Scaling of Street Network Centrality with City Population
R. L. Fagundes a,1 , G. G. Piva a , A. S. Mata a , F. L. Ribeiro a, · 2025
Read Scaling of Street Network Centrality with City Population by R. L. Fagundes a,1 , G. G. Piva a , A. S. Mata a , F. L. Ribeiro a, (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-12006.txt.
Introduction
Fagundes and co-authors study how the accessibility of street networks scales with city size, in a 2025 paper covering more than five thousand Brazilian cities. Accessibility here means how close, on average, each point in the network lies to all others. Urban scaling laws describe much of how cities change as they grow, but the street side of the story was little explored.
The team finds that average closeness falls as population rises, following a power law with an exponent near 0.38. That fall is slower than a grid would give, which points to a network richer than two-dimensional. Big cities, the authors argue, trade closeness in the core for better links overall: expansion pushes access toward the edges while shortcut routes keep the city navigable. The results join a known urban model with network measures, offering a frame for weighing efficiency against fair access as cities grow.
Catalog record
| CREATOR | R. L. Fagundes a,1 , G. G. Piva a , A. S. Mata a , F. L. Ribeiro a, |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 12 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-12006
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Scaling of Street Network Centrality with City Population R. L. Fagundes a,1 , G. G. Piva a , A. S. Mata a , F. L. Ribeiro a, Abstract Urban scaling laws reveal how cities evolve as their populations grow, yet the role of street network accessibility in this process remains underexplored. We analyze over 5,000 Brazilian cities to establish a scaling law linking average closeness centrality ⟨ c C ⟩ \langle c_{C}\rangle —a measure of structural accessibility in street networks—to population size N N . Our results demonstrate that ⟨ c C ⟩ \langle c_{C}\rangle decays sublinearly as N − σ N^{-\sigma} ( σ ≈ 0.38 \sigma\approx 0.38 ), indicating that larger cities redistribute accessibility from cores to peripheries while maintaining navigability through hierarchical shortcuts. This scaling arises from the fractal interplay between infrastructure and population, characterized by a network dimension d ≈ 2.17 d\approx 2.17 , which exceeds that of a 2D grid. The slower decline in closeness centrality ( σ < 0.5 \sigma<0.5 ) reflects a trade-off: urban expansion reduces proximity but enhances connectivity through optimized path diversity, fostering economic dynamism. By integrating the Molinero & Thurner model with network centrality metrics, we provide a framework to reconcile infrastructure efficiency with equitable accessibility in growing cities. Keywords: scaling laws, urban systems, networks, infrastructure, centrality measures. 1 1 footnotetext: Corresponding author: [email protected] I Introduction Cities are hubs of social, economic, political, and cultural activity, where interactions between individuals and their environment shape emergent urban dynamics. These interactions drive socioeconomic output, with road networks forming the physical backbone that enables their propagation across urban scales. Crucially, while both human interactions and infrastructure underpin urban efficiency, they exhibit divergent scaling behaviors with respect to city population size Bettencourt2007 ; Ribeiro2017 ; Meirelles2018 ; Marc2019 ; Dong2020 ; Molinero2021 ; Rybski2022 . Empirical studies reveal that many urban metrics, from social interactions schlapfer2015 to road network properties Molinero2021 , scale with population size N via a power law: Y\sim N^{\beta}, (1) where the exponent \beta distinguishes between regimes of increasing returns ( \beta>1 , e.g., GDP, innovation), economies of scale ( \beta<1 , e.g., road networks), and linearity ( \beta=1 , e.g., water consumption) Bettencourt2007 ; strano2016rich ; Molinero2021 . These scaling patterns reflect fundamental principles of complex systems: cities self-organize through non-trivial agent interactions, exhibit multifractal properties in infrastructure and population distributions Murcio2015 ; Zhou2020 ; Stepinski2020 , and resemble dissipative systems exchanging energy and matter with their environment Hidalgo2021 ; Broekel2023 . Theoretical frameworks link these dynamics to mechanisms such as mobility-productivity trade-offs, fractal infrastructure, and rank-size distributions Bettencourt2013 ; Yakubo2014 ; Molinero2021 ; Ribeiro2023 , offering insights into mobility patterns, policy design, and infrastructure resilience. Within this context, street networks play a dual role: they enable socioeconomic interactions while embodying hierarchical complexity. Vehicular flows concentrate unevenly across streets, reflecting structural hierarchies that influence accessibility, multimodal transport use, and emissions Lammer2006 ; Bassolas2019 . Centrality measures, particularly closeness centrality, quantify this hierarchy by identifying streets critical to connectivity and urban function Freeman1977 . Closeness centrality, for instance, correlates with public-oriented activities like retail services porta2009correlating and informs resilience analyses, as disruptions to central roads disproportionately impact traffic Robson2021 ; Nair2019 . Yet, despite its relevance, a scaling law linking average closeness centrality to population size remains unestablished—a gap limiting our understanding of how infrastructure adapts to population growth and complexity. This study addresses this gap by analyzing street networks in over 5,000 Brazilian cities to derive an empirical scaling law for average closeness centrality. Building on Molinero & Thurner’s model Molinero2021 , we propose a theoretical framework connecting population-driven infrastructure demands to structural accessibility. Our results deepen the interplay between scaling laws and network science, offering insights into urban adaptation and informing equitable infrastructure planning. II Materials and Methods The study leverages data from the IBGE API, including 2022 population estimates from Brazil’s most recent census population2022 . Street networks for over 5,000 Brazilian cities were extracted using the osmnx Python package OpenStreetMap , with urban boundaries defined by satellite-derived 2015 urban area data embrapa2015 . This combination ensures spatial alignment between population metrics and infrastructure topology. Each city’s road network is modeled as an undirected, weighted spatial graph G=\{E,V\} , where E represents street segments (edges) and V denotes intersections or endpoints (nodes). Nodes are embedded in 2D space via coordinates (x,y) , preserving geographic fidelity. For each city, n=|V| and e=|E| define the graph’s size and connectivity. Shortest path distances \ell_{ij} between nodes i and j underpin key metrics. The farness of node i is f_{i}=\sum_{i\neq j}\ell_{ij} , and its closeness centrality is c_{i}=1/f_{i} , quantifying proximity to all other nodes. The network’s average shortest path length \ell scales with n in a d -dimensional lattice as \langle\ell\rangle\propto n^{1/d} mata2020 , implying dimensionality-dependent accessibility. Formally, closeness centrality is defined as murray1965 : c_{C}(i)=(n-1)\sum_{i\neq j}\frac{1}{\ell_{ij}}, (2) where n-1 normalizes by reachable nodes. Particularly in planned cities, the average closeness \langle c_{C}\rangle\propto 1/\langle\ell\rangle thus scales as: \langle c_{C}\rangle\sim n^{-1/d}. (3) This scaling reflects the network’s spatial efficiency, with lower d (e.g., d=2 for planar grids) correlating with slower decay in accessibility. Closeness centrality has dual interpretive value: it signals transportation cost savings at locations (nodes) Sevtsuk2012 and predicts facility distributions (e.g., residential, retail) shi2024 . Originating in social network analysis bavelas1950 , its adaptation by Murray and Freeman emphasized efficiency and interdependence in spatial systems murray1965 ; freeman1978 . Figure 1: The average degree a), scaling relations between nodes and average shortest path with a positive slope of 0.456\pm 0.003 b), and average closeness centrality with a negative slope of -0.428\pm 0.005 c). Street networks with closeness centrality heatmaps for four Brazilian cities: d) Belo Horizonte (MG), e) São Paulo (SP), f) Salvador (BA), and g) Rio de Janeiro (RJ). III Results The histogram in Figure 1 a) reveals that Brazilian cities exhibit quasi-regular street networks, with an average node degree \langle k\rangle=3\pm 0.3 . This suggests a semi-planned urban structure where most intersections connect three streets. Figure 1 b) demonstrates a power-law scaling between the average shortest path length \langle\ell\rangle and the number of nodes n , expressed as \langle\ell\rangle\sim n^{1/d} . The derived network dimension d\approx 2.17 (vs. d=2 for a regular grid) indicates enhanced navigability, as path lengths grow slower than in strictly planar lattices. This deviation from d=2 arises from occasional high-degree nodes that act as shortcuts, reducing overall connectivity distances. Compared to UK street networks ( \langle k\rangle\approx 2.4 Arcaute2016 ), Brazilian networks balance grid-like regularity with limited randomness, avoiding the logarithmic scaling ( \langle\ell\rangle\sim\ln n ) typical of random graphs. Figure 2: Empirical scaling of average closeness centrality \langle c_{C}\rangle versus population N . The green line denotes the power-law fit \langle c_{C}\rangle\sim N^{-\sigma} . On the other hand, closeness centrality \langle c_{C}\rangle declines with n as \langle c_{C}\rangle\sim n^{-1/d} (Fig. 1 c), signaling a gradual redistribution of accessibility from central to peripheral areas. Visualizations of four cities (Figs. 1 d-g) confirm that nodes in central urban zones exhibit higher closeness centrality, aligning with heatmap patterns of proximity. Figure 2 establishes the empirical scaling of \langle c_{C}\rangle with population N : \langle c_{C}\rangle\sim N^{-\sigma},\quad\sigma=0.38\pm 0.01. (4) This power-law decay reflects reduced proximity between urban