The East Aegean Sea strong earthquake sequence of October–November 2005: lessons learned for earthquake prediction from foreshocks

Abstract. The seismic sequence of October–November 2005 in the Samos area, East Aegean Sea, was studied with the aim to show how it is possible to establish criteria for (a) the rapid recognition of both the ongoing foreshock activity and the mainshock, and (b) the rapid discrimination between the foreshock and aftershock phases of activity. It has been shown that before the mainshock of 20 October 2005, foreshock activity is not recognizable in the standard earthquake catalogue. However, a detailed examination of the records in the SMG station, which is the closest to the activated area, revealed that hundreds of small shocks not listed in the standard catalogue were recorded in the time interval from 12 October 2005 up to 21 November 2005. The production of reliable relations between seismic signal duration and duration magnitude for earthquakes included in the standard catalogue, made it possible to use signal durations in SMG records and to determine duration magnitudes for 2054 small shocks not included in the standard catalogue. In this way a new catalogue with magnitude determination for 3027 events was obtained while the standard catalogue contains 1025 events. At least 55 of them occurred from 12 October 2005 up to the occurrence of the two strong foreshocks of 17 October 2005. This implies that foreshock activity developed a few days before the strong shocks of 17 October 2005 but it escaped recognition by the routine procedure of seismic analysis. The onset of the foreshock phase of activity is recognizable by the significant increase of the mean seismicity rate which increased exponentially with time. According to the least-squares approach the b-value of the magnitude-frequency relation dropped significantly during the foreshock activity with respect to the b-value prevailing in the declustered background seismicity. However, the maximum likelihood approach does not indicate such a drop of b. The b-value found for the aftershocks that followed the strong shock of 20 October 2005 is significantly higher than in foreshocks. The significant aftershock-foreshock difference in b-value is valid not only if the entire aftershock sequence is considered but also if only the segment of aftershocks that occurred within the first 24-h or the first 48-h after the mainshock of 20 October 2005 are taken into account. This difference in b-value should be examined further in other foreshock-aftershock sequences because it could be used as a diagnostic of the mainshock occurrence within a few hours after its generation.