The December 2002-2003 effusive event at Stromboli volcano, Italy: Insights into the shallow plumbing system by petrochemical studies.

On 28 December 2002, the persistent Strombolian activity at Stromboli was interrupted by the sudden onset of lava emissiononto the Sciara del Fuoco slope, a horseshoe-shaped depression on the NW flank of the volcano. The effusive episode went on until22 July 2003 and produced a cumulative volume of lavas of 11×106 m3 ranking the event as the largest occurred in the past30 years. The eruptive vents were mainly located in the NE sector of the Sciara del Fuoco depression, at an elevation of 550-600 ma.s.l. On 30 December, the eastern portion of the Sciara del Fuoco collapsed producing a tsunamigenic landslide. On 5 April 2003,a paroxysmal eruption occurred at the summit craters during which crystal-poor pumiceous products were emitted. The paroxysmdid not interrupt the lava emission. The Strombolian activity at the summit craters gradually resumed starting from March 2003 andfully recovered by the end of July.Periodic sampling of the lava at the active vents was carried out during the entire effusive event. All the analysed samples areshoshonitic basalts (SiO2 48.5-50.4 wt.%; K2O 2.1-2.4 wt.%) in the range of composition observed in the products erupted duringthe past 20 years. They bear about 50 vol.% zoned crystals of plagioclase An90-60, diopside-augite and olivine Fo70-73 in acompositionally homogeneous shoshonitic groundmass. Bulk rock major and trace element contents measured with differentmethodologies at different laboratories show only minor variations. Sr and Nd isotope ratios are close to those of the crystal-richscoriae erupted in the previous years.Despite of small compositional variations related to the emptying of the zoned topmost portion of the conduit, the data showthat the switch from Strombolian explosive to effusive activity is not associated with changes of the textures and composition of theerupted products. Slight but somehow systematic variation of trace elements and isotope ratios between products erupted beforeand after the 5 April eruption are likely accounted by limited mixing between the fresh, volatile-rich, crystal-poor, magma eruptedas pumice during the paroxysm, and the volatile-poor, crystal-rich magma feeding the lava flow. The uniform composition of theerupted lava indicates the presence of a large volume of well-mixed, crystal-rich, homogeneous magma residing in the shallowplumbing system of the volcano.Two possible trigger mechanism of the effusive event are here proposed: (i) a discrete input of fresh magma into the lower partof the shallow magmatic system occurred some months before the eruption and was followed by crystallisation, degassing and mixing with the crystal-rich shallow magma which re-homogenised the system. These processes eventually led to the rise of themagma at a higher level and failure of the conduit walls. (ii) The onset of the effusion may represent the consequence of a gradualrise of the magma level in the conduits occurred in the past two decades. The process of progressive refilling, initiated after the1985 effusive eruption, ultimately culminated on 28 December 2002 with the failure of the conduit wall and the opening of vents100-150 m below the summit craters.