Unsupervised classification of evolving metropolitan street patterns

• Type: Article in International Conference Proceedings Book
• Date: 2013

Authors

Serra, M (Author) Other; Gil, J (Author) Other; Paulo Pinho (Author) FEUP

Conference proceedings International

Title: 2013 International Space Syntax Symposium

9th International Space Syntax Symposium, SSS 2013

31 October 2013 through 3 November 2013

Indexing

Scopus - 2 Citations

Other information

Authenticus ID: P-00Q-FXA

Abstract (EN): The definition of urban typomorphologies is fundamental both for the description as well as for the prescription of urban form. However, urban typomorphologies are traditionally defined through time-consuming analytical procedures, which are based mainly on the personal knowledge and ability of the analyst and on ad-hoc generalizations. Such generalizations are also usually context-dependent and apparently not fit to deal with contemporary metropolitan and suburban forms, whose emergent and very different morphologies have until now eluded stringent classifications. In this paper we explore the possibility of defining urban typomorphologies through the use of non-supervised classification techniques, focusing on urban street patterns. We illustrate the method through the analysis of the expansions of the street network of Oporto's metropolitan area, over the last sixty years. We have modelled such network using space syntax's axial mapping technique on four different historical moments (2011, 1998,1977, and 1951). By comparing these four axial models and by cross-checking them with old cartographies and current orthophotomaps, we were able to isolate and to identify each and every urban development intervention entailing the creation of new streets (as represented by their axial lines), occurring within the metropolitan network along the study's time-span. Using this dataset of street-network interventions (amounting to 4208 objects), we quantified a certain number of geometrical and topological attributes describing each intervention's street-layout morphology. This database was then explored by non-hierarchical classification techniques and a number of significant clusters were extracted. We conclude by showing that such clusters correspond to clearly different morphotypes of street-layouts, with different distributions over time. Based on these morphotypes we propose a new taxonomy of metropolitan street network expansions. We suggest that their identification can provide a solid basis to support the formulation and implementation of planning and urban design policies.

Language: English

Type (Professor's evaluation): Scientific