Interaction model application for urban planning

Abstract

Urban growth is a terrific issue for the next centuries. Monitoring the urban development in order to assure sustainable cities in the future is an absolute necessity. Decision support in this domain needs some spatial information enabling to forecast the urban development trends. Satellite imagery may provide reliable information assessing the different states of urban growth, but it is necessary to anticipate where future built areas might be located. This paper presents the interest of interaction models, more precisely the potential model, to define future development zones. This model is applied to the Strasbourg area (France) using remotely sensed data. Two images are used, one to initiate the model processing and the other to validate the results. GIS capabilities are applied to refine the results, adding traffic network in the study.

Introduction

The urban population in the world has grown rapidly in the last century, increasing from 14% at the beginning of this century to 25% in 1950. Between 1950 and 1985, the number of people living in cities has more than tripled. The urban growth rate is of 0.8% a year in the developed countries, but it has reached 3.6% a year in the 1980s (UNDP, 1991). Even if urban growth patterns have not been consistent over time and across the various geographic areas, it is necessary to monitor this urban development in order to avoid possible environmental problems due to urban sprawl and traffic increase (land-use changes, increase of atmospheric pollution, etc.). One of the difficulties to cope with such a situation is to forecast locations of future development. Of course, local authorities apply urban management tools to define and restrict future land-uses. In France, these tools are the plan d’occupation du sol (POS) or the schéma directeur d’aménagement et d’urbanisme (SDAU) or more applied tools like the zone d’aménagement concertée (ZAC). These tools are associated with cities strategic planning decisions depending both on the use of all pieces of land adjacent to actual urban areas and on the restriction of some land-uses which impact may be harmful to sustainable development.

Remotely sensed data are often used to get a general view of urban areas but are seldom integrated in a landscape modelling approach. Nevertheless, satellite imagery can be a valuable help in designing decision support products allowing city development visions. The link between the forecasting issue and the satellite image is introduced here in the use of an interaction model, the potential model. Interaction models and especially potential model have often been used for demographic or social purposes, in order to study the relationships between population or services and locations, or between users and services points. Defining the interactions between land-use characteristics, Donnay, 1995, Donnay, 1994, Donnay and Lambinon (1997) and Binard et al. (1997) have promoted the application of such model on remote sensing products in order to define limits of urban areas. Another use of potential model also dealing with remote sensing has been proposed by Weber and Hirsch (1997) with the integration of potential model results and geographic information system (GIS) to assess development trends of urban settlements over years. Referring to natural or human constrains, it allows for merging different types of interactions among various landscape objects associated to natural elements (land-cover characteristics) or to political, economical decisions like protected areas for instance.

The application presented in this paper has been conducted on the Strasbourg urban region (France) on the German border, in order to assess change trends during 20 years. The method allows comparison with the actual city development outlines and the estimated trends. A time analysis has been set up to evaluate the spatial changes occurring between two dates (1986–1998). Two satellite images have been processed: the first one to get a initial state of the land-cover surface of Strasbourg and the surroundings (SPOT XS 1986), and the second one to perform time-series analysis and to assess the accuracy of the potential model application (SPOT XS 1998). Ancillary data are mixed to the results in order to strengthen our knowledge on the studied area. The potential model application is used to get a mapping of spatial change possibilities, depending on initial land-use/land-cover categories. The expected results are the identification and spatial location of Strasbourg development trends.

The paper is divided into three parts. The first presents the objectives of the study. The second describes the area, the methods and the material used. Finally, the third provides some results. They are commented and compared with classification result of the 1998 image.