Comparing and Clustering Ensemble Forecast Members to Support Strategic Planning in Air Traffic Flow Management

Strategic air traffic flow management addresses predictions of significant capacity/demand imbalances two or more hours in the future. The current planning process for mitigating the imbalances resulting from large-scale weather events relies heavily on the translation of weather forecasts into traffic impact. However, with a typical look-ahead time of 6 to 8 hours, the major challenge results from the uncertainty in weather outcomes. Although probabilistic forecasts are available to depict the uncertainty, specifying effective strategies for delay mitigation requires more explicit traffic impact information in both space and time dimensions. In our recent work, the ensemble members from the Short Range Ensemble Forecast product have been used to represent as a wide range of deterministic weather scenarios for the en route airspace. While each of the scenarios is considered to have the same likelihood of occurrence, some of them demonstrate similar characteristics of impact so that the identification of the representative weather scenarios is possible. Such a limited but representative number of scenarios can significantly reduce the effort required by decision makers to develop mitigation strategies for each scenario. In this study, we will extend the research scope by integrating the weather impact prediction in both enroute and terminal airspace. The capacity reduction models developed for en route sectors and airports will be employed to facilitate estimating weather impact as well as the system-wide delays under a fast-time simulation tool. Performance metrics will be explored for classifying weather scenarios and tested with an ad hoc clustering algorithm. The difference among clustering results will be analyzed and discussed.