Probabilistic Prediction of Aggregate Traffic Demand Using Uncertainty in Individual Flight Predictions

Abstract

(Abstract) Federal Aviation Administration (FAA) air traffic flow management (TFM) decision-making is based primarily on a comparison of deterministic predictions of demand and capacity at National Airspace System (NAS) elements such as airports, fixes and en- route sectors. The current Traffic Flow Management System (TFMS) and its decision- support tools ignore the stochastic nature of the predictions. Taking into account uncertainty in predictions and moving from deterministic to probabilistic TFM is an important part of the NextGen program that will help TFM specialists make better and more realistic decisions. This paper uses current TFMS data to analyze how uncertainty in prediction of arrival times for individual flights translates into uncertainty in prediction of aggregate traffic demand counts at arrival airports. A methodology was developed for probabilistic prediction of aggregate 15-minute demand counts by using the probability distributions of arrival time predictions for individual flights. A key element of the methodology is that the aggregate demand counts are predicted from extended sets of flights with the estimated times of arrival (ETAs) in both the interval of interest and several adjacent intervals. Numerical examples are presented that illustrate the difference between deterministic and probabilistic traffic demand predictions.