Quantifying Uncertainty of Ensemble Transport and Dispersion Simulations Using HYSPLIT

Uncertainty associated with identifying the source location of nuclear pollutants in the atmosphere after a nuclear fallout using a numerical model is often difficult to determine. Uncertainty can originate from input data (meteorological and emissions), internal model error, physics parameterizations, and stochastic processes. This study uses the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model with ground and airborne data from the 1983 Cross Appalachian Tracer Experiment (CAPTEX) (Ferber et al. 1986) to quantify uncertainty in ensemble dispersion simulations. This is accomplished by utilizing an ensemble of forward trajectory simulations with varying initial conditions and physical parameterizations (e.g. turbulent velocity, boundary layer stability and mixed layer depth). Model rank spread is determined and then compared with the statistical model rank from the CAPTEX experiment (Stein et al. 2015) to ascertain uncertainty. Further consideration is given to the correlation coefficient between measured and predicted concentrations. Effectively quantifying and communicating uncertainty is crucial in providing probabilistic results in nuclear treaty monitoring.