Towards probabilistic forecasts of volcanic ash transport in the atmosphere

Abstract:

Satellite based remote sensing techniques are the primary means of identifying the location of volcanic ash during eruption events. This information is then used to initialize ash dispersion models, which employ meteorological fields to forecast the future locations of ash. Remote sensing of ash in tropical regions is especially challenging due to the difficulty of demarcating ash from the ice and water in convective clouds which results in frequent missed and false detections of ash. Dispersion models also contain uncertainties that arise from uncertainties in the meteorological fields and the source term such the height of the ash column. Communicating forecasting uncertainties is becoming increasingly important to stakeholders for the purposes of improved risk management. For these reasons the Bureau of Meteorology is engaged in research with the aim of providing probabilistic forecasts of ash in the near future based on ensembles of dispersion model simulations. The ensembles are constructed to reflect two fundamental uncertainties. Firstly, uncertainties in the meteorological fields are incorporated by the use of the Bureau’s ensemble model system which issues a set of 24 meteorological forecasts representing possible states of the atmosphere. Secondly, uncertainties in the model parameters such as the ash column height are also incorporated. This is accomplished by running a suite of dispersion model simulations, each with a different value of the model parameter. Pattern correlations are then used to quantify the match between the model and observations. The model parameters which provide the best matches between model and observations are then employed in the ensemble to issue probabilistic forecasts of ash. This process can also ameliorate errors due to incorrect or missing model physics. The efficacy of these techniques shall be demonstrated by the use of several case studies.