Probabilistic volcanic ash cloud simulations: Characterizing the uncertainty and moving into the operational environment

Thursday, 18 December 2014
Peter Webley1, Abani K Patra2, Marcus I Bursik3, E Bruce Pitman4, Jonathan Dehn1, Tarunraj Singh2, Puneet Singla2, Elena Ramona Stefanescu2, Reza Madankan2, Solene Pouget5, Matt Jones6, Donald Morton7,8 and Christopher G Hughes5, (1)University of Alaska Fairbanks, Geophysical Institute, Fairbanks, AK, United States, (2)SUNY Buffalo, Department of Mechanical & Aerospace Engineering, Buffalo, NY, United States, (3)SUNY Buffalo, Department of Geology, Buffalo, NY, United States, (4)SUNY Buffalo, Department of Mathematics, Buffalo, NY, United States, (5)SUNY Buffalo, Deptartment of Geology, Buffalo, NY, United States, (6)SUNY Buffalo, Center for Computational Research, Buffalo, NY, United States, (7)University of Alaska Fairbanks, Arctic Region Supercomputing Center, Geophysical Institute, Fairbanks, AK, United States, (8)Boreal Scientific Computing, LLC, Fairbanks, AK, United States
The aim of this presentation is to illustrate how a probabilistic simulation workflow can be built to provide real-time products, and therefore be used in a real-time decision support environment for volcanic plumes and ash clouds. Generating simulations of volcanic plumes and ash clouds requires knowledge of the eruption characteristics and the potential atmospheric variability, both above the volcano and at distal locations downwind as the ash cloud disperses. Often the initial event characteristics, such as plume height, particle size distribution, and eruption rate have a level of uncertainty associated to them. Therefore, there is a need to simulate the full spectrum of eruption inputs to a dispersion model, to produce a complete range of downwind ash concentrations and atmospheric mass loadings, each with their own probability. Numerical weather prediction (NWP) data are generated as ensemble members also, each representing a potential state of the atmosphere. Therefore, one needs to couple the NWP ensemble members to the eruption uncertainties to build the probabilistic ash cloud concentrations and mass loadings. Here, we show the results of a Lagrangian dispersion model, Puff, used with a one-dimensional volcanic plume model, BENT, to characterize the input variability, and coupled with ensemble NWP data to build a full probabilistic set of ash cloud simulations. We show how these real-time simulations can be generated and then compared to remote sensing data for calibration/validation of the simulations' probabilistic output and how end users could apply the developed products in their operational environment.