Generating Representative Weather-Impact Futures for Strategic Traffic Flow Management: Airport Capacity Prediction

Motivated by needs in strategic traffic flow management, we study the problem of forecasting airport capacity profiles over a full-day time horizon. Specifically, we present a case study of Atlanta’s Hartsfield-Jackson International Airport (KATL), which explores use of 1) convective weather presence/absence near the terminal, 2) cloud ceiling height, and 3) wind speed as predictors of the total airport capacity. It is shown that terminal-area convective weather is a primary cause of capacity reduction, and also that low ceilings are a sensitive and specific predictor of reduced capacity. Wind speeds are shown to be less specific predictors, but nevertheless modulate capacity. Using these analyses, a preliminary predictive model that stochastically generates capacity profiles is proposed. This model leverages Terminal Aerodrome Forecasts of wind and ceilings, as well as convective-weather scenarios obtained from an ensemble-forecastderived weather simulator, in generating possible profiles. Model-generated forecast profiles for Atlanta Hartsfield airport are compared with the historical profile for a particular bad-weather day of interest, September 26, 2010.