The transport behaviour of the Stromboli’s volcanic plume inferred from combined scanning-DOAS and UV Camera observations

Abstract

The volcanic SO2 flux has long been known as an especially useful tracer of the rate of magma supply into the shallower portions of volcano plumbing systems, and as such is central to volcano monitoring. SO2 fluxes from active volcanos are widely measured globally with networks of scanning spectrometers coupled with the DOAS technique and UV camera systems. However, observations have traditionally been challenged by a variety of technical and methodological caveats and limitations, to overcome which an intercomparison between, and eventually integration of, different observational techniques are required. The ability of scanning-DOAS spectrometers to scan the plume from distance and from different viewing directions, allows to fully capture the “bulk” gas plume, utilising the robustness of DOAS technique to significantly reduce radiative transfer and light dilution issues. Disadvantages exist, however, especially arising from assumptions that must be made on geometry and height of the plume and, most importantly, using wind speed as a proxy for the real plume velocity. UV Cameras, in contrast to scanning spectrometers, are typically deployed near the active vents and measure the two-dimensional SO2 distribution in the plume with a greater spatial and temporal resolution, applying optical flow algorithms to the acquired images and estimating the plume vertical velocity profile at source. However, the proximity to the active vents often causes the plume to be only partially imaged by UV Cameras, due to morphological complexity of volcano summits or unfavourable wind directions causing underestimation of the measured SO2 flux. Furthermore, targeting the proximal and most optical opaque portion of volcanic plumes cause UV Cameras measurements to be more prone to errors related to radiative transfer issues. All remote sensing techniques aimed to measure SO2 fluxes can effectively ‘measure’ only gas concentrations and retrieve a gas flux by combining these measurements with estimates of the gas movement speed. To measure a gas flux away from the source (as scanning-DOAS networks do), a conservative flux model is traditionally assumed, which considers a continuous and stationary-type source. Based on this assumption, it is possible to measure a flux in distal analysis sections, by relating it to the distal-measured gas concentration, as the flux is retrieved as the product of the gas concentration with its velocity. Furthermore, in order to maintain a gas flux equal to that at source, it is assumed that gas concentrations are linearly scaled with speed, an aspect that has been little considered in the past. Here, we compare ~9 years (2014 to 2022) of SO2 flux records at Stromboli obtained through (i) a near-vent UV Camera system and (ii) a network of DOAS spectrometers scanning the distal bulk plume. Our combined near-vent and distal observations offer a unique opportunity to test models of gas plume dilution in the atmosphere and the local emissive source behaviour. We find a large (133 t/d on average) systematic offset between the SO2 flux time-series streamed by the two observational techniques. Using empirical evidence, we propose this SO2 flux mismatch to be dominantly caused by the non-continuous emission behaviour of the Stromboli's volcanic plume. If gas at Stromboli is emitted from source in discrete parcels (“puffs”) rather than continuously, then the extent of plume dilution upon downwind atmospheric transport does not scale linearly with wind speed. Hence the SO2 flux cannot be accurately retrieved from combination of wind velocity and SO2 integrated column densities. If our hypothesis of non-continuous plume behaviour is not unique to Stromboli, but generally valid, then SO2 fluxes can have been overestimated at volcanoes globally. Our work suggests that a combined approach, integrating (a) gas velocities derived from optical flow algorithms implemented into UV Camera