Clustering Days with Similar Airport Weather Conditions

On any given day, traffic flow managers must often rely on past experience and intuition when developing traffic flow management initiatives that mitigate imbalances between aircraft demand and weather impacted airport capacity. The goal of this study was to build on recent efforts to apply data mining classification and clustering algorithms to vast archives of historical weather and air traffic data to identify patterns and past decisions that can ultimately inform day-of-operations decision-making. More specifically, this study identified clusters of hours for which the probability of imposing a Ground Delay Program was similar. The identification of these similar hours was a two-step process. In the first step, historical weather data, scheduled arrival rates and traffic flow management logs were used to estimate the hourly probability of a Ground Delay Program being implemented at both Chicago O’Hare International Airport and Newark Liberty International Airport. These hourly probabilities were subsequently clustered in the second-step to identify collections of hours in which the likelihood of a Ground Delay Program was similar. For Chicago, a total of five clusters were identified, while thirteen clusters were identified for Newark. This large difference reflects the increased diversity in air traffic and weather conditions that result in Ground Delay Programs being implemented at Newark, as compared to Chicago. For both airports, clusters associated with nighttime and early morning operations were clearly identifiable, and the probabilities associated with Ground Delay Programs occurring for these clusters was very low, as expected. Additional clusters representing typical “bad weather” days were also identified where the probabilities of a Ground Delay Program being implemented was high, but these were rare and accounted for less than 3% of all the hourly events. Somewhat surprising was the large number of clusters identified for Newark where the weather and traffic conditions were similar, but very inconsistent Ground Delay Program usage was observed. Finally, the frequency of occurrence of Ground Delay Programs, Ground Stops and Miles-in-Trail restrictions implemented for each of the clusters was analyzed. Based on the results, it does appear as if the usage of Miles-in-Trail, Ground Delay Program and Ground Stop restrictions correlates well with the probabilities of Ground Delay Program occurrence. Furthermore, the results demonstrate that it is feasible to use historical weather and air traffic archives to provide guidance on the types of traffic management initiatives to implement in response to the weather and traffic conditions impacting an airport.