Navigation – Plan du site

AccueilRubriquesEspace, Société, Territoire2020French urban systems: a relationa...


French urban systems: a relational approach

Sandrine Berroir, Nadine Cattan, Frédéric Dobruszkes, Marianne Guérois, Fabien Paulus et Céline Vacchiani-Marcuzzo
Traduction de Alvin Harberts
Cet article est une traduction de :
Les systèmes urbains français : une approche relationnelle


This article explores the interactions between the French cities from a multi-dimensional and multi-scale approach. Most of the time, these relations are analyzed in a segmented way, focusing solely on one dimension among a multifaceted reality. To take into account the great diversity of exchanges, we simultaneously consider several types of inter-city links, reflecting individual mobility, economic relations and scientific partnerships, on which we apply flows selection methods and a scoring approach. These methods also allow distinguishing several interdependent scales of territorial dynamics: first, twenty six local urban systems are identified. Comparisons between these systems are conducted according to their more or less polarized organization and to the nature of the flows within these systems. On the other hand, the horizontal relations between regional urban centers are analyzed taking into account the range and diversity of interactions and also according to the way local and regional urban systems connect to each other. Finally, relations with Paris are highlighted depending on the specialization of these links and the size of the cities involved.

Haut de page

Notes de la rédaction

This article is the translation of: Les systèmes urbains français : une approche relationnelle

Texte intégral

From cities and conurbations to urban systems

1The issue of what is at stake for cities in a globalised system of flows is intricately linked to certain postulates that are at the basis of urban theories and principles of territorial planning action. Today, neither the city nor the metropolis give any information on the territorial and metropolitan dynamics in progress. In a world dominated by exchange, a change of perspective in know-how is necessary because territorial changes question the very meaning of urban and metropolitan spaces defined in very narrow perimeters, following an areolar reading and management of territories, in contiguity and continuity.

2To understand the changes that are taking place in urban and metropolitan systems is to understand that we have entered the era of mobile territorialities and networked societies. Only a relational approach to spaces makes it possible to integrate the transformations that territories are going through today. In this perspective, territories must be thought in terms of interdependence and articulation and no longer in terms of distribution and location. This conceptual posture leads to centering the debate on public policies in territorial systems of interfaces that are considered in the combination of place and link, network and territory, gateways and corridors.

3While all experts agree that space is network, it must be recognized that the integration of mobility and links is still incomplete in the design of the organization and development of territories. This research offers to define territorial dynamics within the prism of an approach by urban system where the very foundation of the urban is the link, an approach which conceives territories explicitly by the relations which articulate the cities between them. In terms of action strategies, this refers to a reflection on partnerships, complementarities and cooperation between territories.

4Today, knowledge of the spatial and territorial configurations of urban systems is partial, even biased. The works on the reality of the networking of French cities are certainly numerous, but remain scattered and dispersed. They are always segmented according to the nature of the link between cities or continue to approach trade in a sectoral manner according to a single mode of transport.

5This research aims at combining the figure of the network city with that of the city-territory. The study goes beyond taking into account flows that are always considered as structuring according to an approach that favours an economic interpretation of territorial development. It broadens the spectrum of links by integrating exchanges that are often perceived as non-structuring. In addition, it is developing an unprecedented comprehensive approach to urban systems that simultaneously crosses very diverse types of interurban exchanges, thereby giving great robustness to the reticulated configurations highlighted.

6This multidimensional analysis of urban systems shows that an approach of spatial planning and development based on proximity is outdated. The organization and action frameworks are multi-scale and three territorial levels are involved in each urban system in France: proximity, inter-metropolitan transversality and connection to Paris.

7The article first presents a state of the art knowledge on city networks. In the second part, it explains the approach and methods used to develop this comprehensive relational approach, which makes it possible to jointly integrate several types of exchanges (7 indicators) and analyse them together between all French urban areas1 (350). The spatial configurations of urban systems at the intersection of the three levels of interdependence are then discussed.

Dominant approaches to urban systems

8The works on urban systems in France and on urban networks on a European scale are numerous. The topic of their structuring raises questions for scientists and spatial planners. In most of these works, metropolises and the largest cities in general are at the center of concerns because they are considered, in an economic approach to territories, as the major drivers of development. As a result, they are perceived as essential vectors for the dynamics and networking of territories. However, in a large body of work on French or European urban systems, the focus is essentially on the concentration of populations, executives, senior corporate services activities, headquarters of transnational firms, equipment or international institutions. This certainly makes it possible to provide knowledge on the spatial organisations of cities. But these works invariably lead to the reproduction of the same dual images of the organisation of spaces: a dominant center and dependent peripheries (Cattan, 2009, 2010). As a result, they do not show the real change that "the territory of networks gives way to a real networked territory" (Veltz, 1996), a reality which, according to the author, is a challenge to representation and imagination. As an echo, Peter Taylor (2002) denounces the paradox of research on cities: while the very essence of world cities lies in their mutual links, he stresses that the work is too often reduced to case studies or comparative analyses evaluating the strengths and weaknesses of a selection of cities. A synthesis of the existing literature on French urban systems in their national or European connections leads to distinguish four main approaches.

Classifying cities

9The first approach, which is overwhelmingly dominant, is "benchmarking". The aim is to evaluate the strengths and weaknesses of cities on the basis of different socio-economic criteria and thus to set up classifications and typologies of cities. Major international organisations (OECD), European programmes (ESPON), national or regional institutes (IAU-Ile-de-France, FNAU, etc.), national administrations (DATAR/CGET) but also private organisations (MasterCard, Global Power City Index Mori Memorial Foundation, etc.) or certain national or international press bodies (e.g. Monocle) attach the greatest importance to this. They are also supported by the work of researchers who often respond to the demand for evaluation of these systems from the above-mentioned organisations and institutions (Cattan et al., 1994; Vandermotten, 2009; Rozenblat, Cicille, 2003; Taylor, 2004; Halbert et al., 2012). In these studies, the approaches are typological and based mainly on stock variables. They lead to the elaboration of categories of cities, to propose urban ranking lists of the most attractive cities for the populations (quality of life, residence, etc.), or the most global cities (headquarters of multinationals, air traffic, foreign investments, etc.). This work provides interesting models for understanding how cities position themselves in global, national and regional economies. They also make it possible to better qualify the diversity of metropolitan profiles by taking broader account of urban functions (Halbert et al., 2012). However, they sometimes offer an incomplete view of urban systems since in most cases, only the largest cities are taken into consideration.

Evaluating inter-urban accessibility

10The second approach questions the role of transport infrastructure in the form of urban networks. These studies focus on differentiated accessibility to various metropolitan services and functions (Spiekermann, Wegener, 2008). They also make it possible to evaluate temporal proximities, such as possible long-distance links over a day (L'Hostis, Bozzani, 2010). In these accessibility assessments, other works measure space-time contraction and induced changes in urban hierarchies (Bretagnolle et al., 2002). These works often have an innovative methodological aim, which is concretized by original cartographies such as anamorphoses or multimodal chronocards (Cauvin, 2000; L'Hostis, 2003). A number of them put the effects of Euclidean distance into perspective and stress the importance of taking time into account when understanding networks and travel.

Modeling city dynamics in a system

11The third perspective of urban systems analysis is part of a theoretical and modelling approach. Research in this framework places long time at the center of settlement systems analysis (Pumain, 2000; Bretagnolle et al., 2007, 2009; Paulus, 2007; Sanders, 1997). They highlight the strong adaptive capacities that characterize these settlement systems, in particular by referring to the path dependence effects. The notion of resilience and robustness of these complex systems are particularly deepened by modelling and validating the factors and mechanisms that contribute to this sustainability. In many cases, this work questions the diffusion of innovation and the interaction between cycles of economic growth and urban growth, in different types of urban systems, in order to highlight what concerns similar processes and what is characteristic of national specificities (Pumain et al., 2015).

Thinking of cities in a relational approach

12The relational approach of urban systems differs from the approaches presented above by explicitly emphasizing the links between places. The vast majority of the work under such an approach develops sectoral analyses of trade. Most often, it is people's mobility that is retained as an indicator structuring relations between cities, such as home-work commuting (Berroir et al., 2008; Gingembre, Beaude, 2014), residential migration (Baccaïni, 2005) or air passenger flows (Cattan, 2004; Dobruszkes, 2008; Derruder et al., 2010), depending on the level of analysis.

13In order to better integrate the challenges of the information and knowledge society, other studies mobilise more specific linkage indicators, such as scientific partnerships (Berroir et al., 2005, 2009; Comin, 2009; Massard, 2003), student mobility (Baron, Berroir, 2007) or co-publications between researchers (Matthiessen et al., 2006; Grosseti et al., 2013). The networks highlighted by these links give rise to more diversified modes of structuring urban systems. By placing cities in a more polycentric network, these analyses call into question the very foundations of spatio-economic theories, in their principles of polarization, competition or dependence.

14Moreover, intangible exchanges, such as telephone flows, social networks or the Internet, have become essential for understanding the networking of cities. Yet research is still emerging, due to the difficulty of accessing data. A landmark study in this field has led to regionalisations of urban systems based on interurban telephone flows, an aspect never taken into consideration at these levels (Damette, 1994). The exceptional growth in these flows shows how interesting it is to approach urban systems with this type of indicator (Blondel et al., 2009).

15This last approach reveals a necessary change in perspective to better grasp the diversity of exchange channels, which is at the very basis of the complexity of territorial systems.

Addressing urban systems in their relational dynamics

  • 1 In the definition given by INSEE, based on the perimeters defined in 1999, there are 354 urban area (...)
  • 2 French urban systems are analysed within the national framework. It would certainly have been desir (...)
  • 3 In this paper, the terms "urban area" and "city" will be used equivalently by convention.

16In this article, we propose an approach to territorial dynamics through an analysis of the relationships between urban spaces that highlights multi-dimensional and multi-scalar functioning, defined as the combination of place and link, territory and network. This approach takes the side of considering all urban areas1 as French urban systems2, because we consider that the capacity to articulate networks of relations does not depend exclusively on large conurbations. Urban areas are increasingly becoming a reference unit for defining what a city3 is today, since mobility is an intrinsic dimension of the urban environment. This reference is also part of a European and international context that privileges the functional area as a relevant urban territory.

17It is thus a question of presenting an encompassing approach of the territories which integrates a broad spectrum of the modalities of the exchange. By going beyond segmented and sectoral visions of territorial dynamics, it reveals the existence of "systemizing" networks. It is therefore a question of restoring the diversity of the spatial organisation of cities into systems in terms of functional dimensions, territorial scales and membership of networks. This perspective has rarely been developed in research to date.

Description of indicators

18The identification of urban systems is based on a selection of exchanges that structure territorial systems. Seven indicators were taken into account simultaneously to describe urban functioning in all its diversity and thus cover the three major and complementary facets of contemporary societies (Table 1).

Table 1: The seven indicators of links between French cities taken into account

Table 1: The seven indicators of links between French cities taken into account

19The mobile and leisure society constitutes the first facet and refers to the recompositions of social times. The mobility of people, through the great variety of reasons, duration and frequency of the journeys they cover, constitutes an essential indicator for analysing the territorial integration of a group of cities at different scales, ranging from living and employment areas to larger territories. Three main indicators reflect this:

  • Home-work commuting are usually considered to be the mobility that most strongly structures the organisation of territories at a local level. Even if these business trips are now in the minority compared to leisure and shopping trips, the location of employment continues to play a major role in travel chains and in the definition of metropolitan polarities. At larger ranges, long-distance commuters, although relatively few in number, tend to occupy an increasing place as a result of the metropolisation process, the increase in travel speeds and the increase in dual activity in couples (Ginger, Baude, 2014).

  • Residential migration (Figure 1) is also a major indicator of mobile and leisure society, in a context of increased fluidity of interurban traffic and greater professional mobility. On an intercity scale, they are closely linked to the phenomenon of long-distance commuters, which are often either a prerequisite for permanent migration or the consequence of a change of residence. Moreover, they may reflect the maintenance of a social network between the old and the new place of residence, or indirectly reflect economic links between cities where different establishments of the same company are located.

    • 4 This index expresses the monthly number of aircraft and TGV connecting two urban areas in spring 20 (...)

    High-speed systems ensure the networking of cities on a larger scale and respond to increasingly varied travel demands. The TGV and the airplane are the two modes of transport capable of transporting passengers over long distances and quickly between cities. A mixed index4 has been created to identify opportunities to travel between two French cities.

Figure 1 Population mobility through residential migration

Figure 1 Population mobility through residential migration

20The second major facet of our contemporary societies, the knowledge and information society can be captured by a set of emblematic flows of new communication and research practices:

  • In an information and knowledge society where access to knowledge has become a decisive factor in the competitiveness and attractiveness of territories, scientific partnerships and the networks they create are a good indicator of territorial change. The research projects of the Research and Development Framework Program (RDFP) have been selected here. Strongly encouraged by public policies and supported by the European Commission, they network various public and private institutions.

  • Economic links in the innovation sector specifically (Figure 2) are a second indicator for addressing a driving process in the evolution of urban systems that generates mobility. Seized by the link between the head office and the establishments of the same group in the specific sectors of innovation, this indicator reveals little-known relational dynamics, more sensitive to niche effects, where the functional specialisation processes of the territories would play a major role.

Figure 2 Economic links through headquarters-establishments links in innovative sectors

Figure 2 Economic links through headquarters-establishments links in innovative sectors

21The third facet concerns the economic and financial aspects that give meaning to the flows of a globalised tertiary society:

  • The most emblematic links are those between the different locations of the establishments of the same company, in particular between its head office and its other establishments, all sectors of activity combined. These links highlight networks of local economic actors and reveal the differentials in the attractiveness of the areas taken into account in the groups' location strategies.

  • At the intersection of economic aspects and the recomposition of social times, second homes (Figure 3) reflect the multiplicity of spaces and temporalities of life. Contrary to popular belief, these residences are not confined to rural areas and coastlines but are located up to 20% in urban areas. The study of the principal residence of secondary residents is rarely addressed; no study of these links exists at the national level, let alone between urban areas.


Figure 3 Leisure mobility by primary and secondary residences

Figure 3 Leisure mobility by primary and secondary residences

From indicators to urban systems: methods for cumulating and identifying systems

23The method developed to identify urban systems using a multidimensional (seven interaction indicators) and multiscalar (three territorial levels to identify urban systems) approach is based on a three-stage analysis of the links between the 354 French urban areas.

Cumulative Indicators

  • 5 This method was preferred for its ease of understanding, compared to statistical methods for data a (...)

24The first step consists in synthetically assessing the intensity of long-distance relationships. It is based on the implementation of a new method of cumulating flow values5, using a scoring value. Assuming that these indicators are all equally important, we apply the same weight to each of them. In addition, to account for the unequal value of flows, each indicator was discretized into five value classes ranging from 0 (for the 20% of the lowest flow values) to 4 (for the 20% of the highest values). Each city pair was therefore characterized by seven partial scores (one score per indicator). It is the sum of these scores that made it possible to define the final score corresponding to the overall intensity of long-distance relations, all indicators combined. Finally, among these links, only the most robust, combining several exchange modalities were retained as constituting an urban system: the score which characterizes them must thus be maximum (4) for at least three of the seven exchange modalities, meaning that urban areas have effective exchanges in several socio-economic domains.

Different flow measurements for different geographical scopes

25Beyond the choice of a cumulative method, we have introduced a distinction in the measurement of cumulated links, which depends on the geographical scope of the relationships:

  • For long-range relations, the selection of the most massive links between distant cities was based on thresholds of volumes of flows exchanged. Indeed, it is these main flows that best highlight long-distance connections, both for transverse links between major regional cities and links to and from Paris. In both cases, the scoring was applied directly to the volume of trade flows, taking into account the effects of city weights on trade intensity.

  • For proximity linkages, we have on the other hand emphasized the preferential links developed between cities, by eliminating size effects. It was the adjustment to a gravity-inspired interaction model that made it possible to evaluate the preferential links between two cities when the observed flows were greater than those estimated by the model, taking into account the emission and reception capacities of the cities (measurement of Chi-2 residuals). Proximity systems were therefore identified in this case by applying the scoring method to preferential links.

Delimitation of proximity systems

  • 6 There are more sophisticated methods (multigraphs) to identify territorial systems taking into acco (...)

26After selecting the local relationships, a third step consisted in identifying the system relationships and presenting the strongest local interdependencies. This delimitation of proximity systems was based on a method linked to graph theory and based on the selection of major flows from each urban area (Nystuen, Dacey, 1961 ; Cattan, 2004 ; Drevelle, 2012)6. In the end, this method makes it possible to identify about twenty local urban systems, each of the 354 French urban areas being integrated into a more or less extensive urban system.

  • 7 Paris is understood as the urban area of Paris, whose perimeter goes beyond the administrative limi (...)

27The different methods used thus make it possible to highlight the variable configurations of the three levels constituting interdependencies between urban areas, at the crossroads of several exchange modalities. Three scales of interdependence constitute urban systems (Figure 4): proximity relations (section 4), transverse relations at the national level and connections with Paris7 (section 5).

France's local urban systems

26 local urban systems

  • 8 See Figure 4 for the case of the Paris urban proximity system.

28Twenty-six local urban systems can be identified, outside Paris' urban area8 (Figure 4.a). All French urban areas are integrated into a system. The smallest system includes five urban areas (Auxerre system, with 200 000 inhabitants), the largest thirty (Lyon-Grenoble system, with four million inhabitants). The configurations of these proximity systems are generally within the limits of the (former) 22 administrative regions, without the statistical model having a priori influenced in this direction. Indeed, the vast majority of links connect urban areas in the same region: this is the case in particular in the two Norman regions, Picardy and Lorraine where only one urban system emerges. On the other hand, in Provence Alpes Côte d'Azur, Brittany and Rhône-Alpes, two urban systems remain well identified within the regions. However, other systems depart from this observation and are structured at the intersection of several regional groups. Two cases attest to these interregional connections. The first concerns small towns located at regional borders that are linked to the urban system of a neighbouring region, such as Angoulême, Royan and Cognac, towns south of Poitou-Charentes that are connected to the Bordeaux urban system or Laval and Mayenne, towns in the Pays de la Loire linked to the Rennes urban system. The second is confirmed by the links between the major poles of two urban systems located in different regions, which can be qualified as transregional. This is the case with the Besançon-Dijon system and, to a lesser extent, the Mans-Alençon system. On this scale of interdependencies, the Besançon-Dijon system is the only one that prefigures the boundaries of the new division into 13 regions (law voted in January 2015).

29The urban area of Paris integrates the five local systems of Rouen, Caen, Orléans-Tours, Amiens and Le Mans-Alençon which have become constitutive of the great "Parisian" system. The latter has 70 urban areas and a total population of nearly 17 million (Figure 4.a). Other local urban systems close to the capital region are not included in this vast system. These are the Auxerre, Reims and Troyes systems. This does not reveal an absence of links (notably home-work links) but the cities of these systems do not have a sufficient diversity and intensity of links with Paris to be integrated into the Parisian system. This old and well-known asymmetry in relations between the west and east of the Paris Basin is confirmed here by the multidimensional approach to exchanges between cities (Damette, 1994; Gilli, 2005).

  • 9 The intensity of the links varies from 0 to 28 (7 indicators * 4 classes)

Figure 4 The three scales of interdependence of French urban systems9

Figure 4 The three scales of interdependence of French urban systems9
  • 10 The relative centralisation index was developed by network sociologists. Its objective is to evalua (...)

30Beyond this diversity of forms of extension, proximity systems present varied spatial organisation structures, which reflect contrasting operating methods in networks. The research and policy communities are always looking for an optimal form of urban networks to improve the efficiency of their operation both in economic and governance terms. The respective advantages of more or less polarized models remain at the center of the debates. A synthetic image of these structures is given by the degree of polarization of each of the 26 systems (Figure 5). This measure, derived from graph theory (relative centralisation index) varies from 0 (low system polarisation) to 1 (maximum polarisation)10 (Degenne, Forsé, 2004, p. 154).

Figure 5 The variety of configurations of urban proximity systems

Figure 5 The variety of configurations of urban proximity systems

31Overall, five subsets can be distinguished. Bordeaux appears as the most polarized system followed by Limoges, Reims and Clermont-Ferrand, with a very mono-centric pattern, this first group can be represented by the image of the star. A second group, composed of Toulouse, Strasbourg, Montpellier, Auxerre and Marseille, is also very polarized by the main city but significant links connect the other cities of the system. Moderately polarized, i.e. developing diversified and numerous links on privileged axes between at least two urban areas of the system, the systems of Brest, Pau-Bayonne, Annecy and Caen for example form the third subgroup. The operation is more polycentric in Dijon-Besançon, Orléans-Tours, Lyon-Grenoble, as in Nantes-Saint-Nazaire. The five systems of Nice, Amiens, Le Mans-Alençon, Tours and Poitiers are distinguished by a very low degree of polarisation meaning a very strongly meshed organisation of the system. The Parisian system is weakly polarized since only 16 links pass through the capital, out of the hundred that the system counts.

Qualifying local urban systems

  • 11 Results from a Hierarchical Ascendant Classification (with Euclidean distance) based on the share o (...)

32The nature of the linkages and their association within each system provides information on the degree of diversity or specialization of urban systems11 (Figure 6). Very specialised relationships on a single type of indicator are rare, with the exception of those based on scientific partnerships and high-speed transport. For example, Marseille is strongly connected with Avignon and Toulon by high-speed rail. The exchanges between Rennes and Lannion but also Vannes and Laval are distinguished by a marked specialization in scientific partnerships. Beyond these rare and specific links, three less specialized types characterize long-distance exchanges. Among these very numerous links, the first type is based more on economic specialisations, the second is characterised by the crossing of links in innovative economic sectors and home-work commuting, while the third combines population, residential and leisure mobility.

33On the scale of local urban systems, the combination of these more or less specialised links shows a fairly great overall diversity within each system. There are two main profiles (Figure 6). The most widespread, which concerns 16 out of 26 systems, combines the three least specialised links previously described, i.e. mobility of people (daily, residential or leisure), with economic operating links encompassing all the more innovative activities and sectors. For example, in this profile we can find the systems of the largest cities, like Toulouse, Bordeaux, Strasbourg or Clermont-Ferrand, and the smallest ones like Reims, Caen and Limoges. The second, more complete and diversified profile, integrating all types of links and in particular scientific links, concerns the remaining ten urban systems: Lille, Rennes, Nantes-Saint-Nazaire, Montpellier, Marseille, Brest and to a lesser extent Lyon, Orléans-Tours, Nice and Pau-Bayonne. Both types of profiles are associated with both polycentric and polarized configurations.

  • 12 CAF takes into account the 7 study indicators (Table 1).

Figure 6 Categorising local urban systems12

Figure 6 Categorising local urban systems12

34As for the urban system of Paris, it has a profile similar to that of most other local systems. The links with Paris are essentially characterized by the relative importance of population, daily, residential and leisure mobility (the Ile-de-France cities of Nemours, Provins, Montereau-Fault-Yonne, Chartres or Beauvais) and economic links between headquarters and establishments in innovative sectors (Orléans, Beauvais, Compiègne or Rouen). More specific links show the importance of high-speed connections with Le Mans or Tours and scientific partnerships with Caen. The links between other cities (outside Paris) have similar characteristics in terms of mobility of people and economic functioning.

Transverse inter-system relations

Longer distance relations

35French urban systems cannot be conceived at this level of proximity alone. The analysis also highlighted the strength of the transverse links between the regional urban centers, the largest in particular, but medium-sized cities are also concerned (Figure 4.b). These links are currently still very little studied. Yet they constitute a major dimension of urban systems. Indeed, the inter-system linkage is very robust: it covers more than four link indicators for a large number of systems, such as Marseille, Nice, Bordeaux, Toulouse, Rennes, Nantes, Lille, Strasbourg, Lyon and Grenoble. Lyon, in particular, and the large cities of the south appear to be major pivots of these transverse relations. This transverse linkage also concerns the urban systems of Tours-Orléans, Clermont-Ferrand and Dijon-Besançon, but to a lesser extent and on a smaller number of indicators. These structures take place in a certain reciprocity between the different urban areas, thus referring to increasingly reticulated and polycentric integration processes at these territorial levels.

36For most local urban systems, the linkage to cross-links is made through the largest city in the system (Figure 7). The intensity of polarization by the main city varies according to regional contexts. Lyon-Grenoble, which has the largest number of transverse links - 53 in total - is characterised by an average polarisation of its transverse links (one third from Lyon). Marseille-Aix also has an average polarization index for its transverse lashing. The five urban systems of Montpellier, Bordeaux, Nantes, Toulouse and Rennes, which rank third in terms of the size of their cross connections, have contrasting interconnection methods. Indeed, if the Montpellier and Rennes systems are anchored on a national scale by several poles that make up their urban system, the other three connect essentially through the largest city in the system, such as Toulouse, which polarizes 85% of the system's links. With less than 15 transverse links, the other systems present very diverse situations, from the very polarized system like Metz-Nancy, Rouen, Clermont-Ferrand and Limoges, to the most diversified such as for Poitiers, Le Mans-Alençon and Troyes.

Figure 7 Transverse links in local urban systems

Figure 7 Transverse links in local urban systems

The transverse links of four systems: Lyon, Toulouse, Montpellier and Nancy-Metz

37The transverse links of four systems were chosen to qualify the modalities of the transverse interconnections, according to the role played by the main urban pole of the system and the scope and variety of the links. The Lyon-Grenoble system covers the entire national territory, being deployed at the same time at a very great distance from the largest metropolises, from many medium-sized cities but also with the major centers of neighbouring regions. Lyon's urban system is much diversified and covers all sectors, in terms of mobility, economic exchanges and scientific partnerships. At great distances, one can distinguish economic links in innovative sectors as well as research partnerships, while the regional anchorage with the cities of the southeast is strongly reinforced by leisure mobility (links between main residences and second homes). Toulouse's urban system, with a smaller number of transverse links, nevertheless presents a range of more selective, far-reaching relationships at the level of the main metropolises of the territory and with a strong regional anchorage. Nevertheless, we can observe that relations are particularly strong in the southern part of the territory, towards Lyon, Bordeaux, Marseille or Pau-Bayonne or Montpellier. The links are mainly in the economic, research and residential mobility sectors. Leisure mobility is more limited in this scope. Montpellier, in third position in terms of total number of transverse links, is a weakly polarised system (less than one link out of two concerns the main pole), allowing the emergence of other anchoring poles, Nîmes and Perpignan in particular, and presenting links of lesser importance, with most of the cities of the south.

38With a small number of total transverse links (10), the Nancy-Metz system is mainly linked to the east, with Strasbourg and only 4 metropolises (Lille, Lyon, Marseille and Toulouse) are linked to Nancy, despite the bipolar configuration of the proximity system. These transverse links are mainly linked to population mobility, especially leisure (with second homes) but also to economic exchanges in innovative sectors with Lyon.

Relations with Paris

39Relations with the urban area of Paris are the third constituent dimension of all French urban systems. They are most often carried out by several poles and not only by the main pole of local urban systems (Figure 4.c). The strong links between the largest urban centers in each of the local systems and Paris must therefore not mask the strong interconnections that also exist between Paris and the medium-sized or even smaller cities in each of the systems. For example, while Nantes has links with Paris on more than five types of exchanges, Saint-Nazaire, Cholet and La Roche-sur-Yon are also linked to Paris by at least four different indicators. A string of medium-sized towns in the Rhône valley, Brittany and Languedoc-Roussillon also develop strong links with Ile-de-France. For some cities, this means that the Paris Region polarization is a driving force behind their integration into the national system as well as in their capacity for international openness. For example, some cities could not have claimed to participate in European programmes without this link with Parisian dynamics. Five urban systems derogate from this more polycentric rule of anchoring with Paris, connecting via the main pole of the proximity system. These are the systems of Nice, Marseilles, Toulouse, Bordeaux and Caen in particular.

40Three types of links are particularly selective in linking French cities with Paris: scientific partnerships, high-speed links and economic exchanges, particularly in innovative sectors. All major metropolises cover the full range of possible exchanges with Paris with maximum link intensities. Medium-sized regional metropolises maintain diversified but less intense links overall. In most cases, one of the three selective links is missing, as in La Rochelle, Limoges or Saint-Etienne. Smaller cities, on the other hand, maintain links with Paris through several types of population mobility, in particular residential migration and second homes.

Elements of conclusion

41By adopting a relational approach that encompasses territories, urban systems are approached in a pioneering way, at the crossroads of a wide range of links. The operating modes of territories appear very diversified because they integrate the multi-dimensional and multi-scale aspects of urban systems. Three levels of interdependence are involved in the organisation of each of the French urban systems: local relations, cross-system relations and connections to Paris. At the local level, the spatial configurations of urban systems are often multipolar, such as for Lille, Lyon-Grenoble, Nantes and Rennes. This means that the urban environment is changing and that the notion of metropolis-network is becoming more and more relevant. At the cross-system level, the confirmation of horizontal pole-to-pole relationships where structuring takes place in relative reciprocity reinforces the perspective of related remote and reticulated urban networks. Still little analysed in the spatial literature and little understood by public actors, these remote systems are a prefiguration of what could be a true polycentric articulation of regional urban areas. Finally, the study also shows that the deployment of these horizontal relations between regional metropolises is entirely compatible, on the one hand, with the maintenance of strong polarisation by the Paris Region metropolis and, on the other hand, with the strengthening of local territories around these metropolises.

42These new representations of urban and territorial dynamics underline the need to take into account, at both national and local levels, all the scales at which interdependencies and linkages between places are deployed. They say that the theoretical and operational conception of the city and the urban environment must be understood both by the networks in proximity and in connections woven by these entities, because each level has a socio-political interest. By jointly integrating the diversity of modes of inhabitation, production and circulation, the urban systems resulting from this research define relevant configurations for changing spatial planning and development policies and thinking new modes of territorial public action that integrate the effects of an increasingly widespread multi-membership of cities in several systems.

Haut de page


Baccaïni B., 2005, "Enquêtes annuelles de recensement : résultat de la collecte 2004 - Des changements de région plus fréquents qui bénéficient aux régions du Sud et de l’Ouest", Insee Première, No.1028, juillet.

Baron M., Berroir S., 2007, "Paris et le système universitaire français : mythe et réalités", Annales de Géographie, n° 655, 3-22.

Berroir, Cattan N., Saint-Julien Th., 2009, "Les masters en réseau : vers de nouvelles territorialités de l’enseignement supérieur en France", L’Espace géographique, No.1, .42-58.

Berroir S., Cattan N., Saint-Julien Th., 2005, "La mobilité des étudiants entre les universités franciliennes", Les Cahiers de l’Institut d’aménagement et d’urbanisme de la région Île-de-France, 76-85.

Berroir S., Mathian H., Saint-Julien Th., Sanders L., 2008, "La mobilité dans la construction du polycentrisme métropolitain", in Thériault M., Des Rosiers Fr., (ed.), Information géographique et dynamiques urbaines - Traité IGAT, chapitre 1, Editions Lavoisier.

Blondel V.D., Deville P., Morlot F., Smoreda Z., VanDooren P., Ziemlicki C., 2011, “Voice on the Border: Do Cellphones Redraw the Maps?”, ParisTechReview, November 15.

Bretagnolle A., Paulus F., Pumain D., 2002, "Time and space scales for measuring urban growth", Cybergeo : Revue européenne de géographie, No.219, 12 p. DOI : 10.4000/cybergeo.3790

Bretagnolle A., Pumain D., Vacchiani-Marcuzzo C., 2007, "Les formes des systèmes de villes dans le monde", in M.F Mattéi, D. Pumain (coord.), Données Urbaines, 5, Paris, Economica, Collection Villes, 301-314.

Bretagnolle A., Pumain D., Vacchiani-Marcuzzo C., 2009, “The Organization of Urban Systems”, in D. Lane, D. Pumain, S. Van der Leeuw, G. West (ed.), Complexity perspectives on innovation and social change, chapter 5, ISCOM, Springer, Methodos Series, 197-220.

Cattan N., 2010, "Le système urbain français", in : Cailly L., Vanier M. (dir), La France, une géographie urbaine, Paris, A. Colin, 365 p.

Cattan N., 2010, "Des sociétés et des territoires mobiles", Territoires 2040, No.1, 71-74.

Cattan N., 2004, "Le monde au prisme des réseaux aériens", Flux, No.58, 32-43.

Cattan N., Pumain D., Saint-Julien Th., Rozenblat C., 1994, Le système des villes européennes, Paris, Anthropos, coll. Villes, 204 p.

Comin M-N., 2009, Réseaux de villes et réseaux d'innovation en Europe : structuration du système des villes par les réseaux de recherche sur les technologies convergentes, Thèse de doctorat, Université de Paris 1, 724 p.

Rozenblat C., Cicille P. 2003, Les villes européennes, Paris, La Documentation française, 94 p.

Damette F., 1994, La France en villes, Paris, DATAR, La Documentation française, 271 p.

Degenne A., Forsé M., 2004, Les réseaux sociaux, 2e édition, Paris, Armand Colin, 295 p.

Derudder, B., Taylor P., Ni P., De Vos A., Hoyler M., Hanssens H., Bassens D., Huang J., Witlox F., Shen, W. & Yang X., 2010, “Pathways of Change: Shifting Connectivities in the World City Network, 2000-2008”, Urban Studies, vol. 47, No9, 1861-1877.

Dobruszkes F., 2008, Libéralisation et desserte des territoires : le cas du transport aérien européen, Peter Lang, coll. Action publique, 285 p.

Drevelle, M., 2012, "Structure des navettes domicile-travail et polarités secondaires autour de Montpellier", M@emonde, No.107.

Ducruet, C., 2012, "Multigraphes, multiplexes, et réseaux couplés", Groupe f.m.r.,

Gingembre J., Baude J., 2014, "Les mobilités domicile-travail dans les réseaux d’agglomérations", EchoGéo, No.27,

Gilli F., 2005, "Le Bassin parisien. Une région métropolitaine", Cybergeo : Revue européenne de géographie, No.305. DOI: 10.4000/cybergeo.3257

Grossetti M., Eckert D., Jegou L., Gingras Y., Lariviere V., Milard B., 2013, "Cities and the geographical deconcentration of scientific activity: a multi-level analysis of publications (1987-2007)", Urban Studies , vol 51, issue 10. DOI :

Halbert, L., Cicille, P., Rozenblat, C., 2012, Quelles métropoles en Europe ? Des villes en réseaux, La Documentation française, DATAR collection Travaux No.16, 108 p.

L’Hostis A., 2003, "De l'Espace contracté à l'espace chiffonné : Apports de l'animation à la cartographie en relief des distances-temps modifiées par les réseaux de transport rapides", Revue internationale de Géomatique, No.13, 69-80

L'Hostis A., Bozzani S., 2010, Contactability measures for assessing urban competitiveness, Glasgow, European Transport Conference

Massard N., Autant-Bernard C., Riou S., 2003, Production de connaissances et innovation : positionnement et enjeux pour la MIAT Bassin Parisien ; Rapport final, Saint-Etienne, 118 p.

Matthiesen C., Schwarz A., Find S., 2006, “World Cities of Scientific Knowledge: Systems, Networks and Potential Dynamics. An Analysis Based on Bibliometric Indicators”, Urban Studies, vol. 47, No.9. DOI

Nystuen J.D., Dacey M.F., 1961, "A graph theory interpretation of nodal regions, Papers and Proceedings of the Regional Science Association, vol. 7, 29-42.

Paulus F., 2007, "Trajectoires économiques des villes françaises entre 1962 et 1999", in : Pumain D., Mattéi M.-F. (coord.), Données Urbaines 5, Paris, Anthropos, 211-223.

Pumain D., Swerts E., Cottineau, C., Vacchiani-Marcuzzo C., Ignazzi A., Bretagnolle A., Delisle F., Cura R., Lizzi L. et Baffi S., 2015, "Multilevel comparison of large urban systems", Cybergeo : European Journal of Geography, No.706. DOI: 10.4000/cybergeo.26730

Pumain D., 2000, "Settlement systems in the evolution", Geografiska Annaler: Series B, Human Geography, vol. 82, No.2. 73-87.

Sanders L., 1997, "Durability of settlements systems: a long term perspective", Cybergeo: European Journal of Geography No.31, 10 p. DOI: 10.4000/cybergeo.1547

Spiekermann K., Wegener, M., 2008, "The shrinking continent: accessibility, competitiveness and cohesion". in: Faludi A. (ed): European Spatial Research and Planning. Cambridge, MA, Lincoln Institute of Land Policy, 115-140.

Taylor P. J., Walker D. R. H., Beaverstock J. V., 2002, "Firms and their Global Services Networks" in: Sassen S. (ed), Global Networks, Linked Cities, New York, Routledge, 93-115.

Taylor P. J., 2004, World city network A global urban analysis, London, Routledge, 256 p.

Vandermotten C., 2009, "Les villes qui gagnent, les villes qui perdent. Forces et faiblesses des régions métropolitaines européennes", Futuribles, No.354. DOI : 10.1051/futur/200935461

Veltz P., 1996, Mondialisation, villes et territoires, Paris, PUF, 262 p.

Haut de page


1 In the definition given by INSEE, based on the perimeters defined in 1999, there are 354 urban areas in France.

2 French urban systems are analysed within the national framework. It would certainly have been desirable to integrate international links, particularly on a European scale, where cross-border situations modify and shape the configurations of certain urban systems. But difficulties in the current availability of exchange data at these levels have prevented the development of this research outside national borders.

3 In this paper, the terms "urban area" and "city" will be used equivalently by convention.

4 This index expresses the monthly number of aircraft and TGV connecting two urban areas in spring 2010 (sources: OAG, SNCF). For TGV services, city pairs were only considered when at least part of the journey was made on a high-speed line.

5 This method was preferred for its ease of understanding, compared to statistical methods for data analysis. However, the latter have also been tested and have led to very similar results.

6 There are more sophisticated methods (multigraphs) to identify territorial systems taking into account several types of links between two summits (Ducruet, 2012). In our case, the small size of the graphs considered made the use of such methods less necessary.

7 Paris is understood as the urban area of Paris, whose perimeter goes beyond the administrative limits of the Ile-de-France region.

8 See Figure 4 for the case of the Paris urban proximity system.

9 The intensity of the links varies from 0 to 28 (7 indicators * 4 classes)

10 The relative centralisation index was developed by network sociologists. Its objective is to evaluate the forms of spatial organization according to a degree of centralization. It is calculated according to the degree of centrality of the nodes, that is, here, the number of links that link an urban area to other urban areas. The formula is written:

Image 10000201000000970000003E891FE203BCE7555B.png

where, for a given set of links (here internal preferential links to a local urban system):
- Ci is the number of links linking the urban area i to the other urban areas of the local urban system
Cmax is the largest Ci found in the urban proximity system;
n is the number of urban areas in the local urban system
C varies from 0 for a non-polarized network to 1 for a perfectly starred network.

11 Results from a Hierarchical Ascendant Classification (with Euclidean distance) based on the share of each indicator in the cumulative score of preferential flows.

12 CAF takes into account the 7 study indicators (Table 1).

Haut de page

Table des illustrations

Titre Table 1: The seven indicators of links between French cities taken into account
Fichier image/png, 37k
Titre Figure 1 Population mobility through residential migration
Fichier image/png, 218k
Titre Figure 2 Economic links through headquarters-establishments links in innovative sectors
Fichier image/png, 149k
Titre Figure 3 Leisure mobility by primary and secondary residences
Fichier image/png, 208k
Titre Figure 4 The three scales of interdependence of French urban systems9
Fichier image/png, 311k
Titre Figure 5 The variety of configurations of urban proximity systems
Fichier image/png, 16k
Titre Figure 6 Categorising local urban systems12
Fichier image/png, 177k
Titre Figure 7 Transverse links in local urban systems
Fichier image/png, 24k
Haut de page

Pour citer cet article

Référence électronique

Sandrine Berroir, Nadine Cattan, Frédéric Dobruszkes, Marianne Guérois, Fabien Paulus et Céline Vacchiani-Marcuzzo, « French urban systems: a relational approach », Cybergeo : European Journal of Geography [En ligne], Espace, Société, Territoire, document 807, mis en ligne le 12 octobre 2020, consulté le 31 octobre 2020. URL : ; DOI :

Haut de page


Sandrine Berroir

Maitre de Conférences, Université Paris-Diderot, UMR Géographie-cités, Paris, France

Nadine Cattan

Directrice de recherche, CNRS, UMR Géographie-cités, Paris, France

Frédéric Dobruszkes

FNRS Research Associate, Université Libre de Bruxelles, IGEAT, Bruxelles, Belgium

Articles du même auteur

Marianne Guérois

Maitre de Conférences, Université Paris-Diderot, UMR Géographie-cités, Paris, France

Articles du même auteur

Fabien Paulus

Maitre de Conférences, Université de Strasbourg, LIVE, Strasbourg, France

Articles du même auteur

Céline Vacchiani-Marcuzzo

Maitre de Conférences, Université de Reims Champagne-Ardenne, UMR Géographie-cités, France
EA Habiter

Articles du même auteur

Haut de page

Droits d’auteur

Licence Creative Commons
La revue Cybergeo est mise à disposition selon les termes de la Licence Creative Commons Attribution - Pas d'Utilisation Commerciale - Pas de Modification 3.0 non transposé.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search