The 5 April 2003 vulcanian explosion at Stromboli volcano (Southern Italy): analysis of seismic data. Luca D’Auria, Flora Giudicepietro, Marcello Martini and Rosario Peluso Istituto Nazionale di Geofisica e Vulcanologia - Osservatorio Vesuviano March 13, 2006 Abstract On 5 April 2003 at 07:13 GMT (09:13 local time) a violent vulcanian explosion occurred at Stromboli volcano. At the time of the event an erup- tive crisis was ongoing at the volcano with a lava flow outpouring along the Sciara del Fuoco flank. The seismic signals related to the event were recorded by 8 permanent broadband stations and gives informations about the eruption kinematics. An ultra-long-period signal, that we intepret as the effect of the ground tilt on the broadband sensors, starts about 4 min before and terminates about 1 min after the explosion. On the basis of the radial pattern of tilt directions we conclude that this signal is the effect of the deformation of the volcanic edifice, due to the rising of a batch of magma, its ejection and the magma column readjustment. About 1 min before the explosion we observe an high frequency signal that we believe to be related to the vesiculation of the rising batch of gas-rich magma. At 07:13:35 GMT a powerful very-long-period signal, marking the onset of the explosive fragmentation, is recorded. This is confirmed by a blast wave following few seconds later. The remaining seismic signal (more than 3 min), shows an higher frequency content being related only to the fall of ballistic ejecta and to landslides along Sciara del Fuoco. In conclusion we propose the implementation of an early warning system for the short-term forecast of such explosions, based on the real-time automatic detection of the tilt signals preceeding such events. 1 The 2002-2003 eruption and the 5 April ex- plosion The 2002-2003 eruption at Stromboli, marked a period of anomalous volcanic activity. The typical, persistent, moderate explosive activity at the summit craters, stopped when a lava flow started pouring out of an eruptive fissure along the Sciara del Fuoco flank [4, ]. The first occurrence of an anomalous explosive event was reported on 3 April. This event, that ejected large blocks outside the crater walls [3, ], was accompanied by a seismic signal (Fig.5) having a waveform different from the typical activity [5, ]. 1