Search

Les auteurs mettent en œuvre trois méthodes de modélisation prospective appliquées à des données géoréférencées haute résolution portant sur l’occupation du sol en milieu montagnard méditerranéen : approche SIG, modèle linéaire généralisé et réseaux neuronaux. Une validation des modèles est entreprise par la prédiction de l’occupation du sol à la dernière date connue. Les résultats obtenus sont, dans le contexte de la dynamique spatio-temporelle de systèmes ouverts, encourageants et comparables. Les scores de prédiction correcte se situent autour de 73 %. L’analyse des résultats porte notamment sur la localisation géographique, les types d’occupation du sol concernés et les écarts à la réalité des résidus. Un croisement des trois modèles souligne le degré élevé de convergence et une relative similitude des résultats issus des deux approches statistiques comparées au modèle SIG supervisé. Des travaux en cours concernent la mise en œuvre des modèles sur d’autres sites et le repérage des points forts respectifs afin de développer un modèle intégré. The authors apply three methods of prospective modelling to high resolution georeferenced land cover data in a Mediterranean mountain area: GIS approach, generalized linear model and neural networks. Land cover prediction to the latest known date is used to validate the models. In the frame of spatial-temporal dynamics in open systems, results are encouraging and comparable. Correct prediction scores average around 73 %. The results analysis focuses on geographic location, land cover categories and parametric distance to reality of the residuals. Comparing the three models shows the high degree of convergence and a relative similitude of the results obtained by the two statistical approaches compared to the GIS supervised model. Further steps are the application of the models to other test areas and the identification of respective advantages to develop an integrated model.

This paper focuses on the concept of a sustainable city and its theoretical implications for the urban system. Urban sustainability is based on positive interactions among three different urban sub-systems : social, economic and physical, where social well-being coexists with economic development and environmental quality. This utopian scenario doesn’t appear. Affluent economy is often associated with poverty and criminality, labour variety and urban efficiency coexist with pollution and congestion. The research subject is the analysis of local risk and opportunity conditions, based on the application of a special definition of risk elaborated and made operative with the production of a set of maps representing the multidimensional facets of spatial organisation in urban sustainability. The interactions among the economic/social and environmental systems are complex and unpredictable and present the opportunity for a new methodology of scientific investigation : the connectionistic approach, processed by Self-Reflexive Neural Networks (SRNN). These Networks are a useful instrument of investigation and analogic questioning of the Data Base. Once the SRNN has learned the structure of the weights from the DB, by querying the network with the maximization or minimization of specific groups of attributes, it is possible to read the related properties and to rank the areas. The survey scale assumed by the research is purposefully aimed at the micro-scale and concerns the Municipality of Milan which is spatially divided into 144 zones.