Enhancing urban resilience versus contagious diseases: An efficient clustering method for urban transportation networks

Abstract

Abstract Lockdowns in urban scale are shown to be a last resort during pandemics. Despite the effectiveness of this strategy in preventing the spread of disease, the economic necessities of citizens and the social and psychological difficulties created during lockdowns are severe challenges to their long-term implementation. Thus, officials pursue versions of lockdown that promote health-livelihoods dichotomy at the same time. This paper proposes a method for clustering cities such that, at the same time, the essential connections among residents and their basic sustenance (grocery and medical services) are maintained, and social bubbling is warranted. Clusters of a network are set of nodes densely connected to each other and sparsely connected to the rest of the network. Because the network is relatively sparse on the borders of clusters, they are cost-effective places for implementing control strategies such as regional lockdowns. As our clustering method is hierarchical, it allows different levels of clustering. As the number of clusters increases, the limitations get stricter, the physical range of trips gets shorter, and more social distancing is applied. The lower number of clusters results in more freedom but may result in higher virus spread risk. We apply our method to a large city (Isfahan, Iran). We assign a daily origin-destination matrix to our real network and use daily traffic flow among pairs of nodes as the proxy for their correlation and interaction. Results show that setting the minimum number of clusters (i.e., four) would ban 25% of daily non-work trips while increasing the number of clusters to 27 would ban 70% of those trips.