Assessing the reliability of the EMSC testimonies database and its potential use for the generation of near-real-time ShakeMaps in the Aegean area

We investigate the possibility of combined interpretation of macroseismic and strong-motion data for recent large earthquakes in the Aegean area. We employ macroseismic information derived from EMSC testimonies, as well as strong-motion information extracted from online sources provided by two Greek institutes (ITSAK and GEIN-NOA). The EMSC testimonies database (https://www.seismicportal.eu/testimonies-ws/) is a widely used inventory for the damage distribution of significant earthquakes. The collected data were first compared with the predicted macroseismic intensities using the empirical relation of Papazachos and Papaioannou (J Seismol 1:181–201, 1997) While the correlation between the observed and modeled data was found to be satisfactory, a systematic bias is evident for very high and very low values intensities derived from the reported EMSC testimonies. A Monte Carlo simulation approach was employed to identify the source of this bias, suggesting that it is a result of the large scatter of the EMSC data and the limits of the macroseismic scale used. To minimize this effect, a spatial grouping and smoothing approach was adopted for the EMSC dataset, resulting in significantly improved correlations with the available independent strong motion estimates, such as PGA and PGV. Using this correlation, we demonstrate through several examples that it is possible to reconstruct the main features of the damage pattern for strong earthquakes in the Aegean. This is achieved by jointly analyzing rapidly crowdsourced EMSC data and strong motion information, after appropriate processing of the raw macroseismic dataset.