|Title||Volcanic sulphur dioxide plume forecasts based on UV-satellite retrievals for the 2011 Grimsvotn and the 2010 Eyjafjallajokull eruption|
|Year of Publication||2012|
|Authors||Flemming, J, Inness, A|
|Secondary Title||Technical Memorandum|
The sulphur dioxide plumes released by the eruptions of the Icelandic volcanoes Eyjafjallajökull in May 2010 and Grímsvötn in May 2011 were studied using observations from GOME-2, OMI and SCIAMACHY and modelled with the Integrated Forecasting System (IFS) of ECMWF. The satellite retrievals of SO2 total columns (TCSO2) were (i) used to estimate source strength and injection height of the two eruptions and (ii) assimilated by the IFS to obtain initial conditions for subsequent forecasts. The observed plume location agreed well between the retrievals from the different instruments but only GOME-2 observations provided complete spatial coverage. The source strength was deduced from the observations within the area covered by the plume simulations. The applied methodology led to emission estimates of 0.26 Tg over a period of 25 days in May 2010 and 0.32 Tg over 2 days in 2011. The injection height was assessed by finding the largest overlap between the observed plume and an ensemble of plume forecasts injected at different levels. GOME-2 TCSO2 retrievals were assimilated with ECMWF's 4D-VAR algorithm to provide the initial conditions for plume forecasts. The SO2 analyses captured the plume maxima well but exaggerated the plume area. Plume forecasts were evaluated by means of hit-rate and plume-size statistics for different TCSO2 thresholds. Plume forecasts using either the source parameters only or the initial conditions only agreed reasonably with the observations but using both led to the best forecast performance. The initialisation improved in particular the forecast of the Grímsvötn plume several days after the end of the eruption. The correctness of the injection height was crucial during the eruption of Grímsvötn because of a pronounced vertical wind shear. The developed forecast and assimilation system can be applied for near-real-time forecasting of volcanic SO2 plumes.