Identifying coherent flows in air traffic

When tracks of air traffic between a given city pair are collected over time and plotted, the resulting picture typically exhibits some internal structure or pattern reflecting a combination of air traffic control (ATC) procedures, airline procedures and weather described by B. Everitt (1981). In this study, we examined recorded track data between several city pairs and developed an interactive software tool that effects a coarse filtering of the data to provide an initial estimate of internal structure. The original motivation for this study came from our simulation modeling activities for the Federal Aviation Administration (FAA). As part of the FAA's National Airspace redesign (NAR) program, the National Airspace System (NAS) is continually assessed and often restructured to improve overall efficiency. In support of this effort, The MITRE Corporation's Center for Advanced Aviation System Development (CAASD) has modeled proposed design changes in order to estimate their impact. This article begins by describing the clustering algorithm and its output. The algorithm is then applied to several datasets to assess the significance of this output. We end with some conclusions and suggestions for future work.