Revealing Spatiotemporal Correlation of Urban Roads Via Traffic Perturbation Simulation
Baoju Liu,Jun Long,Min Deng,Jianbo Tang,Jincai Huang
DOI: https://doi.org/10.1016/j.scs.2021.103545
IF: 11.7
2022-01-01
Sustainable Cities and Society
Abstract:Unbalanced interactions in cities result to a correlation between urban road segments with spatiotemporal heterogeneity. The accurate detection of this spatiotemporal correlation can significantly contribute to traffic congestion prediction and sustainable urban operations. However, the spatiotemporal nonstationarity of the traffic flow causes dynamic uncertainty in road network correlation. Therefore, we proposed a novel method to evaluate the spatiotemporal correlation of road networks. First, a dynamic chain propagation model of traffic perturbation was constructed to track the propagation process of traffic response. Then, the time and spatial lag operators were designed based on the perturbation response and utilized to detect the local spatiotemporal correlation of the road network. Finally, a new remote correlation measurement model was proposed by capturing the traffic flow contribution. Experiments demonstrated that the average local correlation of an urban road network is 0.83 in first-order neighborhood segments, which decreases to 0.72 in third-order neighborhood segments with the synchronous increase of space-time lag operators. Moreover, the correlations of downstream segments are on average 11.6% higher than those of upstream segments. These results can provide targeted decision support for preventing the potential spread of traffic congestion, reducing urban air pollution, and promoting sustainable urban development.