An Aggregate Sector Flow Model for Air Traffic Demand Forecasting

Abstract

Sector capacity, number of aircraft permitted in a region of the airspace referred to as a sector, is used to limit air traffic to an amount that can be safely handled by a human controller. The traditional approach for predicting traffic demand in a sector involves the simulation of trajectories of individual aircraft. The demand provided by this approach is inaccurate for hourly time horizons. This paper explores the use of linear time varying aggregate models constructed using historical data for predicting sector demand. Since air traffic varies with the seasons, day of the week and weather, multiple aggregate models can be built to represent different situations. The paper evaluates the accuracy of using a single aggregate model and compares it with the improvements that can be achieved by using several aggregate models and selecting the best model based on hypothesis testing. The paper also presents the results of using a combination of probability-weighted predictions from several aggregate models. Evaluation of the prediction errors for all the high-altitude sectors in Indianapolis Center for a month shows that the multiple aggregation models result in smaller errors; errors vary with the sector and do not vary significantly with the prediction interval.