Improving departure taxi time predictions using ASDE-X surveillance data

Flights incur a large percentage of delay on the ground during the departure process; however, predicting the taxi-out time is difficult due to uncertainties associated with the factors influencing it, such as airport surface traffic, downstream traffic restrictions, runway configuration, weather, and human causes. Airport Surface Detection Equipment, Model X (ASDE-X) surveillance data provides high resolution coverage of aircraft surface movement which can be leveraged to address this problem. This paper presents a novel approach which builds an adaptive taxi-out prediction model based on a historical traffic flow database generated using the ASDE-X data. The model correlates taxi-out time and taxi-out delay to a set of explanatory variables such as aircraft queue position, distance to the runway, arrival rates, departure rates and weather. Two prediction models are developed. One treats aircraft movement from starting location to the runway threshold uniformly while the other models aircraft time to get to the runway queue different from the wait time experienced by the aircraft while in the runway queue. The models are evaluated using data from New York´s John F Kennedy (JFK) airport during the summer of 2010. Results show significant improvement in taxi-out predictions as compared to predictions from FAA´s Enhanced Traffic Management System (ETMS).