Novel retrieval of volcanic SO2 abundance from ultraviolet spectra

The recent development of fixed networks of scanning ultraviolet spectrometers for automatic determination of volcanic SO2 fluxes has created tremendous opportunities for monitoring volcanoes but has brought new challenges in processing (and interpreting) the copious data flow they produce. A particular difficulty in standard implantation of differential optical absorption (DOAS) methods is the requirement for a clear-sky (plume-free) background spectrum. Our experience after four years of measurements with two UV scanner networks on Etna and Stromboli shows that wide plumes are frequently observed, precluding simple selection of clear-sky spectra. We have therefore developed a retrieval approach based on simulation of the background spectrum. We describe the method here and tune it empirically by collecting clear, zenith sky spectra using calibration cells containing known amounts of SO2. We then test the performance of this optimised retrieval using clear-sky spectra collected with the same calibration cells but for variable scan angles, time of day, and season (through the course of 1 year), finding acceptable results (~ 12% error) for SO2 column amounts. We further illustrate the analytical approach using spectra recorded at Mt. Etna during its July 2006 eruption. We demonstrate the reliability of the method for tracking volcano dynamics on different time scales, and suggest it is widely suited to automated SO2-plume monitoring.