Abstract
In this study, the site effects on seismic ground motion in the city of Thessaloniki (Greece) are estimated by applying experimental methods on acceleration data and theoretical modeling. The technique of standard spectral ratio (SSR) is applied to a reference station located on rock, while the horizontal-to-vertical spectral ratio (HVSR) technique is applied to earthquake records (entire record length including P and S waves) as well as on noise recordings. In addition, the SSR method is also applied to the vertical components. The results from all methods are compared in terms of resonant frequencies and amplification levels. The fundamental resonant frequency is identified by all methods, while the average amplification level is generally underestimated when the HVSR technique is used. An attempt is made to correlate the site amplifications computed in this study with the observed macroseismic intensities reported at the same sites for the 20 June 1978 earthquake. A relation of the form, 
I = 
+ b · log[SSR(f)] is derived, where I is the intensity increment with respect to the intensity of the reference station and SSR(f) is the mean amplification factor obtained at each station using the SSR technique for a certain frequency band. In the numerical approach, we construct complete strong-motion synthetics using the modal summation method for the P-SV waves up to frequencies of 10 Hz. As input, four point sources are used, located at different distances and azimuths from the stations. Ratios of response spectra of the local 1D over the regional 1D synthetic seismograms are calculated. The obtained mean spectral amplifications are compared with those derived from experimental data, and the two sets are found to be consistent at most stations.