Simulated Radiative Transfer DOAS - A new method for improving volcanic SO2 emissions retrievals from ground-based UV-spectroscopic measurements of scattered solar radiation

Passive Differential Optical Absorption Spectroscopy (DOAS) has become a standard tool for measuring SO2 at volcanoes. More recently, ultra-violet (UV) cameras have also been applied to obtain 2D images of SO2-bearing plumes. Both techniques can be used to derive SO2 emission rates by measuring SO2 column densities, integrating these along the plume cross-section, and multiplying by the wind speed. Recent measurements and model studies have revealed that the dominating source of uncertainty in these techniques often originates from an inaccurate assessment of radiative transfer through the volcanic plume. The typical assumption that all detected radiation is scattered behind the volcanic plume and takes a straight path from there to the instrument is often incorrect. We recently showed that the straight path assumption can lead to column density errors of 50% or more in cases where plumes with high SO2 and aerosol concentrations are measured from several kilometers distance, or where the background atmosphere contains a large amount of scattering aerosols. Both under- and overestimation are possible depending on the atmospheric conditions and geometry during spectral acquisition. Simulated Radiative Transfer (SRT) DOAS is a new evaluation scheme that combines radiative transfer modeling with spectral analysis of passive DOAS measurements in the UV region to derive more accurate SO2 column densities than conventional DOAS retrievals, which in turn leads to considerably more accurate emission rates. A three-dimensional backward Monte Carlo radiative transfer model is used to simulate realistic light paths in and around the volcanic plume containing variable amounts of SO2 and aerosols. An inversion algorithm is then applied to derive the true SO2 column density. For fast processing of large datasets, a linearized algorithm based on lookup tables was developed and tested on a number of example datasets. In some cases, the information content of the spectral data is sufficient to allow the retrieval of additional atmospheric parameters such as the aerosol optical thickness of the volcanic plume. Thus, the SRT-DOAS approach provides a great advancement in the ability to assess and improve the accuracy of SO2 emission rates determined from UV remote sensing instruments. The impact of the new methodology is far-reaching, as other volcanic volatile emissions are often calculated by scaling remotely determined SO2 emission rates by the respective X/SO2 concentration ratios; therefore emission rates of such constituents would be equally affected.