Real-Time Forecasts of Aircraft Departure Queues

Abstract

Safe and efficient management of air traffic requires accurate predictions of aircraft trajectories. In the existing air traffic system, predictions of takeoff times at congested airports are a major source of forecast error, and few facilities exist to monitor and forecast airfield congestion using available data. This paper develops models of aircraft flow to predict departure congestion. In these models, the rate of flow onto the airfield is determined by aircraft pushbacks, while the departure rate is determined by runway capacity. The models are data driven and are updated as new information arrives. Empirical tests using data from Atlanta Hartsfield and Boston Logan airports demonstrate that the models produce accurate predictions of airfield congestion over 10 min to 1 h forecast horizons.