The inclusion of large-scale data-monitoring and data-storage systems has presented researchers with new opportunities for understanding the decisions of air traffic controllers in managing traffic. This paper takes a step in the direction of data-driven analysis by clustering trajectories according to air traffic control decisions (e.g. trajectory changes), as opposed to clustering on spatial positioning reports. The particular focus for this paper is in the immediate vicinity of airports. Unlike in enroute or terminal area airspace where aircraft typically traverse well established routes and procedures, near airports there tends to be greater variability in air traffic decisions (e.g. vectoring and tromboning) to appropriately space aircraft at the runway, ultimately leading to spatial dispersion of aircraft trajectories. Using a hidden Markov model, heading changes within aircraft trajectories are identified, extracted, and translated into variable-length trajectory strings. Comparison of the trajectory strings using edit distance metrics then allows for clustering of the trajectories according to the air traffic control decisions. The classification and clustering process is applied on a set of historical trajectories at Washington National Airport. The resulting clusters foster an understanding of the arrival traffic structure and the decision strategies of controllers.
Clustering Aircraft Trajectories According to Air Traffic Controllers' Decisions
2020-10-11
2097740 byte
Conference paper
Electronic Resource
English
HIERARCHICAL CLUSTERING OF AIRCRAFT SURFACE TRAJECTORIES
TIBKAT | 2020
|Hierarchical Clustering of Aircraft Surface Trajectories
AIAA | 2020
|Clustering Object Trajectories for Intersection Traffic Analysis
TIBKAT | 2020
|