The identification of urban systems: an exploratory spatial data analysis of land use in the Lombardy Region

In this paper we present a data-driven methodology to classify the territories according to the urban structure of municipalities in absence of reliable commuting data. The methodology is grounded on the Exploratory Spatial Data Analysis (ESDA) and the objective is twofold. On the one side we aim at detecting polycentric urban systems by defining their boundaries. On the other side we seek to identify the presence of mono-centric cities in less urbanized areas. Such a classification is necessary to further extend the analysis to the role of peri-urban territories in different urban frameworks. The choice to rely on ESDA is motivated by the necessity to account for spatial association in statistical description of urbanization-related variables. The presence of functional relations among territories is likely to determine, in fact, the spatial clustering of zones with similar characteristics in terms of degree of urbanization. In absence of functional relations data, cluster detection demonstrates to be a powerful tool for the identification of urban systems. The tool is used to study the spatial distribution of soil use for urbanization purposes in the Lombardy Region, in Italy. The Lombardy region provides an interesting example of co-existence of a variety of urban typologies. First of all, the city of Milan, the largest in the region with more than 1300000 inhabitants . The area surrounding Milan is the most urbanized in the region, although very specific agricultural activities also characterize the periphery. Secondly there are other main cities such as Varese, Bergamo and Brescia, still very densely populated – less than Milan – and whose activities are highly interconnected with the economy of Milan. Finally there are other cities such as Cremona, Mantova, Lodi and Pavia located in the countryside in the southern part of the region, characterized by specialized agricultures. Accordingly, for the Lombardy region, we expect peri-urban areas – whatever its definition is – to be differently structured in different parts of the region. The classification of urban typologies thus appears as a necessary step to study further the peri-urban. The methodology is described in section 2, in which the data are also presented. Results are discussed in section 3. Follow conclusion.