Effects of shallow and deep geology on seismic hazard estimates: a case study of pseudo-acceleration response spectra for the northwestern Balkans

Northwestern Balkans represents one of the rare regions where data on both the shallow geology (i.e., local soil conditions) and the deep geology exist for stations that recorded hundreds of strong motion records. The strong motion database used in this study consists of 203 strong motion accelerograms (each with three orthogonal translational components), recorded in former Yugoslavia in the period 1976–1983, from 108 contributing earthquakes. In this paper, the results of a series of regression analyses are presented where empirical equations for scaling pseudo-acceleration response spectra were developed on the basis of 5 subsets of data and using three prediction models. Results of the regression analyses show that for ground motion in the horizontal direction, both the shallow and deep geology site conditions have to be taken into account or else the spectral estimates might be considerably biased. Results show that the shallow geology influences spectral amplitudes the most in the short period range and has much lesser effects for larger periods, while the influence of the deep geology is expressed in a wider range of intermediate to long periods. Results also show that if the prediction model that considers solely the shallow geology effects is used, the spectral peaks that have been related to the deep geology effects will completely vanish for larger source-to-site distances, while in the case when solely the deep geology is considered, the peaks related to the shallow geology will not be visible for any distance. As for the amplitudes in the vertical direction, although both the shallow and the deep geology effects are less expressed than in the horizontal one, they still cannot be neglected—the deep geology effects are visible for a wider range of vibration periods, while the shallow geology has some effects only for periods smaller than ~0.3 s.