Extreme earthquake and earthquake perceptibility study in Greece and its surrounding area

The sample interval for the selection of extreme magnitudes plays an important part in the quality of Gumbel model fitting. A short sample interval can produce many observations, which is helpful in obtaining a reliably fitting model. However a short sample interval can bring many dummy ``observations'', a condition which adversely biases the fitting. The short sample interval also increases the chance to introduce non-independent observations as well, which violates a basic requirement of the Gumbel model. On the other hand, a large time interval not only reduces the number of observations, but also enlarges the observation error. Thus, for Greece, the most suitable parameters of the third Gumbel extreme model are obtained by using a sample interval which produces minimum error. In consideration of the reliability of the seismic data, earthquakes with magnitude M ≥ 5.5 in Greece and its surrounding region after 1900 are used mainly in the present paper. In order to obtain well resolved contour maps with smooth changes a 2°× 2° cell with half-degree overlap strategy was used to scan the region. The most expected largest earthquake for the next fifty, one hundred and two hundred years are estimated for each cell. Likewise, the events with magnitude at a probability of 90\% of non-exceedance over the next fifty, one hundred and two hundred years are estimated for each cell. In parallel to this procedure we also analyze the 67 shallow seismic zones outlined by Papazachos and his colleagues and detail individual zone results where these are obtained. The most perceptible earthquake magnitude for the range of intensities I = {VI}, VII and VIII are also calculated. All results show that the areas around the Hellenic Arc and the Cephalonia Transform Fault for Greece have comparatively high frequency of destructive earthquakes accompanied by a high occurrence probability of moderate earthquakes (M ≥ 5.5).