Robustness Analysis of City Road Network at Different Granularities - Space-Time Integration in Geography and GIScience - page 125

Previous researches have proved that the efficiency and robustness of urban
road networks highly depend on the network structure (Demšar et al. 2008 ). To
provide a practical robustness analysis procedure for urban road networks, we aim at
evaluating the robustness of urban road networks by analyzing urban road network
structure.
Originating from studies of complex networks, the concept of network structural
robustness denotes the decrease of network performance due to the selected removal
of nodes or edges (Holme et al. 2002 ). Many robustness analyses are performed in
the real world artificial networks using complex network methods such as North
American power grid (Albert et al. 2004 ), water distribution networks (Yazdani and
Jeffrey 2011 ). Although the urban road network is already a network intuitionally, it
is different from the other networks because the granularity of the road networks
varies according to the demands of different applications. In traditional GIS
applications and navigation systems, the urban road network contains road segment
and junctions. From the point of view of urban planning, the urban network is
consisted of roads and their intersections where several road segments are grouped
as one element. Structural diversity occurs at different granularities of road network.
Whether this diversity will cause bias judgment for robustness analysis remains an
important research question.
Thus the fundamental issue about using complex network theory in robustness
analysis in urban road networks is exploring the structural differences at multiple
granularities. Our goal is inquiring the robustness of urban road networks at three
different granularities based on complex network theory to figure out whether and
how different granularities affect the robustness evaluation through an empirical
study in Beijing. Three successive hypothetical attack strategies are carried out to
observe the performances of road network under pressure. In each simulated attack
procedure, nodes are successively removed from the network and the performances
of the network as well as the spatial form of urban road network changes resulted
from the removal of nodes are recorded. These series of spatial forms provide a
clear conception of a spatial change procedure of urban road networks under attacks.
The results show that the performances under attacks vary in different granularity.
Segment-based road network lacks the sensitivity under intentional attacks, and
stroke-based road network cannot represent the split procedure of the urban road
network under attacks. Neither of them is appropriate to evaluate the robustness
of city road networks. Community-based road network can capture the traffic
interactions between adjacent road segments and use local impact area to evaluate
network robustness. We argue that community-based can reflect the network change
situation under attack at a mesoscopic point of view and is suitable for robustness
evaluation of city road networks.
This chapter is organized as follows: In the second section, we will introduce
three models of road network representation in different granularities. In Sect. 8.3 ,
the experimental procedures to evaluate the robustness of urban road network at
multiple granularities are described. The fourth section is devoted to the main results
and we summarized our conclusions and the implications for the further research in
the last section.