| Summary: | During the Fagradalsfjall eruptions, the CALPUFF dispersion model was used operationally to forecast groundlevel exposure to SO2. The initial graphical presentation of the forecast used the color code system used by the Environmental Agency of Iceland to refer to critical SO2 thresholds of concern for human health. The model however was found to predict peaks that were not observed. In reaction to the recognition of this overprediction, two additional maps that showed the most likely areas to be impacted by elevated SO2 concentrations within 6- and 24-hours were also released after the first month of the first eruption. People were encouraged to check the real-time measurements available on the EAI webpage for gas concentrations at the station closest to them in addition to utilizing the forecasts. The small-sized plumes (most common heights between 1 ¿ 2 km asl) are challenging to model as the boundary layer has a dominating impact on them. Additionally, the domain of interest is a small peninsula where very local atmospheric dynamics are not well captured by the meteorological models driving the dispersion code. Various approaches are being tested to improve the model forecast including updating plume height and SO2 flux source terms using different data sources and producing probabilistic maps as well as developing an AI-based algorithm for quantifying the eruption source parameters in near-real time. Our results highlight the challenge in achieving accurate simulations of volcanic plume dispersion. The performance remains to be improved with respect to forecasting reliable ground-level concentrations of volcanic SO2.
|
|---|