Exploring Volcanic Ash Forecasting Techniques Using HYSPLIT and VOLCAT Observations

Abstract:

The aviation industry relies heavily on ash forecasts from Volcanic Ash Advisory Centers (VAACs) to determine safe flight paths around volcanic hazards. Therefore, deterministic forecasts must be produced quickly, providing information about ash height and location to keep airlines informed of the evolving situation. Volcanic ash forecasts for aviation are moving from deterministic forecasts, towards quantitative ash forecasts which provide more information about ash concentration and uncertainty.

We initialize HYSPLIT, a transport and dispersion model, with satellite retrievals of volcanic ash cloud properties like mass loading and cloud top height at the location of the ash cloud. Observations from the VOLcanic Cloud Analysis Toolkit (VOLCAT) developed by NOAA/NESDIS in collaboration with the University of Wisconsin are used to initialize HYSPLIT simulations for the June 2019 eruption of Raikoke, located in the Kuril Islands. VOLCAT uses automated algorithms to identify volcanic clouds from multiple satellite platforms and issues real-time alerts for any detected volcanic activity. For this eruption, VOLCAT observations are available in 10 minutes intervals over 3 days. Each observation is used to initialize HYSPLIT and quickly generate a forecast in a computationally efficient manner. The forecasts are then combined to create quantitative products that communicate forecast uncertainty. We explore the differences between forecasts initialized with VOLCAT observations, with a source term derived from an inverse algorithm, and forecasts initialized as a uniform ash column from the volcanic vent. We find statistically significant improvements to model forecasting performance when utilizing this data-insertion initialization procedure, and discuss generation of a probabilistic product from multiple forecasts. Lastly, development of a testbed HYSPLIT web application at the NOAA Air Resources Laboratory that can implement this new data-insertion forecasting technique is underway.