Assimilating Volcanic SO2 Satellite Data in the Copernicus Atmosphere Monitoring Service Global Data Assimilation System

Kurzfassung

In the middle of September 2014 several European countries experienced high concentrations of sulphur dioxide at ground level. Due to strong European efforts over the last decades to reduce SO2 emissions, high concentrations of SO2 are now quite rare in Western Europe except in specific areas affected by industrial or shipping emissions. French in-situ air quality stations observed high values of SO2, especially along the northwestern coast. However, the hypothesis that these high values could be linked to ship emissions trapped in the lower atmosphere appeared unlikely because they were exceptionally high and observations in the United Kingdom, the Netherlands and Germany also showed high concentrations between 21 and 25 September. The precursor Copernicus Atmosphere Monitoring Service near-real-time forecasting system, as provided by the Monitoring Atmospheric Composition and Climate (MACC) project, explained the situation thanks to its use of satellite observations to constrain the model forecasts. The OMI satellite instrument observed concentrations of volcanic SO2, emitted by the Icelandic Bardarbunga volcano, and these observations were assimilated by the MACC system. The subsequent 5-day forecast then captured the transport of this plume of volcanic SO2 southward. While this case illustrated successful use of OMI satellite data in the data assimilation system, it also showed some of its weaknesses. Only one OMI observation was used for the specific forecast resulting in a plume with concentrations that were too low. OMI observations are only assimilated above a certain threshold value to limit the impact of noisy observations on the analysis, which significantly limits the spatial coverage. GOME-2 also provides SO2 retrievals in near real-time, but these suffer from even higher noise than the OMI observations, which limits the use of these retrievals in an operational assimilation system. Since the particular event in September, a specific volcanic SO2 flag was introduced for the GOME-2 SO2 retrievals allowing a much more optimal use of the observations in the data assimilation. We will present both the original results from the pre-operational OMI assimilation and re-runs with the new GOME-2 SO2 data. Significant improvement of the analysis and forecast will be shown. We will also show current work on and ideas for future improvements of both the modelling and data assimilation aspects of volcanic SO2.