New insights into volcanic processes at Stromboli from Cerberus, a remote-controlled open-path FTIR scanner system

Abstract The ordinary, low intensity activity of Stromboli volcano is sporadically interrupted by more energetic events termed, depending on their intensity, “major explosions” and “paroxysms”. These short-lived energetic episodes represent a potential risk to visitors to the highly accessible summit of Stromboli. Observations made at Stromboli over the last decade have shown that the composition of gas emitted from the summit craters may change prior to such explosions, allowing the possibility that such changes may be used to forecast these potentially dangerous events. In 2008 we installed a novel, remote-controlled, open-path FTIR scanning system called Cerberus at the summit of Stromboli, with the objective of measuring gas compositions from individual vents within the summit crater terrace of the volcano with high temporal resolution and for extended periods. In this work we report the first results from the Cerberus system, collected in August–September 2009, November 2009 and May–June 2010. We find significant, fairly consistent intra-crater variability for CO 2 /SO 2 and H 2 O/CO 2 ratios, and relatively homogeneous SO 2 /HCl ratios. In general, the southwest crater is richest in CO 2 , and the northeast crater poorest, while the central crater is richest in H 2 O. It thus appears that during the measurement period the southwest crater had somewhat more direct connection to a primary, deep degassing system while the central and northeast craters reflect a slightly more secondary degassing nature, with a supplementary, shallow H 2 O source for the central crater, probably related to puffing activity. Such water-rich emissions from the central crater can account for the lower crystal content of its eruption products, and emphasise the role of continual magma supply to the shallowest levels of Stromboli's plumbing system. Our observations of heterogeneous crater gas emissions and high H 2 O/CO 2 ratios do not agree with models of CO 2 -flushing, and we show that simple depressurisation during magma ascent to the surface is a more likely model for H 2 O loss at Stromboli. We highlight that alternative explanations other than CO 2 flushing are required to explain distributions of H 2 O and CO 2 amounts dissolved in melt inclusions. We detected fairly systematic increases in CO 2 /SO 2 ratio some weeks prior to major explosions, and some evidence of a decrease in this ratio in the days immediately preceding the explosions, with periods of low, stable CO 2 /SO 2 ratios between explosions otherwise. Our measurements, therefore, confirm the medium term (~ weeks) precursory increases previously observed with MultiGas instruments, and, in addition, reveal new short-term precursory decreases in CO 2 /SO 2 ratios immediately prior to the major explosions. Such patterns, if shown to be systematic, may be of great utility for hazard management at Stromboli's summit. Our results suggest that intra-crater CO 2 /SO 2 variability may produce short-term peaks and troughs in CO 2 /SO 2 time series measured with in-situ MultiGas instruments, due simply to variations in wind direction.