Investigating reference station conditions in Northeastern Greece

The assumption of reference station conditions is an important one, yet is often taken for granted, especially at seismological stations or those presumably lying on rock. Stations are not always installed on true bedrock or in ideal free-field conditions, and rock formations do not always exhibit negligible amplification. In Greece, there have been no studies of site effects conducted systematically for the ensemble of stations, and relatively few ad hoc field surveys so far to characterise them. We investigate stations belonging to the various broadband and accelerometric networks in Greece whose data are publicly available. We focus on stations from Northeastern Greece, in the regions of Macedonia and Thrace, as they can be of particular interest for the study of background seismicity and seismic hazard that has begun in the Kavala-Prinos area in the framework of EU project COREu. We take the following steps: 1. We first compile all publicly available station metadata, seeking information from external sources, i.e., geology, topography, housing, etc. 2. We then analyse geological maps and provide a description of geological unit and geological age and combine this information with available observations by the operators. 3. Finally, we collect waveforms available from the past decade and perform a detailed analysis to estimate the local site response per station. To do this, the dataset first goes through quality control by visually inspecting and meticulously processing the data on a waveform-specific basis, both in the time and frequency domain. Single-station amplification functions are then estimated from the dataset using the horizontal-to-vertical spectral ratio (HVSR) technique. Since a good reference station should have low and flat amplification without any significant azimuthal variations, we not only assess the features of the mean HVSR per station but also its directional sensitivity, which serves as a means to identify departure from the 1D assumption. We use clustering techniques to group stations with different response characteristics, and finally combine this data-derived characterisation with the previously compiled metadata to evaluate the stations’ overall capacity as reference sites. This work intends to add value and help decrease epistemic uncertainties for future applications within the COREU project in the Kavala-Prinos area that involve seismic monitoring and hazard assessment.