Decoding Urban Archetypes: Exploring Mobility-Related Homogeneity among Cities

Abstract

To make cities more sustainable and livable and to achieve climate targets in transportation, cities around the globe must undergo sustainable transformations. However, disparities in initial conditions pose challenges when trying to implement these sustainable changes. Identifying these differences aids in the comprehension of future developments. In this study, we establish an international comparison by decoding the mobility-related characteristics of cities and determining urban archetypes. Using publicly accessible data, we analyze and classify 96 cities in different countries. Therefore, we utilize principal component analysis to simplify the data. The emerging components serve as input for segmentation. This approach yields nine unique urban archetypes, ranging from Well-Functioning and Ancient Hybrid Cities in Europe to Paratransit and Traffic-Saturated Cities in the southern hemisphere. Our results show that there is a significant advantage to using a multidimensional segmentation basis, which we identify in an extensive literature review. The result is a finer segmentation, which is especially clear for European cities that demonstrate four different clusters. We discuss that the effect of future restrictions on private car usage will vary widely between the urban archetypes.

1. Introduction

The rapid growth of urban areas is leading to uncertainty around the future of motorized individual mobility worldwide. According to the United Nations, the proportion of the population living in cities in 2018 is predicted to double by the year 2050 [1], creating a growing demand for mobility currently not covered by the existing transport infrastructure. Faced with a range of challenges, cities are looking to transform urban mobility by reducing the negative externalities commonly attributed to excessive car usage. Objectives include a reduction in air pollution and greenhouse gas emissions, as well as the implementation of car-free city centers making urban areas more sustainable and livable. However, this sustainable transformation requires a range of specific measures that vary from city to city. Several cities have shown early indications of reconfiguring their built environment around non-motorized transport modes by implementing novel initiatives like the establishment of “superblocks” in Barcelona, the enlargement of the cycling network in Milan and London, or the deceleration of car traffic in Vienna [2].

Nevertheless, this transformation could face challenges to meeting the population’s mobility demands, especially when alternative travel options to private motorized modes are lacking. In order to gain a better understanding of the status quo of urban mobility, a comparison of cities and their different preconditions is necessary. The implementation of this requires a unified and consistent approach. While some studies exist, no comprehensive comparative database currently focuses specifically on motorized individual mobility and its role within urban areas.

This study aims to fill this gap by identifying urban archetypes that can help cities improve their existing infrastructure and develop sustainable mobility solutions. Each archetype refers to a type of city that summarizes the common mobility-related features of their underlying urban areas. We pose the following research questions:

What are crucial dimensions of urban mobility in order to identify urban archetypes of cities?

How do these urban archetypes differ in terms of the role of the car in urban spaces, and how may this change in the future?

To answer these research questions and to identify relevant dimensions as a basis for segmentation, we perform a comprehensive literature review of different approaches to the segmentation of cities. Based on these dimensions, our study compares 96 cities in various countries and continents and classifies them into similar urban archetypes using an explorative clustering method. Cities are described using a set of 26 indicators, with a strong focus on urban mobility.

The following sections describe the outcome and conclusions of our analyses. They are structured as follows: after a literature review and an outline of the database, we describe our methodology for identifying urban archetypes. An in-depth analysis of the cluster and the various formed urban archetypes follows. Based on the cluster solution, we take a closer look at the archetypes specifically located within the European region. This approach will deliver deeper insights regarding future car use and restrictions in Europe. Finally, we discuss our results and draw a conclusion.

2. Literature Review

Clustering is a common and scientifically valid approach when investigating differences and commonalities between cities. More than 80 years ago, Harris [3] was the first researcher to empirically differentiate 984 American cities based on national employment statistics. Using a rule-based cluster algorithm, cities were assigned to one or multiple of the following classes, defining their core industrial functions: Manufacturing, Transportation, Tourism, Wholesale, Retail, Education, Mining, or Diversified. This clustering method established itself as a standard procedure in subsequent studies [4,5].

Bruce and Witt [6] supplemented this analysis of socio-economic indicators with socio-demographic parameters, defining 13 economic city types. Similar analyses of urban prosperity are still relevant to this day [7,8,9]. Hill et al. [8], for example, implemented a segmentation of 508 cities to analyze the regional distribution of wealth across the US. Furthermore, Martin et al. [9] describe a segmentation of 300 cities to support state-level policy decisions. Cities within the same cluster are supposed to react as similarly as possible to fiscal interventions.

In addition to various economic classifications, other approaches provide information on the different geographical features to describe the spatial patterns, layouts, and structures that characterize a city [10,11,12,13]. Huang et al. [11] identified seven spatial metrics to describe said measures of urban shaping using satellite imagery from 77 cities. The four archetypes that emerged in the process indicate urban shapes that vary greatly depending on their region. Thompson [12] used data from 30 megacities to illustrate how different urban systems attempt to solve local traffic problems. The five archetypes, Full Motorization, Weak Centre, Strong Centre, Low Cost, and Traffic Limitation, define distinct forms of urban infrastructure.

A growing body of available data and methodological developments continues to provide new insights into various areas of urban research [14,15,16,17]. For example, Kenworthy and Laube [18] published the Millennium Cities Database for Sustainable Transport, the first international comparison of cities based on a wide range of indicators of urban mobility. The data collection from 1995 contains 230 standardized indicators from 100 cities.

In addition to analyzing the sustainability of various urban transport systems in Kenworthy [19], the Millennium Cities Database forms the data basis of various other studies [20,21]. Based on the database, Priester et al. [22] identified 13 relevant factors for the quantitative mapping of urban mobility in 41 megacities. These factors include their car dependency, traffic fatalities, traffic congestion, and public transit use. The study lists a total of seven city typologies: Paratransit Cities, Auto Cities, Non-Motorized Cities, Hybrid Cities, Traffic-Saturated Cities, Low-Motorized Cities, and Transit Cities.

Oke et al. [23] presented one of the most comprehensive urban segmentations in mobility research, subdividing 331 cities in 124 countries. Relevant indicators include urban shaping, economic performance, and urban mobility behaviors. The broad spectrum of the study is intended to support a large number of actors from politics and business in the development of new mobility solutions and an associated reduction in emissions. In total, the study presents 12 clusters.

A segmentation on a national level can be found in Klinger et al. [24]. The study used data from 44 German cities to cluster different “urban mobility cultures”. In addition to established dimensions such as the socio-demographic and economic characteristics, as well as the travel behavior of the population, the study additionally considered various psychographic indicators to represent the subjective dimensions of urban mobility. Subjective dimensions are described as those characteristics whose assessment depends exclusively on the individual perception of the surveyed population. The multidimensional consideration of objective and subjective indicators describes, in part, strong contradictions between using certain mobility offerings and their general association in the population. However, the lack of comparable psychographic data collection essentially prevents a comparison within international studies.

The role of urban sustainability and energy use is also the focus of several other classifications. Shell [25] used six city archetypes to forecast future urban energy demand. Acuto et al. [26] analyzed sustainability-promoting measures in terms of their sectoral and economic origins. Cantuarias-Villessuzanne et al. [27] identified three “smart city strategies” based on seven predefined “smart city dimensions” [28].

In addition to the areas of use illustrated so far, cluster analyses possess a high economic relevance. The procedure is often used to identify suitable target markets for new products and services [29,30,31]. For example, Lang et al. [30] used an analysis of the urban spatial structure and socio-economic parameters to identify suitable cities for the deployment of autonomous vehicle fleets.

In contrast to the beforementioned approaches, various other segmentations are implemented within individual cities. These usually form part of analyses of the intra-urban spatial structure [32,33] or the urban population [34,35]. As a rule, however, the objective of these cluster analyses is not a direct comparison of cities. For this reason, they will not be discussed further in the remainder of this study.

The abovementioned indicators within the described segmentation approaches can be summarized into five dimensions: 1. Socio-economic indicators provide information on a city’s economic development and its influence on social, political, ecological, and spatial processes. 2. Socio-demographic indicators enable a more precise description of the urban population, its social structure, and the regional level of development of a city. 3. Geographical input values describe the spatial layout of a city. The orientation of the transport network or divergent land use are established measures for segmentation. Urban mobility is of central importance in this work. 4. The dimension of Mobility describes the dependence on individual means of transport or congestion of the regional infrastructure. These form essential indicators in earlier clustering approaches. 5. Psychographic indicators map subjective motives within the urban population. This dimension is key to achieving a holistic view of urban areas.

Figure 1 gives a non-exhaustive overview of the different segmentation approaches, facilitating a better comprehension of the various indicator dimensions. Missing arrows to individual groups do not mean that such approaches do not exist. Often, a clear separation of approaches is only possible to a limited extent.

Figure 1. Overview of the conceptual approach of existing urban segmentations [3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,22,23,24,25,26,27,29,30,31].

Our review shows that there is currently no segmentation approach, to our knowledge, that covers all five dimensions within an international comparison of cities. However, in order to gain a better understanding of the different manifestations of urban mobility, a holistic approach is required. This includes the selection of a comprehensive database as well as the consideration of a diverse set of cities. The following section provides a detailed description of the applied methodology.

3. Data Collection

The selection of appropriate indicators is a crucial challenge in the clustering process, closely related to the choice of cities to be considered in the study. As we aimed to achieve a multidimensional segmentation framework, we faced two constraints on selecting indicators: (1) the selected indicators had to represent various areas of the urban environment and (2) the choice of cities had to enable an international comparison. Varying data sources can demand a weighing up of the inclusion of certain indicators against a subsequent reduction in the total number of comparable cities. The goal of achieving a diversified analysis is additionally met with the issues outlined in Murphy’s “curse of dimensionality” [36], which describes an increase in complexity associated with the inclusion of more factors. Hence, a too broad database can have negative implications for the interpretability of the solution. A preliminary step was taken to eliminate highly correlated values within an initial set of 57 indicators to counteract this issue. The resulting dataset used to segment the cities is summarized in Table 1, comprising 26 distinct indicators. The selected indicators encompass all five dimensions identified in the previous chapter, namely socio-economic, socio-demographic, psychographic, geographic, and mobility (see Table 1). Notably, all values, except psychographic indicators, pertain to the corresponding administrative levels of the examined cities. The data were normalized to a uniform scale through min–max scaling to enable comparability of all indicators. In total, a set of 96 cities was found to provide sufficient data and serve as our initial base for segmentation. Of these cities, 43 are located in Europe, 17 in North America, and 19 in the Asian/Pacific region. The remaining 17 cities are mainly situated within the southern hemisphere. A complete listing of all cities under consideration can be found in Table A1 in Appendix A.

Table 1. Overview of the indicators used in the clustering process.

Indicator	Unit/Range	Description	Year	Source
Mobility Dimension
Micromobility sharing options	0–3	Number of categorically different micromobility offerings within a city	2023	NUMO [37]
Modal split bike	%	Share of trips made by bicycle	2017–2020	UITP [38] Dixon et al. [39]
Modal split car	%	Share of trips made by car	2017–2020	UITP [38] Dixon et al. [39]
Modal split public transit	%	Share of trips made by public transit	2017–2020	UITP [38] Dixon et al. [39]
Motorization rate	0–9999	Average number of registered cars per 1000 inhabitants	2018–2021	Eurostat [40] Cities official statistics
Numbeo traffic index	0–1000	Composite index of time spent in transit and the corresponding dissatisfaction with time spent traveling	2023	Numbeo [41]
Time lost in traffic	%	Additional time spent on transport due to traffic compared to a city’s baseline conditions without congestion	2022	TomTom [42]
Geographic Dimension
Built-up area per capita	m2	Area built up by covered surfaces per capita	2015	European Commission [43]
Geographical agglomeration	%	Size of a city’s urban area, including connectivity to nearby cities and suburbs, with 100% indicating a densely populated area and 0% indicating an isolated location of the city	2023	OpenStreetMap [44]
Urban street network orientation	%	Quantifying a city’s adherence to a single-grid road network, with 100% indicating full adherence to geometric principles and 0% indicating no adherence at all	2019	Boeing [45]
Socio-economic Dimension
Access to public transit	%	Share of population with access to a public transit stop within walking distance (500 m for buses and/or 1000 m for rail-based services)	2020	UN Habitat [46] Data-Driven EnviroLab [47]
Buildings over 35 m per 1,000,000 inh.	0–1000	Number of buildings with a height of more than 35 m in the urban area per 1,000,000 inhabitants (inh.)	2023	Skyscraper Source Media [48] World Population Review [49]
CO2 emissions per capita	t	A city’s carbon emissions per capita	2018	Moran et al. [50]
Cost-of-living index	0–200	Living cost compared with the average cost of living in New York City (=100); a city with a cost-of-living index of 120 is estimated to be 20% more expensive than New York City	2023	Numbeo [41]
Gini coefficient	%	Measure of statistical dispersion intended to represent income or wealth inequality	2022	The World Bank [51]
GRP per capita	USD	Monetary measure of the market value of all final goods and services produced and sold per resident	2021	Neffke et al. [52]
Innovation index rating	0–60	Assessment of the baseline conditions for innovation in a city, including digital transformation, technology applications, startups, sustainability, and others	2022	2thinknow [53]
License caps	0; 1	Existence of regulatory policies limiting the issuance of new license plates to residents	2020	Zhuge et al. [54]
Metro stations per 1,000,000 inh.	0–1000	Number of metro stations per 1,000,000 inhabitants	2023	Rohde [55] World Population Review [49]
Monthly costs of public transit	USD	Cost of a monthly pass for the usage of public transit	2023	Numbeo [41]
Rail network length per 1000 inh.	m	Total length of all tracks of a city’s railway network (metro and tram) per 1000 inhabitants	2023	Rohde [55] World Population Review [49]
Road pricing	0; 1	Existence of charging mechanisms for the use of parts of the urban road network	2023	Sadler Consultants [56]
Student city index	0–100	Assessment of the best urban destinations for international students	2023	Quacquarelli Symonds [57]
Traffic fatalities per 100,000 inh.	0–100,000	Number of annual traffic fatalities per 100,000 inhabitants	2022	ITF [58]
Socio-demographic Dimension
Population growth per year	%	Annual increase in the number of inhabitants within a city	2023	World Population Review [49]
Psychographic Dimension
Individualism index *	0–100	Quantification of the preference for a loose social fabric in which the individual only cares for oneself and one’s immediate family	2001–2020	Hofstede Insights [59]

* Indicator value on national level.

4. Methodology

In this section, we present our methodological approach to identifying urban archetypes. First, we applied a principal component analysis (PCA) to reduce the complexity of the 26 indicators that served as inputs for our segmentation. Second, we performed a clustering in order to segment cities.

4.1. Principal Component Analysis

The 26 indicators used as segmentation criteria were intended to address different aspects of a city. As a preliminary analysis of our segmentation, and to reduce the complexity, we used a PCA method based on the selected indicators. Before performing the PCA, we conducted Bartlett’s test of sphericity (<0.001), which indicates that the data are suitable for the PCA as that analysis proves that the correlation matrix, which underlies the analysis, is statistically significant [60]. Table 2 shows the results of the PCA. Based on the scree plot (elbow criterion) and Kaiser’s criterion, five factors were extracted: one describing the car dependence, one the prosperity, and one the traffic maturity. In addition, we identified a fourth factor regarding traffic saturation and a fifth regarding the quality of public transit.

Table 2. Loadings of the selected indicators on the five principal components.

	Principal Component
	1	2	3	4	5
Indicator	Car Dependence	Prosperity	Traffic Maturity	Traffic Saturation	Quality of Public Transit
Cronbach-α	α = 0.83	α = 0.88	α = 0.81	α = 0.63	α = 0.72
Motorization rate	0.746	−0.102	0.199	0.034	−0.067
Individualism index	0.716	0.175	0.245	0.024	0.004
Modal split public transit	−0.711	0.208	0.280	0.331	−0.056
Modal split car	0.875	0.045	−0.216	0.098	−0.042
Built-up area per capita	0.785	0.190	−0.146	−0.025	0.018
Urban street network orientation	0.494	0.187	−0.406	0.203	0.093
License caps	−0.479	0.393	−0.407	−0.267	−0.054
Cost-of-living index	0.246	0.674	0.377	−0.229	−0.104
GRP per capita	0.331	0.579	0.255	−0.160	0.176
CO2 emissions per capita	0.294	0.618	−0.150	−0.007	−0.061
Monthly costs of public transit	0.382	0.620	0.227	−0.106	−0.073
Innovation index rating	−0.104	0.856	0.054	0.146	0.126
Student city index	−0.146	0.783	0.438	0.053	−0.206
Geographic agglomeration	−0.239	0.465	−0.317	0.067	0.274
Access to public transit	−0.132	0.037	0.804	−0.045	0.074
Traffic fatalities per 100,000 inh.	0.096	−0.262	−0.752	−0.003	−0.099
Annual population growth	−0.114	−0.132	−0.673	−0.075	−0.067
Gini coefficient	0.169	−0.074	−0.560	0.426	−0.079
Modal split bike	−0.391	0.027	0.030	−0.668	0.181
Time lost in traffic	−0.486	−0.049	0.038	0.595	0.009
Numbeo traffic index	0.052	−0.029	−0.380	0.722	0.053
Road pricing	−0.239	0.244	0.254	0.566	0.022
Metro stations per 1,000,000 inh.	−0.100	0.158	0.099	0.029	0.834
Buildings over 35 m per 1,000,000 inh.	0.247	0.371	−0.134	0.188	0.518
Micromobility sharing options	0.067	−0.211	0.373	−0.012	0.550
Rail network length per 1000 inh.	0.023	−0.095	0.154	−0.212	0.733

The indicators that have the strongest possible effect on the respective factors are used to describe each principal component. For this purpose, all indicators that show a loading >0.45 are considered [61]. The factor car dependence indicates the degree to which an urban transportation system requires the use of motorized individual mobility. The concept of traffic maturity elucidates the degree to which equitable access to secure and publicly available transportation alternatives is available to all individuals. Traffic saturation characterizes the congestion of the urban road network. Quality of public transit describes the availability of high-quality alternatives to cars through rail-bound offerings (e.g., metro). Prosperity gives an impression of a city’s economic and technological possibilities.

All five factors were found to have sufficient internal consistency (reliability), with a Cronbach’s alpha of 0.83 (car dependence), 0.88 (prosperity), 0.81 (traffic maturity), 0.63 (traffic saturation), and 0.72 (quality of public transit), respectively. The low value for Cronbach’s alpha of traffic saturation (0.63) only shows acceptable reliability [62]. While this is sufficient overall, it must be considered in further analyses.

4.2. Clustering

Segmentation approaches can be broadly classified into supervised (a priori) and unsupervised (post hoc) methodologies. The former is applicable when the segments are known or predefined and aims to categorize new observations into these segments. The latter finds its use when the objective is to identify homogeneous groups within multivariate data [60,63]. In transportation, k-means cluster analysis is one of the most used unsupervised cluster analysis methods [34,35,64,65,66]. It is also used in previous city segmentations [16,24,26]. Therefore, we used the method to identify city archetypes based on the five principal components. Using the cluster packages factoextra and bios2mds in the software R (v4.3.1), the silhouette score [67] and the elbow criterion [68] of the underlying data were determined. Both values implicate a good fit of a 9-cluster solution.

5. Results

The combination of five principal components and an unsupervised cluster analysis yielded nine distinct clusters out of 96 cities, each representing a unique urban archetype with discernible characteristics related to a specific group of cities within our database. A complete listing of all cities and their respective archetypes can be found in Table A1 in Appendix A.

In the following, these urban archetypes are analyzed by comparing their means for the cluster characteristics. Based on these specific attributes, we further describe and name each archetype.

5.1. Cluster Description

Table 3 presents the expressions of the five principal components and the mean values of all individual indicators for each identified cluster. CL 1 shows the lowest value in traffic saturation and the lowest time lost in traffic. Additionally, CL 1 and CL 2 exhibit a well-established traffic system with superior accessibility to public transit. CL 3 has a low car dependence with a low motorization rate and good access to public transit. In contrast, CL 4 has a strong car dependence and rather high motorization rates. Notably, CL 3 and CL 4 exhibit only negligible divergence in terms of traffic saturation. A high level of prosperity and excellent rail network infrastructure distinguishes CL 5. However, it also exhibits an elevated reliance on automobiles, implying a hybrid urban structure. CL 6 exhibits the least reliance on cars. Nonetheless, the level of traffic maturity remains significantly low. Moreover, this cluster exhibits the second-highest rate of traffic fatalities and the second-poorest accessibility to public transit. CL 7 boasts the third-highest traffic maturity level and excellent public transit access. However, a limited number of rail-bound mobility offerings indicates a high reliance on road-based public transit, mainly carried out by buses. CL 8 is characterized by the poorest accessibility to public transit and the lowest score for traffic maturity, coupled with the highest frequency of traffic fatalities. CL 9 displays the highest level of traffic saturation, with minimal dependence on cars. The travel time loss in this cluster can be attributed to various other modes of transport.

Table 3. Characteristics of the nine clusters based on the five main principal components and averages of their respective cluster-forming indicators.

Cluster (CL)	1	2	3	4	5	6	7	8	9
Principal Components
Car dependence	0.03	0.15	−1.12	1.46	1.32	−1.77	0.62	−0.53	−1.10
Prosperity	0.16	−0.81	1.37	0.68	1.63	0.39	0.36	−1.21	−0.79
Traffic Maturity	0.78	0.80	0.47	−0.57	−0.55	−2.02	0.57	−1.47	−0.11
Traffic Saturation	−1.29	−0.16	0.43	0.29	0.81	−0.77	−0.19	0.33	1.76
Quality of public transit	0.89	−0.04	−0.48	−0.10	2.24	−0.14	−1.00	−0.71	−0.51
Cluster-forming Indicators *
Motorization rate per 1000 inh.	377	478	249	563	422	208	454	250	297
Individualism index	68.4	60.2	44.3	82.4	91.0	20.0	81.4	39.2	34.7
Modal split public transit [%]	22.6	26.8	51.9	13.7	14.9	30.1	18.9	25.2	48.2
Modal split car [%]	34.4	45.1	21.7	75.4	75.2	23.5	55.4	39.9	23.8
Built-up area per capita [m2]	170	131	88	333	335	74	211	72	61
Urban street network orientation [%]	4	7	5	44	42	20	9	11	9
License caps [% of cluster]	0	0	14	0	0	80	0	0	0
Cost-of-living index	77.8	59.9	82.2	73.5	85.0	47.9	82.0	38.0	35.9
GRP per capita [USD]	57,161	40,465	51,214	56,357	69,954	23,342	45,829	21,184	23,821
CO2 emissions per capita [t]	8.29	6.31	14.9	15.43	16.6	4.94	10.92	6.6	3.42
Monthly costs of public transit [USD]	82.1	40.1	69.5	81.5	100.8	34.0	100.2	23.8	30.1
Innovation index rating	44.9	41.2	51.1	47.6	52.0	47.8	44.4	37.7	42.8
Student city index	71.1	59.4	92.0	64.3	72.6	61.7	80.5	48.7	61.1
Geographic agglomeration [%]	41	26	76	51	81	86	28	30	44
Access to public transit [%]	93	91	86	61	76	53	91	50	80
Traffic fatalities per 100,000 inh.	1.72	2.53	2.07	6.39	5.04	11.84	3.02	18.73	7.38
Annual population growth [%]	0.69	0.47	0.4	1.01	0.9	2.17	1.12	2.22	0.94
Gini coefficient [%]	30.6	32.9	34.1	38	41.5	38.2	34.3	42.7	44.2
Modal split bike [%]	17.7	4.1	6.6	1.2	1.4	14.6	4	4.5	2.4
Time lost in traffic [%]	24.7	28.8	35.9	26.7	29.6	38.0	31.5	34.8	46.6
Numbeo traffic index	101.1	131.8	148.9	188.7	203.4	165.6	155	204.4	222.7
Road pricing [% of cluster]	12	17	71	11	43	0	8	8	67
Metro stations per 1,000,000 inh.	40.6	17.8	18	8.6	71.9	14.2	1.6	4.8	6.5
Buildings over 35 m per 1,000,000 inh.	140	54.2	100.9	274.5	630.4	27.8	69.8	22.7	35.4
Micromobility sharing options	2.12	2.11	1	1.22	1.71	1	1.17	0.67	1.11
Rail network length per 1000 inh. [m]	141.5	58.6	26.6	30.9	143.4	23.5	17.8	7.6	10.7

* Description of indicators in Table 1.

Based on the analysis of the cluster-forming indicators, we can summarize each cluster’s characteristic features (see Table 4). In addition, each cluster is given a concise name to simplify the understanding of clusters as urban archetypes.

Table 4. Overview of the urban archetypes and their characteristics.

	Cluster/ Urban Archetype	Characteristics	Key Example Cities
1	Well-Functioning Cities (n = 18)	▪ Good access to public transit in the form of rail and bus offerings ▪ Highly efficient mobility while retaining a moderate motorization rate ▪ Evenly distributed usage of all transport modes with a notably high bicycle usage	Amsterdam, Copenhagen, Rotterdam, Munich
2	Ancient Hybrid Cities (n = 18)	▪ Despite having high access to public transit within the city, the car still plays a dominant role ▪ Cities are often geographically isolated from other urban areas	Madrid, Rome, Barcelona, Athens
3	Transit Cities (n = 7)	▪ High reliance on public transit ▪ Often using regulatory measures to reduce car usage (e.g., road pricing)	Tokyo, Paris, Singapore, London
4	Auto Cities (n = 8)	▪ Car makes up almost a single mode of transport ▪ Congestion levels remain moderate due to car-centered infrastructure	Los Angeles, Toronto, Dubai, Houston
5	Hybrid Commuter Cities (n = 8)	▪ Car dependence remaining high due to low expansion of rail network in outer regions of the city ▪ Reliance on car puts significant stress on mobility infrastructure	New York, San Francisco, Chicago, Boston
6	Dynamic Megacities (n = 5)	▪ More than 60% of all trips are walking or public transit trips ▪ Common usage of regulatory measures (e.g., license caps) to reduce the amount of newly registered cars to ease stress on infrastructure	Beijing, Shanghai, Shenzhen, Guangzhou
7	Car and Bus Cities (n = 11)	▪ Heavily reliant on motorized mobility ▪ Public transit operates mostly through bus services ▪ Car continues to be of great importance, especially for inhabitants for trips out of the cities	Montreal, Auckland, Sydney, Manchester
8	Paratransit Cities (n = 12)	▪ Low coverage of public transit leads to various challenges in cities’ mobility ▪ Rapidly growing population puts existing mobility services under stress	Kuala Lumpur, Johannesburg, Bangalore, Delhi
9	Traffic-Saturated Cities (n = 9)	▪ Public transit mostly covered by buses ▪ Few alternative options for other modes of transport direct all traffic onto the road network	Mumbai, Istanbul, Mexico City, Buenos Aires

To facilitate a better comprehension, Figure 2 provides an overview of the manifestation of the five primary components in each urban archetype and highlights some pertinent cluster representatives. A detailed comparison of the individual cities based on their principal component values can be found in Figure A1, Figure A2, Figure A3 and Figure A4 in Appendix B.

Figure 2. Expressions of the five principal components in each urban archetype.

5.2. Geographical Location

The geographic distribution of the nine urban archetypes exhibits substantial heterogeneity within the locations of their associated cities. Figure 3 illustrates an accumulation of Well-Functioning and Ancient Hybrid Cities located in Europe, which also encompasses a diverse mix of Transit Cities as well as Car and Bus Cities. Auto Cities and Hybrid Commuter Cities are predominantly situated in North America, while all South American cities fall within the Traffic-Saturated categorization. Paratransit Cities predominantly exist in eastern regions, with a few individual representatives in Africa. The highest degree of regional concentration is observed in the exclusively Chinese Dynamic Megacities cluster. Most of the remaining cities are categorized as Transit Cities and Car and Bus Cities, with the latter primarily concentrated near coastal areas. Notably, this finding warrants further investigation.

Figure 3. Geographical location of the nine clusters.

5.3. Cluster Validation

A content validation approach is implemented to ensure the cluster solution’s conformity with established city segmentations. This entails examining the intersections of the identified urban archetypes with the existing city segmentation proposed by Priester et al. [22]. The selected study is particularly suitable for a comparative analysis, as it shows considerable overlap in the selection of indicators and its main points of interest, emphasizing the role of motorized individual mobility in urban areas. An essential precondition for sampling is that the cities must be included in both the study by Priester et al. [22] and our study: this applies to a subsample of 35 cities. The segmentation proposed by Priester et al. [22] was compiled based on a sample of 41 cities. Figure 4 visualizes the breakdown of the two studies in a Sankey diagram. Here, transitions from one segmentation approach to the other are shown in proportion to quantity. The greater the match between two segments, the broader the quantity flows between the two groups.

Figure 4. Sankey diagram comparing the distribution of the urban archetypes across the city typologies from Priester et al. [22].

The Sankey diagram shows strongly overlapping cluster assignments of the two segmentation approaches. Particularly noteworthy is a distinct subdivision of the city typology “Hybrid Cities” from Priester et al. [22]. The cluster is divided into a total number of seven urban archetypes. Other urban archetypes, such as Dynamic Megacities and Traffic-Saturated Cities, are consistent with their respective counterparts in Priester et al. [22].

The more detailed subdivision of the large city typology Hybrid Cities can be considered a significant advantage of our urban archetypes’ methodology. This highlights that using all dimensions listed in Section 2 enables a more comprehensive consideration of cities. Overall, the comparison indicates a good consistency of the urban archetypes with the existing study of Priester et al. [22]. We identify similar clusters with similar compositions of cities, which demonstrates that the cities possess a certain stability in their development.

6. Discussion

6.1. Future Trajectories of Urban Archetypes

The classification of urban archetypes is a temporal snapshot, prompting inquiry into the evolution of cities with respect to their future car use. Regional differences are apparent; archetypes have distinct development trajectories and face various challenges. European cities are showing an increased willingness to transform through a multitude of mobility options, coupled with a deprioritization of the car. In line with their ambitious climate targets, European policymakers intend to pursue these targets at the urban level. European cities like Ancient Hybrid Cities and Well-Functioning Cities prioritize a livable city through road space allocation moving towards improved bicycle infrastructure and high-quality dwelling areas. Cities place less emphasis on examining the traffic-related impacts of measures when it serves the goal of a more livable city.

Due to their high population density and absolute number of people, Asian cities have fundamentally different needs. In order to attain livability and a more sustainable transport system, they need to focus on leveraging efficiency and stringent regulations. Predominantly Asian archetypes, such as Dynamic Megacities or Transit Cities, initially prioritize the efficiency of their transportation systems due to high demand. Smart city approaches are especially relevant here. Improvements in the cities’ livability are a subsequent result.

In contrast, North American archetypes like Auto Cities and Hybrid Commuter Cities primarily focus on managing road traffic, lacking sufficient mobility alternatives. Taken together, transformational activity in North American archetypes tends to be low.

The examination of potential development trajectories is particularly intriguing when it comes to car use. European cities, consisting of four different urban archetypes, show the greatest degree of uncertainty regarding their future car usage. Therefore, a comprehensive exploration of the European archetypes will be conducted below.

6.2. In-Depth Analysis of European Cities

For the final discussion of the urban archetypes, the Well-Functioning Cities, Ancient Hybrid Cities, Public Transit Cities, and Car and Bus Cities clusters represented in Europe will be used as examples to explain the extent to which the segmentation of cities can be used to create uniform strategies and mobility solutions. As an example, an analysis of the role of the private car in the respective clusters will be carried out. The extent to which the degree of car dependence of each cluster is related to its regulatory constraint on car use needs to be examined. Cities use a wide variety of regulatory measures to reduce urban car use (urban vehicle access restrictions, UVAR). Within the framework of the ReVeAL project (Regulating Vehicle Access for Improved Liveability) initiated by CIVITAS [56,69], the possible actions of cities are divided into five groups. A distinction is made between road user charges, environmental zones, zero-emission zones, emergency measures that allow cities to impose restrictions on car use under certain conditions, and other restrictions on traffic spaces [56]. To better understand the differences in car dependence within the selection of European cities, their respective principal component value is scaled relative to the distribution within the smaller sample group (n = 43).

Figure 5 shows a qualitative ranking of archetypes according to their regulation of urban vehicle access and car dependence. A complete list of the measures employed by each city can be found in Table A1. The archetype Transit Cities with the lowest car dependence also shows the highest intensity of regulatory measures. In addition to the presence of road pricing, environmental and zero-emission zones are established measures to accelerate the transition towards an electric car fleet in London. Paris follows suit, announcing the introduction of initial zero-emission zones in 2030. Within Europe, Car and Bus cities form the counterpart of Transit Cities. The highest car dependence of the comparison archetypes is paired with the lowest regulation of urban car use at the same time. Individual cities use environmental zones in their urban core. However, except for the city of Dublin, these remain passable. Overall, we see an expected negative correlation between car dependence and restrictions. Strong alternatives to the car are necessary to obtain any room for maneuvering at all.

Figure 5. Qualitative ranking of the car dependency and number of regulatory measures within the clusters represented in Europe.

7. Limitation of the Approach

The methodology applied and the discussion of the results presented facilitate a better understanding of urban archetypes around the world. Through our segmentation, we are able to identify strong differences between cities and their potential for increased livability and sustainable transport. Nevertheless, our findings should also be considered against the backdrop of their limitations. Firstly, in this study, we examined only 96 cities. The restriction in terms of the number of cities analyzed and their regional distribution results from the study’s broad database for segmentation. This database is undoubtedly an advantage in terms of the quality of the results. Nevertheless, the unavailability of data for a larger number of cities is a limitation. It is likely that we could find additional archetypes, for example, by including more cities in Africa. Secondly, we only captured cities at one point in time within our study. Our discussion shows that push and pull measures can change cities and their classification. Thirdly, we look at cities only as a whole, neglecting different archetypes of neighborhoods or municipalities within the cities. Moreover, due to limited data accessibility, we were restricted to analyzing cities at the administrative level, which poses a constraint on the comparability of the cities. Fourthly, with deterministic clustering methods, an object, e.g., a city, can only be assigned to exactly one cluster, even if it also fits into another cluster to a certain extent. This limitation must be taken into account when considering individual cities in the clusters. There are also differences within the clusters.

8. Conclusions

Enriching existing city segmentations, we segmented 96 globally distributed cities, based on a selection of 26 indicators, into nine urban archetypes. This indicates the extent to which cities differ regarding their mobility needs and challenges. This differentiation required a comprehensive understanding of the various aspects affecting urban mobility. Our findings address this first research question by stating the critical dimensions of urban mobility in order to identify urban archetypes of cities. The comprehensive literature review shows that five dimensions are relevant for achieving a holistic city segmentation. Due to the broad database of 96 cities, these dimensions could be considered within five principal components. A comparison with the study of Priester et al. [22] shows the strength of our approach, as a more differentiated representation of Hybrid Cities is possible. To answer the second research question, our findings also show what role the private car currently plays in different urban areas and how this may be subject to change in the future. Based on the results of the cluster analysis, we illustrated the influence of these differences in urban areas on the development of uniform strategies. A subsequent comparison of the archetypes within the European area also validated the usability of our segmentation in a smaller, regionally more limited setting. Urban transformation trajectories vary, and we see an expected negative correlation between car dependence and future regulatory measures to reduce car usage. Strong alternatives to the car are necessary to obtain any room for change at all.

Due to their broad database, the urban archetypes provide a well-suited base for further analyses of the implications of car traffic. The status quo is thereby mapped via the nine clusters. Based on the clusters, various analyses of car use, prosperity development, or urban emission development can be derived in this way. Similarly to the process described in Section 6.2, a holistic analysis of all urban archetypes and their respective restrictions of urban car usage can be conducted in further research.

The assessment of urban archetypes has unveiled geographical variations among North America, Asia, and Europe, which arise from multiple factors such as urban size, the presence of alternatives to the car, weather conditions favoring cycling, and, significantly, the level of maturity of the transportation network. Our study serves as a base for future scientific work focusing on cities and, at the same time, provides an important orientation for practitioners to initiate new urban projects for a more sustainable transport system.

We expect that a proportion of the cities considered in this study will continue to transform into smart, livable cities, while others will remain in their current status quo for now. This fact must be considered, as continuous monitoring of cities is necessary in order to identify this transformation. Future research consists of forecasting the development of cities and the transferability of measures. For this purpose, further cities must be assigned to urban archetypes via supervised clustering methods.