{The Kozani earthquake (Ms = 6.6) of 13 May 1995 is the strongest event of the decade in Greece and occurred in a region of low seismic activity. Using regional data and the strong-motion record at the Kozani station, we relocate the mainshock at 40.183° N and 21.660° E, beneath the Vourinos massif at a depth of 14.2 km. We also compute a focal mechanism by body-waveform modeling at teleseismic distance, which confirms a normal mechanism. The most likely plane strikes 240° ± 1° N and dips 40° ± 1° N with a centroid depth of 11 ± 1 km. Modeling of the strong-motion record at Kozani confirms that nucleation started at the eastern termination of the bottom of the fault.Six days after the mainshock, we installed a network of 40 portable seismological stations for one week around the epicentral region. Several thousand aftershocks were recorded, among which we locate 622 with a precision better than 1 km. We compute 181 focal mechanisms that mostly show normal faulting. The aftershock seismicity is restricted between 5 and 15 km depth and defines a plane dipping north at an angle of about 35°, consistent with the mainshock mechanism. Seismic activity with the same pattern of normal fault mechanisms is also seen on an antithetic fault connected to the main one at 12 km depth, which cuts the ground surface north of the Vourinos ophiolite massif in the Siatista valley. These results suggest two possibilities for the active fault plane; either it is the Deskati fault that is flat and dips with a constant angle, and therefore the surface breaks are secondary features, or, more likely, it is the Paleohori fault that is new, of listric shape, and located ahead of the Deskati fault, which was not active during the earthquake.}

We present the results of a multidisciplinary study of the Ms = 6.2, 1995, June 15, Aigion earthquake (Gulf of Corinth, Greece). In order to constrain the rupture geometry, we used all available data from seismology (local, regional and teleseismic records of the mainshock and of aftershocks), geodesy (GPS and SAR interferometry), and tectonics. Part of these data were obtained during a postseismic field study consisting of the surveying of 24 GPS points, the temporary installation of 20 digital seismometers, and a detailed field investigation for surface fault break. The Aigion fault was the only fault onland which showed detectable breaks (< 4 cm). We relocated the mainshock hypocenter at 10 km in depth, 38 ° 21.7 ′ N, 22 ° 12.0 ′ E, about 15 km NNE to the damaged city of Aigion. The modeling of teleseismic P and SH waves provides a seismic moment Mo = 3.4 1018 N.m, a well constrained focal mechanism (strike 277 °, dip 33 °, rake − 77°), at a centroidal depth of 7.2 km, consistent with the NEIC and the revised Harvard determinations. It thus involved almost pure normal faulting in agreement with the tectonics of the Gulf. The horizontal GPS displacements corrected for the opening of the gulf (1.5 cm/year) show a well-resolved 7 cm northward motion above the hypocenter, which eliminates the possibility of a steep, south-dipping fault plane. Fitting the S-wave polarization at SERG, 10 km from the epicenter, with a 33° northward dipping plane implies a hypocentral depth greater than 10 km. The north dipping fault plane provides a poor fit to the GPS data at the southern points when a homogeneous elastic half-space is considered: the best fit geodetic model is obtained for a fault shallower by 2 km, assuming the same dip. We show with a two-dimensional model that this depth difference is probably due to the distorting effect of the shallow, low-rigidity sediments of the gulf and of its edges. The best-fit fault model, with dimensions 9 km E–W and 15 km along dip, and a 0.87 m uniform slip, fits InSAR data covering the time of the earthquake. The fault is located about 10 km east-northeast to the Aigion fault, whose surface breaks thus appears as secondary features. The rupture lasted 4 to 5 s, propagating southward and upward on a fault probably outcropping offshore, near the southern edge of the gulf. In the shallowest 4 km, the slip – if any – has not exceeded about 30 cm. This geometry implies a large directivity effect in Aigion, in agreement with the accelerogram aig which shows a short duration (2 s) and a large amplitude (0.5 g) of the direct S acceleration. This unusual low-angle normal faulting may have been favoured by a low-friction, high pore pressure fault zone, or by a rotation of the stress directions due to the possible dip towards the south of the brittle-ductile transition zone. This fault cannot be responsible for the long term topography of the rift, which is controlled by larger normal faults with larger dip angles, implying either a seldom, or a more recently started activity of such low angle faults in the central part of the rift.