Centrality anomalies for the domestic air transportation networks in the USA: an empirical benchmark

Air transportation systems are a foundational infrastructure for the human's society. The lack of systematic and detailed investigation on a large amount of records for air flights has blocked seriously the deep understanding of the systems. By using the American domestic passenger flight records from 1995 to 2020, we constructed the air transportation networks and calculated the betweenness and the eigenvector centralities for the airports. It is found that in terms of eigenvector centrality, 15-30% airports in the unweighted and undirected networks behave anomalous. The anomalies disappear after considering the information of link weights or directionalites. Five widely used models for air transportation networks are evaluated, results for which tell us that the spatial constraints are required to eliminate the anomalies detected by the eigenvector centrality, and provide us some references for selecting the parameters in the models. We hope the empirical benchmarks reported in this paper can stimulate much more works on theoretical models for air transportation systems.

Automated summary

Air transportation networks were constructed using American domestic passenger flight records from 1995 to 2020, and betweenness and eigenvector centralities were calculated for airports. Anomalies were observed in 15-30% of airports in the unweighted and undirected networks based on eigenvector centrality, but disappeared when considering link weights or directionalities. Five widely used models for air transportation networks were evaluated, showing that spatial constraints are necessary to eliminate anomalies detected by eigenvector centrality, providing references for parameter selection in the models. The empirical benchmarks reported in this paper are expected to stimulate more theoretical modeling works on air transportation systems.