Uncertainties in Initial Plume Modeling: What is needed prior to a volcanic eruption?

Abstract:

Low-probability high-intensity volcanic events are likely to have the greatest impact on us all. Pre-event scenario planning can help us to understand the likelihood of event type, size and duration while observational data can be used, once an event occurs, as calibration/validation to the pre-event modeling. During the 2010 Eyjafjallajökull eruption uncertainty in initial plume dynamics and ash mass lead to significant variability in modeling the downwind ash clouds. To forecast the potential dispersal patterns from volcanic eruptions assumptions are made on eruption start time; initial plume dynamics; erupted mass; and erupted/dispersing particle size distributions. There is a need for pre-event inputs to the dispersion models used in operational ash cloud forecasting. There are a number of different initial plume models available and each has its own set of defined inputs that are needed to model the eruption plume. There is a need for updated rapidly produced scenarios daily at time of unrest and updated as soon as the event occurs. While validation with satellite data is key, the models can forecast ash in locations that could be below detection limits that are still relevant for mitigation strategies and aviation safety. So, are we sufficiently prepared to perform the number of initial plume modeling simulations to capture all the potential variability in the dispersion model inputs? Can a systematic approach be developed for application within the operational environment and for use by the volcanic ash advisory centers? Here we will present approaches to building a real-time workflow that encompasses the initial plume uncertainties and discuss how we as a community can develop tools to be implemented in an operational environment. The aim is to a build a complete system to provide the simulations of the ash cloud location and concentrations along with confidence levels that can then be used by the aviation community in their decision support systems.