36th General Assembly of the European Seismological Commission (ESC 2018), Valletta, Malta, 2 - 07 September 2018, pp.449
Modeling of near-surface attenuation has significance in a wide range of seismological applications such as site characterization, sitespecific hazard assessment for critical facilities and ground motion predictions. Kappa (k) factor is often used to represent the rate of the attenuation in spectral amplitudes at high frequencies (f > 10 Hz) and it is also one of the essential parameters for stochastic strong ground motion simulation method. Although the origin and physical components of k is still under debate, it is considered to be a combination of site and path effects. Moreover, a source component could also contribute to variability in kappa estimates. Zero-distance kappa value (k0) is computed as an indicator of the spectral attenuation due to top soil layers, eliminating the path effects. In this study, we used the recentlycomplied strong ground motion database of Turkey which consists of records from mostly small and moderate-size earthquakes with a smaller number of records from large events. k factors are manually computed from the S-wave portion of both horizontal and vertical components for each record as we observed variations in k values for different components. Next, magnitude and epicentral distance dependence of the k values are studied. We use multivariate linear regression analyses to describe the relationships between the studied kappa dataset and selected independent variables such as the site class, distance and magnitude. Kappa estimates reveal that there is a significant scattering at the majority of the investigated sites. The scattering of these initial kappa estimates could be attributed to the influence of the sourceand wave propagation effects associated with the complex tectonic structure of the region.