Data-driven evaluation of a flight re-route air traffic management decision-support tool
Summary
Air traffic delays in the U.S. are problematic and often attributable to convective (thunderstorms) weather. Air traffic management is complex, dynamic, and influenced by many factors such as projected high volume of departures and uncertain forecast convective weather at airports and in the airspace. To support the complexities of making a re-route decision, which is one solution to mitigate airspace congestion, a display integrating convective weather information with departure demand predictions was prototyped jointly by MIT Lincoln Laboratory and the MITRE Corporation. The tool was deployed to twelve air traffic facilities involved in handling New York area flights for operational evaluation during the summer of 2011. Field observations, data mining and analyses were conducted under both fair and convective weather conditions. The system performance metrics chosen to evaluate the tool's effectiveness in supporting re-route decisions include predicted wheels-off error, predicted wheels-off forecast spread, and hourly departure fix demand forecast spread. The wheels-off prediction errors were near zero for half the flights across all days, but the highest 10% errors exceeded 30 minutes on convective weather days. The wheels-off forecast spread exceeded 30 minutes for 25% of forecasts on convective weather days. The hourly departure demand forecast spread was 9 flights or less for 50% of departures across all days except one. Six out of the seven days having the highest hourly departure demand forecast spreads occurred in the presence of long-lived weather impacts.