WEATHER NORMALIZATION FOR EVALUATING NATIONAL AIRSPACE SYSTEM ( NAS ) PERFORMANCE

The need to benchmark air traffic management performance, predict future performance, and improve our understanding of air traffic operations has created a growing literature on the statistical modeling of delays in the National Airspace System (NAS). This paper contributes to that literature in three distinct ways. First we introduce an innovative delay metric that avoids the distortions created by schedule padding and can be decomposed into different flight phase components. Second, we examine closely how daily variation in weather-impacted traffic affects operational performance. Third, we consider the impacts of weather forecast errors as well as realized weather. Our results demonstrate the value of the new delay metric, show that simple daily averaging adequately captures the effect of weather-impacted traffic, and reveal that falsepositive weather forecast errors are a significant source of delay in the NAS.