Air Traffic Analysis Could Help Limit Spread of DiseaseJune 7, 2005
According to a Northwestern University analysis of air traffic data from airports around the world, the busiest airports are not necessarily the most important for getting international travelers from their starting points to their final destinations. This insight could be used to identify cities whose airports may play a disproportionate role in spreading infections such as SARS or influenza.
The study, published by the Proceedings of the National Academy of Sciences (PNAS), is the first time the truly critical hubs in the global air transportation network have been identified. The authors created a map of the 25 most “centrally important” cities and found them to be distributed uniformly across continents. In addition to helping limit the spread of disease, knowing the main routes people are traveling and the gateway cities they pass through could aid in regulating airline competition and combating terrorist attacks.
“While large airports with high volumes of traffic are important, small facilities can be crucial in the spread of contagious diseases,” said Luís A. Nunes Amaral, associate professor of chemical and biological engineering, who led the study.
“Port Moresby, which is a relatively small airport, is the main gateway for everybody in Papua, New Guinea, and many islands in the Pacific. Most people are never going to use airports in those locations, but if you have a disease starting in Papua, New Guinea, or the surrounding islands, then Port Moresby would be the airport through which the disease would propagate to the rest of the world.”
Amaral and his colleagues analyzed air traffic data from one week in November 2000 with more than 500,000 flights between 27,000 pairs of cities worldwide. (A total of 3,883 cities were included in the study.) They looked at two important factors: centrality (the number of routes that go through an airport, or how crucial the airport is to worldwide traffic) and connectivity (the number of non-stop flights from an airport, or how much air traffic there is).
They found the air transportation network to be scale-free, which means that connectivity is unbalanced: a few nodes are highly connected with others, while most are only sparsely connected. The Internet is an example of a scale-free network.
Yet, unlike what is expected for scale-free networks, the researchers found that a busy airport does not always mean it is important globally. For example, although Paris and Frankfurt have similar traffic levels, Paris is more important for worldwide travel. (Paris was ranked number one in connectivity and centrality, while Frankfurt was third in connectivity but only eighth in centrality.) The researchers determined this is because the global network is actually comprised of smaller communities defined by political as well as geographic considerations.
“The community structure of the air transportation network means that some small airports have greater importance than one would have predicted,” said Amaral, a physicist with expertise in computer-based modeling. “Sao Paulo, Brazil, ranks only 53 in number of direct connections. However, it is, together with Buenos Aires, Argentina, the most important hub in South America, making it more important for global connectivity than more connected places such as Denver, Colorado.”
The busiest or most connected airports were located mainly in Western Europe and North America. Chicago, with one of the world’s busiest airports in O’Hare International Airport, ranked sixth in terms of connections and 13th in centrality.
Other authors on the PNAS paper are Roger Guimerà, a postdoctoral fellow in Amaral’s lab; Adrian Turtschi, from Avanade Deutschland GmbH, Kronberg, Germany; and Stefano Mossa, from the European Synchrotron Radiation Facility, Grenoble, France.
The research was supported by the National Institutes of Health under grant 5-K25-GM069546.
For more information and rankings of the cities used in the study, go to <http://amaral.northwestern.edu/Press/Guimera-2005-PNAS.html>.