could be sufficiently captured by an 1D approach. Moreover, it is shown that if soil non-linearity is properly accounted for, it may play an important role on the estimated ground response, since non-linear approach seems to produce lower amplification levels with regard to 1D and 2D equivalent-linear approaches. As a general trend across the investigated site, for the entity of the various scenarios, the numerically calculated response spectra exceed EC8 seismic code provisions.