Defining Polycentric Urban Areas Through Commuting Cohesion in France - Methods for Multilevel Analysis and Visualisation of Geographical Networks - page 184

Gordon & Richardson , 1996 ; Sultana , 2000 ), Canada ( Bourne , 1989 )orthecom-
parative study on Stuttgart and Turin ( Binder, Haag, & Rabino , 2003 ). According
to these studies, the polycentric structure of major urban areas is now a self-evident
truth. A polycentric structure can be defined by the existence of one or several sub-
centers in addition to the city center. In our study, the commuters' flows illustrate the
existence of these sub-centers, which are seen as a point of attraction for commuters.
The aim of our work is to characterize these polycentric structures and to analyze
their properties. Therefore, our approach, which is based on the visualization of the
network of commuters using graphs and applying a measurement index, is quite
different from the approaches used by existing studies.
The analysis of daily commuting patterns, represented as weighted, directed
graphs, leads to new measurements and perspectives, potentially assisting in the
process of urban planning. The flows of commuters are represented by the edges
between places of residence and those of employment, seen as nodes of the graph.
Three types of data are added as weights on the edges: the number of commuters
that pass between the municipalities; the portion, in percent, of these commuters
who are in the working population in their municipality of residence; and the
same percentage of these commuters who are in the working population in the
municipality of their employment.
In this study, we investigate the home to work commuting patterns among 36,000
French municipalities drawn from the census data for 1975, 1982, 1990 and 1999.
For these 4 years, two scales of investigation are developed.
We need to highlight some inconsistencies in this data. First, we do not distin-
guish between effective and official residences, leading to incongruence between the
graphs built on this data and the actual situation. A data cleaning performed at the
national level and associated with a coloration of the graph based on its structural
properties leads to the delineation of cities through the commuters' flows and is
applied in the first stage of this paper.
Second, we show the impact that regional transportation had on the spread of the
commuters' network around the French major cities. Comparing the commuters'
graphs from four censuses and their cohesion, defined using the average strength
index, we highlight the reticular extension of the cities' delineation, their spatial
discontinuities and their overlap. At the same time, we show the development
of urban sprawl. The number of municipalities connected, directly or indirectly,
to a central municipality grows from 1975 to 1999. These municipalities are
growing progressively more distant from the city center. However, each municipality
is not necessarily either strongly or directly linked to primary city center. We
see urban polycentric structures develop through subnetworks that function like
subsystems. In this second section, we will apply different indexes to characterize
these subcenters. We also explain a method for building a graph that combines
the data from the four censuses with the aim of creating an evolving study of the
commuters' network.
In the last stage of this paper, we investigate a new approach that considers
the commuters' graphs to be scale-free networks. We explain this position by the
structure of the commuters' graph: a few important municipalities import a large