Classification and computation of aggregate delay using center-based weather impacted traffic index

The total aggregate delay experienced by aircraft in the United States is a measure of performance of the air traffic management system. The concept of weather weighted traffic index (WITI), the number of aircraft affected by weather, is used to develop linear regression models for aggregate delay. WITI is a two-dimensional spatio-temporal quantity. The current delay models represent WITI as a function of time by aggregating its spatial variation. A piecewise linear regression model provides a better estimate of aggregate delay than a single linear regression model. However, the piece-wise model assumes that a day can be classified as a low, medium or high delay day. Classification methods using only time varying WITI information are not accurate. This paper recovers the regional information lost in the spatial aggregation process by modeling WITI as a function of time and air traffic control centers. The behavior of Center level WITIs is used to classify expected delay on a given day into a day with low, medium or high aggregate delay. The integrated method used to improve delay estimates is examined using traffic and weather data for the period April to August for the years 2004, 2005 and 2006. The classification methodology, together with the three-step regression model, provides a comprehensive approach to estimate delay.