Characterizing the Brazilian airspace structure and air traffic performance via trajectory data analytics

This paper presents a data-driven approach for multi-scale characterization of the Brazilian airspace structure and air traffic operational performance from aircraft tracking data recorded by surveillance systems. Unsupervised learning is performed with a flight trajectory clustering analysis to automatically identify spatial traffic patterns in both the terminal and the en route airspace for major origin-destination pairs of the Brazilian air transportation system. Based on the as-flown route structure learned, quantitative metrics are developed to describe the structural efficiency of the airspace and the operational efficiency of the traffic flows. For this, actual flight trajectories are projected onto reference nominal trajectories in space and time. The results allowed for cross-route comparisons of air traffic flow efficiency across multiple flight phases as well as for the identification of causal factors for trajectory deviations from nominal routes. An interactive data analytics tool is also created to output performance statistics and air traffic visualizations. With the provision of a systematic data-driven approach for characterizing actual air traffic operations, the analytics framework is envisioned to assist airspace design and performance monitoring processes and to provide the basis for developing predictive capabilities in support of traffic flow management.