In addition to rhythmic slug-driven Strombolian activity, Stromboli volcano occasionally produces discrete explosive paroxysms (2 per year on average for the most frequent ones) that constitute a major hazard and whose origin remains poorly elucidated. Partial extrusion of the volatile-rich feeding basalt as aphyric pumice during these events has led to consider their triggering by the fast ascent of primitive magma blobs from possibly great depth. Here I examine and discuss the alternative hypothesis that most of the paroxysms could be triggered and driven by the fast upraise of CO2-rich gas pockets generated by bubble foam growth and collapse in the sub-volcano plumbing system. Data for the SO2 and CO2 crater plume emissions are used to show that Stromboli's feeding magma may originally contain as much as 2 wt.% of carbon dioxide and early coexists with an abundant CO2-rich gas phase with high CO2/SO2 molar ratio (≥ 60 at 10 km depth below the vents, compared to ∼ 7 in time-averaged crater emissions). Pressure-related modelling indicates that the time-averaged crater gas composition and output are well accounted for by closed system decompression of the basalt–gas mixture until the volcano–crust interface (∼ 3 km depth), followed by open degassing and crystallization in the volcano conduits. However, both the low viscosity and high vesicularity of the basaltic magma permit bubble segregation and bubble foam growth at deep sill-like feeder discontinuities and at shallower physical boundaries (such as the volcano–crust interface) where the gas-rich aphyric basalt interacts with the unerupted crystal-rich and viscous magma drained back from the volcano conduits. Gas pressure build-up and bubble foam collapse at these boundaries will intermittently trigger the sudden upraise of CO2-rich gas blobs that constitute the main driving force of the paroxysms. Deeper-sourced gas blobs, driving the most powerful explosions, will be the richest in CO2 and have highest CO2/SO2 ratios. This mechanism is shown to account well for the dynamic, seismic and petrologic features of Stromboli's paroxysms and, hence, to provide a potential alternative interpretation for their genesis and their forecasting. Enhanced bubble foam leakage prior to a paroxysm, or foam emptying in several steps, should lead indeed to precursory upstream of CO2-rich gas and increasing CO2/SO2 ratio in crater plume emissions. The recent detection of such signals prior to two explosions in December 2006 and March 2007 strongly supports this expectation and the model proposed in this study.
Times Cited: 5