Novel Atmospheric Ensemble Modeling Techniques Applied to Long-Range Transport

Ensemble modeling (EM) has become an essential tool for assessing the uncertainty in atmospheric model predictions.  Most attention has centered on temperature and precipitation forecasts, but improved wind and turbulence prediction for atmospheric transport and dispersion is also of interest.  Airborne transport models are commonly driven by mesoscale atmospheric models, whose accuracy is limited by model biases, available data, and natural variability.  These errors are compounded with time and render predictions of long range transport increasingly less accurate.  Ensemble modeling quantifies model uncertainty by providing a range of possible atmospheric end states.  However, ensembles are still subject to underlying biases and have yielded only marginal improvement because of model complexity and casual application to special situations.  This research has used the first European Tracer Experiment (ETEX) as a testbed to compare standard EM modeling with two novel techniques: (1) a physics-based ensemble technique which adapts models to specific geographical locations and time frames, and (2) data assimilation with an Ensemble Kalman filter.  Improvements are discussed using the novel ensemble methods, as well as future research needs.