Network analysis reveals patterns behind air safety events

• A complex network representation of air traffic is proposed.
• Aircraft are mapped into nodes, pairwise connected according to their distance.
• Resulting topologies are able to forecast the appearance of unsafe situations.

Complex networks have been extensively used to study the topological and dynamical characteristics of transportation systems, although far less attention has been devoted to the analysis of specific problems arising in everyday operations. In this work, the use of a network representation is proposed for studying the appearance of Loss of Separation events, a kind of safety occurrence in which two aircraft violate the minimal separation while airborne. The topological analysis of networks representing the structure of traffic flows allows identifying situations in which the probability of appearance of such events is increased. Beyond these specific results, this work demonstrates the usefulness of the complex network approach in the analysis of operational patterns and occurrences.