Robustness of city road networks at different granularities

The city road system, as well as other complex systems, can be modeled as networks. Studies investigating the relationship between network structure and functionality provide a novel perspective to investigate the robustness of a city road network by analyzing its structure. The fact that different events may lead to different levels of traffic interruption requires that the investigation on the robustness of city road networks be conducted at multiple granularities. In this study, road networks are modeled at three different granularities: segment-based, stroke-based and community-based model to investigate the response of road networks under punctiform, linear and zonal traffic interruptions respectively. An empirical study with six city road networks around the world shows that the topological structure, especially the diversity level in the betweenness centrality distribution of the network, is more essential to the robustness of the city road network than geographical features and inherent attributes. The performances of city road networks under attacks are consistent in different cities due to the similarity in their topological structures but differ by granularity because city road networks modeled at different granularities have different topological structures. This variation illustrates that the robustness analysis of city road networks should be modeled at the appropriate granularity for a given traffic situation. Furthermore, empirical results suggest that the tolerance threshold for urban traffic is the critical point where the city road network begins to split into several components. The robustness analysis solution proposed in this article can identify these critical points to provide an early warning in urban traffic interruptions.