Indicator of serious flight delays with the approach of time-delay stability

Passenger flight delays, causing much disorder of air traffics, economic losses of airlines, and downgrading the travel quality of millions of people, are ubiquitous phenomena in airports all over the world. Investigation based on real data from the view point of statistical physics is rarely seen. In the present work, big data of such delay records over 20 years accumulated by Bureau of Transportation Statistics in the United States are downloaded and purified by us. We account the departure and arrival records of such flights between certain pair of airports as time series, and rectify them by defining dimensionless velocity of the flights. Furthermore, we find the varying cross-correlations among such time series with the approach of time delay stability, and describe the correlations with temporal networks for correlation states. Deterministic correspondences between the average degrees of temporal networks and delay ratios of passenger flights are verified in different sampling groups of flights with the longest records. The mean degrees of correlation networks usually emerge a peak prior to that of high delay ratios, which serves an indicator for the precaution to serious flight delays.