Insights from  CO2/SO2 gas molar ratio variations and distribution at Stromboli volcano

Gas-sensor-based monitoring stations (aka MultiGAS) near degassing volcanic vents sensibly increased the sampling rate of gas composition measurements. In particular, the CO2/SO2 gas molar ratio was demonstrated to be a good indicator of magma depth thanks to the CO2 and SO2 contrasting solubilities in magma. Numerous high CO2/SO2 gas transients recorded days before an effusive/explosive eruption have been reported in the literature (e.g. Etna, Villarrica, Masaya, Poas). The successful detection of a precursor gas signal in these volcanoes has been favoured by the presence of open or highly permeable closed-conduit (as in the case of Poas) and the instrument’s vicinity to a high-flux degassing vent (high signal/noise ratio). Volcanic gas monitoring in Stromboli represents a particular case. Stromboli is characterised by degassing from multiple vents, which exhibit simultaneous different molar gas ratios during quiescent degassing. The gas emissions during transient strombolian explosions are also different, with higher CO2 contents. Stromboli occasionally exhibits major explosions and paroxysms of greater energy which have often shown a substantial variation of the gas bulk composition weeks before their onset. Finally, the safest sites for a monitoring station in Stromboli are located hundreds of meters from the crater terrace, implying that the detected gas concentrations depend on the wind direction and speed. All these features make the analysis and interpretation of the volcanic CO2/SO2 challenging. This work presents a 2-year-long CO2/SO2 time series recorded at the Stromboli’s summit. We discuss different types of analysis that can be performed to enhance the variations before major and paroxysmal eruptions. We apply an iterative algorithm to estimate the time, number and magnitude of abrupt changes within the CO2/SO2 time series and discuss the origin of such variations. We use an algorithm for finite mixture models on the whole dataset to characterise the source of different gas phases. Finally, we compare the measured CO2 and SO2 concentrations with the wind parameters obtained for the area of Stromboli from the ERA5 reanalysis dataset. Hence, we determine the best conditions for gas ratio measurements and how meteorological conditions may affect the measurements' quality.