Using geographical information system and satellite imagery within a numerical simulation of regional urban growth

Abstract

This paper represents an attempt to combine a geographical information system (GIS) and satellite imagery within a numerical model for the simulation of regional urban growth. The area under study is located in the southeast of France, between the cities of La Ciotat and Toulon. It covers 1600 km². The spatial resolution of the GIS as well as of the remote sensing data is better than one hectare (80 × 80 m²). Two Landsat MSS satellite images were analysed by a supervised maximum likelihood classifier to provide two maps of land use and urban domain for 1976 and 1981. A numerical model for the growth of the density of the population is applied to the urban domain. It takes into account some constraints that are due to the environment as well as the changes of population per city and per year. Seven environmental parameters are defined by pixel and originate from the content of the database, namely, land use, topography, road network, forest properties and forest fire characteristics. The global change in population is given by observations for the first years and by scenarios for the predicted years. The numerical simulation gives maps of the urban domain per year and per scenario. The initial state (year 1976) and the simulated states Tor the years 2000 and 2025 are compared and the differences are discussed briefly.