Extracting the multimodal fingerprint of urban transportation networks
Luis Natera, Federico Battiston, Gerardo Iñiguez, Michael Szell
EExtracting the multimodal fingerprint of urban transportationnetworks
Luis Guillermo Natera Orozco , Federico Battiston , Gerardo Iñiguez , , , Michael Szell , , * ) Department of Network and Data ScienceCentral European University, H-1051 Budapest, Hungary ) Department of Computer Science, Aalto University School of Science, FI-00076 Aalto, Finland ) IIMAS, Universidad Nacional Autónoma de México, 01000 Ciudad de México, Mexico ) NEtwoRks, Data, and Society (NERDS), IT University of Copenhagen, DK-2300 Copenhagen, Denmark ) ISI Foundation, 10126 Turin, Italy ) Complexity Science Hub Vienna, 1080 Vienna, Austria June 8, 2020
Abstract
Urban mobility increasingly relies on multimodality, combining the use of bicycle paths, streets, andrail networks. These different modes of transportation are well described by multiplex networks. Here wepropose the overlap census method which extracts a multimodal profile from a city’s multiplex transportationnetwork. We apply this method to 15 cities, identify clusters of cities with similar profiles, and link this featureto the level of sustainable mobility of each cluster. Our work highlights the importance of evaluating all thetransportation systems of a city together to adequately identify and compare its potential for sustainable,multimodal mobility.
Keywords: multiplex networks, sustainable transport, multi-modal transportation * Corresponding author: [email protected] a r X i v : . [ phy s i c s . s o c - ph ] J un . RESEARCH QUESTION AND HYPOTHESIS The infrastructure of different modes of transportation can be described as a mathematical object, the multiplextransport network (Morris and Barthelemy, 2012; Strano et al., 2012; Barthelemy et al., 2013; Battiston et al., 2014;Gallotti and Barthelemy, 2014; De Domenico et al., 2014; Strano et al., 2015; Aleta et al., 2017; Lee et al., 2017). Acity’s multiplex transport network contains the layer of streets and other coevolving network layers, such as thebicycle or the rail networks, which together constitute the multimodal transportation backbone of a city. Due tothe car-centric development of most cities (Jacobs, 1961), streets form the most developed layers (Gössling et al.,2016; Szell, 2018) and define or strongly limit other layers: For example, sidewalks are by definition footpathsalong the side of a street and make up a substantial part of a city’s pedestrian space (Gössling et al., 2016). Similarly,most bicycle paths are part of a street or are built along the side. Yet, the different layers of a multimodal networktypically serve as diverse channels to permeate a city. Here we consider the transport networks of 15 worldcities and develop an urban fingerprinting technique based on multiplex network theory to characterize thevarious ways in which transport layers can be interconnected, identifying the potential for multimodal transport.Using clustering algorithms on the resulting urban fingerprints, we find distinct classes of cities, reflecting theirtransport priorities.
2. METHODS AND DATA
We acquired urban transportation networks from multiple cities around the world, defined by their administrativeboundaries, using OSMnx (Boeing, 2017). These data sets are of high quality (Haklay, 2010; Girres and Touya,2010) in terms of correspondence with municipal open data (Ferster et al., 2019) and completeness (Barbosa-Filhoet al., 2017). The various analyzed urban areas and their properties are reported in Table 1. Figure 1 shows thedifferent network layers for Manhattan, one of our analyzed cities.
Pedestrian Bicycle Rail Street PopulationNodes Links (cid:104) k (cid:105) Nodes Links (cid:104) k (cid:105) Nodes Links (cid:104) k (cid:105) Nodes Links (cid:104) k (cid:105) Amsterdam 23,321 33,665 2.89 34,529 35,619 2.06 1,096 1,655 3.02 15,125 21,722 2.87 872,680Barcelona 20,203 30,267 3.00 7,553 7,647 2.02 249 249 2.00 10,393 15,809 3.04 1,600,000Beihai 2,026 2,978 2.94 0 0 0.00 59 62 2.10 2,192 3,209 2.93 1,539,300Bogota 81,814 121,038 2.96 9,760 9,651 1.98 166 165 1.99 62,017 91,197 2.94 7,412,566Budapest 73,172 106,167 2.90 10,494 10,318 1.97 1,588 1,964 2.47 37,012 52,361 2.83 1,752,286Copenhagen 30,746 41,916 2.73 13,980 13,988 2.00 276 369 2.67 15,822 20,451 2.59 2,557,737Detroit 47,828 78,391 3.28 3,663 3,626 1.98 20 21 2.10 28,462 45,979 3.23 672,662Jakarta 140,042 191,268 2.73 248 231 1.86 58 54 1.86 138,388 188,637 2.73 10,075,310LA 89,543 128,757 2.88 14,577 14,428 1.98 173 221 2.55 71,091 101,692 2.86 3,792,621London 270,659 351,824 2.60 62,398 60,043 1.92 2,988 3,535 2.37 179,782 219,917 2.45 8,908,081Manhattan 13,326 21,447 3.22 3,871 3,777 1.95 349 436 2.50 5,671 9,379 3.31 1,628,701Mexico 108,033 158,425 2.93 5,218 5,278 2.02 370 364 1.97 95,375 140,684 2.95 8,918,653Phoenix 111,363 157,075 2.82 35,631 35,979 2.02 105 138 2.63 73,688 102,139 2.77 1,445,632Portland 50,878 72,958 2.87 24,252 24,325 2.01 230 340 2.96 35,025 49,062 2.80 583,776Singapore 82,808 110,612 2.67 12,981 12,947 1.99 683 740 2.17 50,403 66,779 2.65 5,638,700
Table 1.
Measures for the administrative area of analyzed cities. The number of nodes, links and average degree ( (cid:104) k (cid:105) )for each layer in all cities of our dataset are highly diverse due to the varying developmental levels and focus of transport.The range of population in the analyzed cities goes from half million people to ten million people living in Jakarta, thisallows to have a range of different sizes and cover different developmental stages. igure 1. (Map plot left) Multiplex network representation of Manhattan with the four analyzed layers of transportinfrastructure (pedestrian paths, bicycle paths, rail lines, and streets), with data from OpenStreetMap. (Right)
Networkinformation for each layer, number of nodes, links and average degree (cid:104) k (cid:105) . Data and code to replicate the results are available in: ( https://doi.org/10.7910/DVN/GSOPCK ), and:( https://github.com/nateraluis/Multimodal-Fingerprint ).We characterize each city as a multiplex network (Boccaletti et al., 2014; Kivelä et al., 2014; Battiston et al.,2017) with M layers and N nodes that can be active in one or more layers in the system. Layers follow a primalapproach (Porta et al., 2006) where nodes represent intersections (that may be present in one or more layers),and links represent streets (denoted by s ), bicycle paths and designated bicycle infrastructure ( b ), subways, tramsand rail infrastructure ( r ), or pedestrian infrastructure ( p ). Construction of these intersection nodes follows thetopological simplification rules of OSMnx (Boeing, 2017).In a multimodal city, we expect to find many transport hubs that connect different layers, such as train stationswith bicycle and street access, i.e. nodes that are active in different multiplex configurations. Here we propose amethod to assess all such combinations of node activities in the system, helping us to learn how well connecteddifferent modes are. For each city, we build a profile based on the combinations of node activities, and refer toit as overlap census (Figure 2). The overlap census captures the percentage of nodes that are active in differentmultiplex configurations and provides an “urban fingerprint” of its multimodality (Aleta et al., 2017). To definethe overlap census formally, given a multiplex transport network with M layers the overlap census is a vector of (2 M ) − components, which accounts for the fractions of nodes that can be reached through at least one layer.In Fig. 2(a) we show a schematic of how the overlap census is built: taking the multiplex network, andcalculating the percentage of nodes that overlap in different configurations. The multiplex approach addressesthe multimodality of a city: it not only counts how many nodes or links there are in each layer, but it showshow they are combined, revealing the possible multimodal mobility combinations in the city. Understanding thepossibilities for interchange between mobility layers provides us with a better understanding of urban systems,showing us the complexity and interplay between layers.3 igure 2. (a) Schematic of multiplex layers in a city (left) and its transformation to the overlap census (right). In theoverlap census, the vertical red line gives a visual separation of the left from the right half where nodes become activein the street layer. High spikes in the right half indicate car-centricity. (b)
Clusters of cities based on similarity of theiroverlap census. We find six different clusters using a k-means algorithm (coloured areas), which explain more than of the variance. (c)
Overlap census for cities in each cluster. The first one corresponds to Amsterdam (the city withmost active nodes in bicycle-only configurations). The Copenhagen-Manhattan-Barcelona-Portland city cluster has manyactive nodes in pedestrian-only and bicycle-only configurations, representing an active mobility city. The clusters of LosAngeles-Bogota and Mexico-Beihai-Jakarta are car-centric.
3. FINDINGS
Even in a multimodally “optimal” city there will be a high heterogeneity of node activities due to the differentspeeds and nature of transport modes, implying, for example, a much lower density of nodes necessary for a trainnetwork than for a bicycle network. Therefore, a good way to assess a city’s overlap census is by comparing itwith the overlap census of other cities. We find similarities between cities via a k-means algorithm fed with fifteenvectors (one per city), where each vector contains the percentages of nodes active in each possible configuration.The algorithm separates the 15 analyzed cities into six different clusters [Fig. 2(b)].On the left half of the overlap census, we show the configurations in which nodes are not active in the streetlayer, while the right half contains car-related configurations [Fig. 2(c)] . These clusters of cities are useful toexplain similarities in infrastructure planning in different transport development paths (Rodrigue et al., 2013;4ouf and Barthelemy, 2014), with clusters of car-centric urbanization (like Mexico, Beihai, and Jakarta) opposedto clusters that show a more multimodal focus in their mobility infrastructure (like Copenhagen, Manhattan,Barcelona, and Portland). In the extreme cluster that contains only Amsterdam, close to of nodes are activein the bicycle layer, whereas in the Mexico-Beihai-Jakarta cluster more than of nodes are active in thestreet-pedestrian configuration. The concentration of nodes in just one configuration informs not only aboutthe mobility character of the city, i.e. Amsterdam being a bicycle-friendly city, but unveils the importance ofexplicitly considering overlooked layers and their interconnections. For example, Singapore, Budapest, London,and Detroit have two main peaks indicating that most of their nodes are either active in the street-pedestrian oronly in the pedestrian configuration. This is not the case in Los Angeles and Bogota, where the majority of nodesare active in the car-pedestrian combination, i.e. the pedestrians have to share most of the city with cars. Ourmultimodal fingerprint unravels how different transport modes are interlaced, helping identifying which layer (orset of layers) could be improved to promote multimodal, sustainable mobility.To summarize, we propose the new “overlap census” method based on multiplex network theory allowing torigorously identify and compare the multimodal potential of cities.
References
Aleta, A., Meloni, S., and Moreno, Y. (2017). A Multilayer perspective for the analysis of urban transportation systems.
Scientific Reports , 7.Barbosa-Filho, H., Barthelemy, M., Ghoshal, G., James, C. R., Lenormand, M., Louail, T., Menezes, R., Ramasco, J. J., Simini,F., and Tomasini, M. (2017). Human Mobility: Models and Applications.Barthelemy, M., Bordin, P., Berestycki, H., and Gribaudi, M. (2013). Self-organization versus top-down planning in theevolution of a city.
Scientific Reports , 3(1):2153.Battiston, F., Nicosia, V., and Latora, V. (2014). Structural measures for multiplex networks.
Physical Review E - Statistical,Nonlinear, and Soft Matter Physics , 89(3):1–16.Battiston, F., Nicosia, V., and Latora, V. (2017). The new challenges of multiplex networks: Measures and models.
EuropeanPhysical Journal: Special Topics , 226(3):401–416.Boccaletti, S., Bianconi, G., Criado, R., del Genio, C. I., Gómez-Gardeñes, J., Romance, M., Sendiña-Nadal, I., Wang, Z., andZanin, M. (2014). The structure and dynamics of multilayer networks.Boeing, G. (2017). OSMnx: New methods for acquiring, constructing, analyzing, and visualizing complex street networks.
Computers, Environment and Urban Systems , 65:126–139.De Domenico, M., Solé-Ribalta, A., Gómez, S., and Arenas, A. (2014). Navigability of interconnected networks under randomfailures.
Proceedings of the National Academy of Sciences , 111(23):8351–8356.Ferster, C., Fischer, J., Manaugh, K., Nelson, T., and Winters, M. (2019). Using OpenStreetMap to inventory bicycleinfrastructure: A comparison with open data from cities.
International Journal of Sustainable Transportation , pages 1–10.Gallotti, R. and Barthelemy, M. (2014). Anatomy and efficiency of urban multimodal mobility.
Scientific Reports , 4(1):6911.Girres, J. F. and Touya, G. (2010). Quality Assessment of the French OpenStreetMap Dataset.
Transactions in GIS , 14(4):435–459.Gössling, S., Schröder, M., Späth, P., and Freytag, T. (2016). Urban Space Distribution and Sustainable Transport.
TransportReviews , 36(5):659–679.Haklay, M. (2010). How good is volunteered geographical information? A comparative study of OpenStreetMap and ordnancesurvey datasets.
Environment and Planning B: Planning and Design , 37(4):682–703.Jacobs, J. (1961).
The death and life of great American cities . Random House, New York, New York, USA.Kivelä, M., Arenas, A., Barthelemy, M., Gleeson, J. P., Moreno, Y., and Porter, M. A. (2014). Multilayer networks.
Journal ofComplex Networks , 2(3):203–271.Lee, M., Barbosa-Filho, H., Youn, H., Holme, P., and Ghoshal, G. (2017). Morphology of travel routes and the organization ofcities.
Nature Communications , 8(1):2229. ouf, R. and Barthelemy, M. (2014). A typology of street patterns. Journal of the Royal Society, Interface / the Royal Society ,11(101):20140924.Morris, R. G. and Barthelemy, M. (2012). Transport on coupled spatial networks.
Physical Review Letters , 109(12):128703.Porta, S., Crucitti, P., and Latora, V. (2006). The network analysis of urban streets : a primal approach.
Environment andPlanning B: Planning and Design , 33(5):705–726.Rodrigue, J.-P., Comtois, C., and Slack, B. (2013).
The Geography of Transport Systems (Third Edition) . Routledge.Strano, E., Nicosia, V., Latora, V., Porta, S., and Barthélemy, M. (2012). Elementary processes governing the evolution of roadnetworks.
Scientific Reports , 2(1):296.Strano, E., Shai, S., Dobson, S., and Barthelemy, M. (2015). Multiplex networks in metropolitan areas: generic features andlocal effects.
Journal of the Royal Society, Interface , 12(111):20150651.Szell, M. (2018). Crowdsourced Quantification and Visualization of Urban Mobility Space Inequality.
Urban Planning , 3(1):1., 3(1):1.