A Multi-Scale Analysis of 27,000 Urban Street Networks: Every US City, Town, Urbanized Area, and Zillow Neighborhood
Summary (3 min read)
Introduction
- Cross-sectional analysis of American urban form can reveal these artifacts and histories through street networks at metropolitan, municipal, and neighborhood scales.
- Second, reproducibility has been difficult when the dozens of decisions that go into analysis—such as spatial extents, topological simplification and correction, definitions of nodes and edges, etc.—are ad hoc or only partly reported (e.g. Porta et al., 2006; Strano et al., 2013).
- First, it describes and demonstrates a new methodology for easily and consistently acquiring, constructing, and analyzing large samples of street networks as nonplanar directed graphs.
- Third, it investigates with large sample sizes some previous smallersample findings in the research literature.
Methodology
- Street networks can be conceptualized as primal, directed, nonplanar graphs.
- Planar graphs may reasonably model the street networks of old European town centers, but poorly model the street networks of modern autocentric cities like Los Angeles or Shanghai with many gradeseparated expressways, bridges, and underpasses (Boeing, 2018b).
- OSMnx is a Python-based research tool that easily downloads OpenStreetMap data for any place name, address, or polygon in the world, then constructs it into a spatially-embedded graph-theoretic object for analysis and visualization (Boeing, 2017).
- The authors retain only the urbanized areas subset of these data (i.e. areas with greater than 50,000 population), discarding the small urban clusters subset.
- The second set of geometries defines their municipal-scale study sites using 51 separate TIGER/Line shapefiles (again, 2016) of US Census Bureau places within all 50 states plus DC.
Street network measures
- The network’s average node degree quantifies connectedness in terms of the average number of edges incident to its nodes.
- It measures the average number of physical streets that emanate from each node (i.e. intersection or dead-end).
- In total, this study cross-sectionally analyzes 27,009 networks: 497 urbanized areas’ street networks, 19,655 cities’ and towns’ street networks, and 6857 neighborhoods’ street networks.
- These sample sizes are larger than those of any previous similar study.
Metropolitan-scale street networks
- Table 1 presents summary statistics for the entire data set of 497 urbanized areas.
- The gridlike San Angelo, TX urbanized area has the most streets per node (3.2) on average, and (outside of Puerto Rico, which contains the seven lowest urbanized areas) the sprawling, disconnected Lexington Park, MD urbanized area has the fewest (2.2).
- The relationship between fine-grained networks and connectedness/gridness is not, however, clear-cut: intersection density has only a weak, positive linear relationship with the proportion of four-way intersections in the urbanized area (r2 ¼ 0:17).
- Densities and average distances such as intersection density and the average street segment length exhibit only moderate heterogeneity.
- Due to the substantial variation in urbanized area size, from 25 to 9000 km2, the preceding analysis covers a wide swath of metropolitan types.
Municipal-scale street networks
- Table 3 presents summary statistics of street network characteristics across the entire data set of 19,655 cities and towns—every incorporated city and town in the US.
- The latter’s small sample size may limit the generalizability of this finding.
- These distributions comprise the lognormal, Gumbel, gamma, exponentiated Weibull, Fréchet, power-law, uniform, and exponential distributions.
- An exception to this general pattern, of course, lies in consistently-sized orthogonal grids filling a city’s incorporated spatial extents.
- The authors find that such cities are not uncommon in the US, particularly between the Mississippi River and the Rocky Mountains: the Great Plains states are characterized by a unique street network form that is both orthogonal and reasonably dense.
Neighborhood-scale street networks
- The authors have thus far examined every urban street network in the US at the metropolitan and municipal scales.
- While the metropolitan scale captures the emergent character of the wider region’s complex system, and the municipal scale captures planning decisions made by a single city government, the neighborhood best represents the scale of individual urban design interventions into the urban form.
- A few neighborhoods have no intersections within their Zillow-defined boundaries, resulting in a minimum intersection density of 0 across the data set.
- Nationwide, the typical neighborhood averages 2.9 streets per intersection, reflecting the prevalence of three-way intersections in the US, discussed earlier.
- Some central San Francisco orthogonal grid networks with many four-way intersections—such as Downtown, Chinatown, and the Financial District—have surprisingly low ANCs: 1.5, 1.3, and 1.6, respectively.
Discussion
- These findings suggest the influence of planning eras, design paradigms, transportation technologies, topography, and economics on US street network density, resilience, and connectedness.
- The median average circuity is lower across the neighborhoods data set than across the municipal set, which in turn is lower than across the urbanized areas set.
- This analysis finds a strong linear relationship, invariant across scales, between total street length and the number of nodes in a network.
- The spatial signatures of the Homestead Act, successive land use regulations, urban design paradigms, and planning instruments remain clearly etched in these cities’ urban forms and street networks today.
Conclusion
- First, it presented empirical urban morphological findings from metric and topological analyses of the street networks of every US city/town, urbanized area, and Zillow neighborhood—particularly focusing on density, connectedness, and resilience.
- Second, its methods demonstrate the use of OSMnx as a new street network research toolkit, suggesting to urban planners and scholars new methods for acquiring and analyzing data consistently and at scale.
- Third, it built on past findings about the distribution of street segment lengths and the relationship between the total street length and the number of nodes in a network.
- This study hasmade all of these network datasets—for 497 urbanized areas, 19,655 cities and towns, and 6857 neighborhoods—along with all of their attribute data and morphological measures available in an online public repository for other researchers to study and repurpose.
Did you find this useful? Give us your feedback
Citations
477 citations
203 citations
60 citations
17 citations
Cites background or methods from "A Multi-Scale Analysis of 27,000 Ur..."
...Boeing (2017a) furthermore shows that statistics of network measures significantly change when the scale of study switches from neighborhood to cities and metropolitan areas....
[...]
...Recent tools such as the one proposed by Boeing (2017b) provide algorithms to operate an extraction of network topology....
[...]
9 citations
Cites background from "A Multi-Scale Analysis of 27,000 Ur..."
...Shannon entropy indicates that the more types of things there are and the more equal each type’s proportional abundance is, the less predictable the type of any single object will be (Boeing 2018b)....
[...]
...Network science provides a lens to explore structure through connectivity (Jiang 2016; Boeing 2017, 2018a; Turnbull et al. 2018)....
[...]
...Boeing (2018b) adapts entropy measures to analyze urban spatial order through street orientations....
[...]
References
18,415 citations
17,647 citations
"A Multi-Scale Analysis of 27,000 Ur..." refers background in this paper
...…OSMnx-calculated measures are discussed here, but extended technical definitions and algorithms can be found in e.g. (Albert and Barabási, 2002; Barthelemy, 2011; Brandes and Erlebach, 2005; Costa et al., 2007; Cranmer et al., 2017; Dorogovtsev and Mendes, 2002; Newman, 2003, 2010; Trudeau, 1994)....
[...]
14,696 citations
"A Multi-Scale Analysis of 27,000 Ur..." refers background in this paper
...Finally, PageRank ranks nodes based on the structure of incoming links and the rank of the source node (Agryzkov et al., 2012; Brin and Page, 1998; Chin and Wen, 2015; Gleich, 2015; Jiang, 2009)....
[...]
10,567 citations
"A Multi-Scale Analysis of 27,000 Ur..." refers background in this paper
...…OSMnx-calculated measures are discussed here, but extended technical definitions and algorithms can be found in e.g. (Albert and Barabási, 2002; Barthelemy, 2011; Brandes and Erlebach, 2005; Costa et al., 2007; Cranmer et al., 2017; Dorogovtsev and Mendes, 2002; Newman, 2003, 2010; Trudeau, 1994)....
[...]
3,551 citations
"A Multi-Scale Analysis of 27,000 Ur..." refers background in this paper
...Such measures extend the toolkit commonly used in urban form studies (Ewing and Cervero, 2010; Talen, 2003)....
[...]
Related Papers (5)
Frequently Asked Questions (2)
Q2. What have the authors stated for future works in "A multi-scale analysis of 27,000 urban street networks: every us city, town, urbanized area, and zillow neighborhood" ?
This study hasmade all of these network datasets—for 497 urbanized areas, 19,655 cities and towns, and 6857 neighborhoods—along with all of their attribute data and morphological measures available in an online public repository for other researchers to study and repurpose.