Atmospheric Dispersion Modelling of Volcanic Ash using Data Insertion

Eruption source parameters in volcanic ash dispersion and transport modelling, such as plume height and eruption rate, can often be highly uncertain. This can lead to significant uncertainties in the position and concentration of the modelled ash cloud downwind of the vent. Methods such as inversion modelling have successfully constrained such uncertainties, but in this work estimation of the eruption source parameters for the atmospheric dispersion model NAME is bypassed by implementing data insertion. Using this method under development, ash cloud properties retrieved from satellite imagery are used to create ash sources downwind from the volcano vent, from which dispersion simulations are initialised. Using the satellite retrievals, a set of simulations are initialised from different times and combined to create forecasts. In other experiments the simulations are sequentially updated using a probabilistic cloud / ash / clear classification scheme to correct the model state over time. Simulations from the Eyjafjallajökull and Grímsvötn eruptions compare well against other established modelling methods and satellite observations.