Comparer deux types de systèmes de villes par la modélisation multi-agents
Résumé
Simpop2 is a generic multi-agent model, conceived in the last years with the collaboration of geographers (UMR Géographie-Cités) and computer scientists (LIP6), and designed for simulating any system of cities in the world. From an evolutionary theory of urban systems, we have extracted different features and properties which characterize networks of cities in different parts of the world and over long time period. In particular, the model has to reproduce two different dynamic features of interacting cities: 1) a general growth trend and spatial expansion (produced through interurban competition for territorial and network resources, which generates emulation towards innovation); 2) a differentiation in city sizes and socio-economic functions (produced through interactions between cities).
The generic model has been instantiated through several applications. A first one is dedicated to the short-term simulation of European cities (1950-2000), with a predictive approach (Eurosim model, Sanders and al. 2007). A second one is dedicated to the long-term simulation of cities in Europe (from 1300 to 2000) and in United States (1650-2000). We show that the model can simulate properly the general dynamics of urban systems, at different scales or observation: the Simpop2 model is able to reproduce the persistency of the hierarchical configuration of the European and American systems (at the macrogeographic level and in the individual trajectories of cities) as well as the dramatic increase of contrasts in city sizes since the first industrial revolution. The results of the simulations also helps to identify a certain number of dynamic properties that are shared by both systems: 1) the dramatic effects of the innovation cycles on the urban growth; 2) the hierarchization of the city sizes that has been linked to the increase of the transportation speed since the transportation revolution; 3) the necessary integration of a new urban function in the model, which represents the emergence of worldwide cities (and that as early as in the 16th-or 17th century in Europe, 18th-19th in the United States). But beyond these similar evolutions, the model also reveals fundamental differences in space filling between the two systems.
The generic model has been instantiated through several applications. A first one is dedicated to the short-term simulation of European cities (1950-2000), with a predictive approach (Eurosim model, Sanders and al. 2007). A second one is dedicated to the long-term simulation of cities in Europe (from 1300 to 2000) and in United States (1650-2000). We show that the model can simulate properly the general dynamics of urban systems, at different scales or observation: the Simpop2 model is able to reproduce the persistency of the hierarchical configuration of the European and American systems (at the macrogeographic level and in the individual trajectories of cities) as well as the dramatic increase of contrasts in city sizes since the first industrial revolution. The results of the simulations also helps to identify a certain number of dynamic properties that are shared by both systems: 1) the dramatic effects of the innovation cycles on the urban growth; 2) the hierarchization of the city sizes that has been linked to the increase of the transportation speed since the transportation revolution; 3) the necessary integration of a new urban function in the model, which represents the emergence of worldwide cities (and that as early as in the 16th-or 17th century in Europe, 18th-19th in the United States). But beyond these similar evolutions, the model also reveals fundamental differences in space filling between the two systems.
Simpop2 est un modèle multi-agent générique élaboré dans les dernières années par des géographes (UMR Géographie-cités) et par des informatiens (LIP6). Il est conçu pour simuler l'évolution de différents systèmes de villes dans la longue durée. Les villes inter-agissent sous des hypothèses de concurrence, formalisées par le fonctionnement spatialisé d'un marché d'échanges. Elles sont rendues hétérogènes par la diversité de leurs fonctions (centrales, territoriales et spécialisées) et par leur plus ou moins bonne insertion dans les réseaux de l'échange. Différents modèles spécifiques ont été construits à partir du modèle générique, par exemple le modèle Eurosim, simulant les villes européennes de plus de 200 000 habitants de 1950 à 2050 à l'aide de différents scénarios prospectifs (Sanders et al., 2005) . Nous présentons ici les résultats d'une comparaison entre les modalités d'évolution des systèmes de villes en Europe (1300-2000) et aux Etats-Unis (1650-2000). Les simulations permettent de mettre en évidence de fortes similitudes (le rôle des cycles d'innovation dans la croissance urbaine, l'impact de l'accroissement de la portée de échanges sur la hiérarchie urbaine, la nécessité du recours à une nouvelle fonction caractérisant les villes de rayonnement international, dès la mise en place du système) mais aussi des différences fondamentales dans les formes d'occupation du sol.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|