Identification and classification of public transport activity centres in Stockholm using passenger flows data
Introduction
Many urban regions worldwide tend to evolve from monocentric to a more complex distribution of activities. At the inter-urban scale, a polycentric policy was adopted by the European Union to support a balanced territorial development (Walsh, 2012). At the intra-urban scale this pattern is driven by urban sprawl, suburbanization, the emergence of specialized employment clusters, shopping centres or other big visitor attractors located on the urban fringe or the combination of several interlining trends. In some cases, such developments may amount to the emergence of a polycentric or multi-centric urban structure (Davoudi, 2003, Baum-Snow, 2010). Whilst these multi-centric concepts remain contested in the urban geography literature, they are essentially defined by the plurality of urban centres, often driven by the decentralization of people, jobs, and services from the core area to sub-centres. The emergence and the spatial distribution of urban centres are facilitated by the underlying transport network. This paper proposes a methodology to identify and classify activity centres based on transport flow data. The spatial and temporal mobility patterns facilitate the analysis of the centres and their relation to the urban activities and transport networks.
Anas et al. (1998) classified centres into two types — old towns incorporating an expanded urban area and newly spawned centres located at nodes of a transportation network. The growth of the second group is seen as the most popular pattern in changing the urban landscape. There is an extensive literature on the impact of urban forms and land-use distribution on travel patterns, such as the influence of urban agglomeration (Garcia-Palomares, 2010) and polycentric urban structure (Schwanen et al., 2001, Casello, 2007) on modal split and commuting patterns. Based on a spatial economics model, Louf and Barthelemy (2013) showed that the monocentric regime becomes unstable as the population grows and that the number of sub-centres grows sub-linearly with population size. External economics of scale underlie the emergence of urban clusters as the spatial concentration of specialized economic activities could foster innovation, spill-over effects and agglomeration benefits. In contrast to these studies, Shearmur and Coffey (2002) questioned the attempts to generalize spatial distribution trends and to imply that urban areas converge to a common spatial development trajectory.
Measuring urban structures and the underlying activity patterns is essential for supporting an evidence-based spatial planning policy. The identification of activity centres and their clustering and characterization will provide planners and policy makers with a better understanding of the existing metropolitan structure and enable them to assess how well it corresponds or diverts from planning policies. As pointed out by Meijers (2008) in the inter-urban context, there is an empirical deficit in the context of spatial planning development that should be addressed by applying a more analytical approach. Previous studies stressed the difficulty of obtaining flow data for analysing urban structure and using transport network attributes or topology indicators as proxies (e.g. EPSON, 2004, Rodrigues da Silva et al., 2014). However, the growing availability of ‘big data’ in the transport sector and in particular travel flow data facilitates the spatial and temporal analysis of urban activity.
This paper presents a methodology which consists of two stages: (1) identifying and (2) classifying activity centres based on mobility data. The objective of this study is to examine how the spatial–temporal distribution of public transport passenger flow could be used to reveal the urban structure dynamics in the case of Stockholm, Sweden. The term public transport activity centres (PTACs) is used rather than urban centre or sub-centre because intermediate interchange hubs cannot be distinguished from genuine demand generators and attractors. PTACs are identified by clustering transport nodes according to their spatial proximity and travel flows. The method proposed in this paper could be applied in different spatial contexts and no particular structure – monocentric or polycentric – is assumed from the outset. A sensitivity analysis is performed to investigate the implications of different parameter values on the number of centres, their spatial distribution and the distribution of passenger flows. PTACs are then classified based on their time-dependent flow profile including magnitude, directness and the distribution of incoming and outgoing flows. It is postulated that urban structure and the spatial distribution of activities are manifested through time-dependent flow profile because activity centres with distinguished functions will yield distinctive travel patterns throughout the day.
The remainder of this paper is organized as follows. The following section reviews the literature on measuring urban structure based on urban geography and economics, spatial analysis and transport flow applications. We then propose a methodology for identifying and classifying PTACs based on time-dependent transport flows. Section 4 presents the context of our case study area – Stockholm Metropolitan Area – and the data available for this analysis. The results are presented and discussed in the context of Stockholm regional and transport development in Section 5. The paper concludes with a discussion on our findings in the context of the urban geography of Stockholm and study limitations.
Section snippets
Literature review
The identification of urban structure is subject to extensive research by urban geographers, urban and regional economists and planners. The two prominent approaches for identifying urban centres could be classified into morphological and functional (or interaction-based) methods. These two approaches essentially correspond to the analysis of densities or mobility patterns, respectively.
Most previous studies undertook a morphological approach. Fractal geometry can be used to investigate the
Methodology
Public transport passenger flows are used in this study to identify activity centres. The individual stations in the public transport network are clustered according to their locations and loaded passenger flow. The identified clusters are thereafter classified into classes based on their time-dependent travel pattern. The methodology applied in this paper is an adaptation of existing methodologies for partitioning interaction matrices which is designed to account for the spatial and temporal
Regional planning in Stockholm
The proposed method is applied to the case study of Metropolitan Stockholm (in Swedish: Storstockholm) defined as Stockholm County. Stockholm is positioned between Lake Mälaren and the Baltic Sea and is built up on a big archipelago. Large green areas (30% of the area), lakes and waterways (additional 30%) form geographical barriers that divide the built-up area. With 2.16 million inhabitants and the fastest growth rate, it is the largest metropolitan area in Sweden. The county covers 6500 km2
Identification
The identification algorithm is applied for the case of Stockholm. A sensitivity analysis was carried out in order to assess the impact of the two parameters — the maximum distance between the primary centre of each urban centre and the farthest station, ρ and; the minimum share of flows that are assigned to centres, δ — specification and select their values. Similarly to the morphological cut-off methods discussed in Section 2 (Riguelle et al., 2007, Anas et al., 1998), the number and
Stockholm — features and trends
Unlike most European metropolitan areas (Riguelle et al., 2007, Veneri, 2013), Stockholm primarily expended by developing satellite towns along its expending rapid public transport system rather than the evolution of a pre-existing hierarchical urban system and the absorption of nearby towns. Nevertheless, the trend observed by Veneri (2013) across Europe towards integration rather than expansion is also observed in Stockholm. The satellite towns led to the decentralization of population but
Acknowledgements
The authors are grateful to Stockholm County Traffic Administration (SLL) for providing the data that enabled this study and in particular to Tengblad Kée and his colleagues.
References (58)
Transit competitiveness in polycentric metropolitan regions
Transp. Res. A
(2007)Sustainable new towns: Stockholm's rail-served satellites
Cities
(1995)Urban sprawl and travel to work: the case of the metropolitan area of Madrid
J. Transp. Geogr.
(2010)- et al.
Subcenters in the Los-Angeles region
Reg. Sci. Urban Econ.
(1991) - et al.
Statistical analysis of the Metropolitan Seoul subway system: network structure and passenger flows
Physica A
(2008) - et al.
The evolving network structure of US airline system during 1990–2010
Physica A
(2014) - et al.
Revealing travel patterns and city structure with taxi trip data
J. Transp. Geogr.
(2015) - et al.
Road accessibility and articulation of metropolitan spatial structures: the case of Madrid (Spain)
J. Transp. Geogr.
(2014) The identification of urban employment subcenters
J. Urban Econ.
(1987)- et al.
The number of subcenters in large urban areas
J. Urban Econ.
(2003)
The topography of metropolitan employment: identifying centers of employment in a polycentric urban areas
J. Urban Econ.
Defining functional urban regions in Bahia, Brazil, using roadway coverage and population density variables
J. Transp. Geogr.
Travel behaviour in Dutch monocentric and polycentric urban systems
J. Transp. Geogr.
Estimating a polycentric urban structure. Case study: urban changes in the Stockholm region 1991–2004
J. Urban Plann. Dev.
Urban spatial structure
J. Econ. Lit.
The validity of the monocentric city model in a polycentric age: US metropolitan areas in 1990, 2000 and 2010
Urban Geogr.
Changes in transportation infrastructure and commuting patterns in US metropolitan areas, 1960–2000
Am. Econ. Rev.
An ex-post CBA for the Stockholm metro
Form follows function? Linking morphological and functional polycentricity
Urban Stud.
Contrasting indications of polycentricism within Spain's Metropolitan Urban Regions
Transportation activity centers for urban transportation analysis
J. Urban Plann. Dev.
Avoiding the crowds: understanding tube station congestion patterns from trip data
Multifractal characterization of urban form and growth: the case of Beijing
Environ. Plann. B: Plann. Des.
Polycentricity in European spatial planning: from an analytical tool to a normative agenda
Eur. Plan. Stud.
Characterizing metro networks: state, form and structure
Transportation
Potentials for polycentric development in Europe
Project Report
Discovering events in the city via mobile network analysis
J. Ambient. Intell. Humaniz. Comput.
From fractal urban pattern analysis to fractal urban planning concepts
Comput. Approaches Urban Environ.
Network accessibility and employment centres
Urban Stud.
Cited by (49)
Unravelling the spatial properties of individual mobility patterns using longitudinal travel data
2022, Journal of Urban MobilityVoting with one's feet: Unraveling urban centers attraction using visiting frequency
2022, CitiesCitation Excerpt :Several past studies went beyond identifying urban centers into systematically characterising those. Cats et al. (2015) proposed a two-step approach for identifying and characterising sub-centers. First, sub-centers are identified using a hierarchical method.
Effects of land use on time-of-day transit ridership patterns
2022, Transportmetrica A: Transport ScienceI-index for quantifying an urban location's irreplaceability
2021, Computers, Environment and Urban SystemsCitation Excerpt :It is generally represented by an ordered point pair of origin (O) and destination (D) (LeSage & Pace, 2008; Shu et al., 2021). Researchers have proposed a range of flow-based locational measures including both simple ones such as inflow, outflow, total flow (Cats, Wang, & Zhao, 2015; Sun et al., 2016), net flow, and net flow ratio (Guo et al., 2012; Xu et al., 2017), and more sophisticated ones such as centrality (Hughes, 1993; Zhong et al., 2014) and entropy (Limtanakool, Schwanen, & Dijst, 2009). Overall, these locational measures are beneficial for the discovery of spatial patterns in flow data and provide insights into the roles of locations in generating interactions.
Multi-modal network evolution in polycentric regions
2021, Journal of Transport GeographyCitation Excerpt :Many urban regions worldwide have evolved or are in the process of being transformed from a monocentric development into a more complex spatial configuration which consists of a multiplicity of urban centers. A wealth of empirical studies has illustrated the emergence of polycentric developments across the world and how sub-centers can be identified using various data such as satellite observed lighting (Tselios and Stathakis, 2020), 3D remote sensing building density (Taubenböck et al., 2017), transport network infrastructure (Liu et al., 2016), residence and workplace locations (Vasanen, 2012), aggregate transport demand flows (Cats et al., 2015) and individual mobility traces (Cats and Ferranti, 2021). There is no consensus in the literature not only in regards to how to measure the spatial parameters of polycentric regions, but even on how to overall define those as well as methodological issues associated with determining their geographical demarcation (Burger and van Oort, 2008) and identifying the number of centers involved (Zhang and Derudder, 2019).
How to identify urban centers/subcenters in metropolises? An analytical review
2021, Habitat International