No abstract

No abstract

Monitoring urban expansion is critical to understanding the effects of urbanization on ecological processes. However, it presents important challenges due to the heterogeneity of the urban areas and the rapid changes in land cover and land use that occur over short time periods. Conventional surveying and mapping methods are not adequate to produce accurate and timely land-use and cover change information on urbanization. To remedy to these problems, object-oriented methods using remote sensing data with very high resolution have been recently developed. The main objective of this paper is to assess the ability of an object-oriented approach to classify urban areas and their spatial expansion with a moderate/high resolution satellite image time series. This paper focuses on the monitoring of urban expansion of a medium-size city (Rennes metropolitan area, North-western France, 384 000 inhabitants) during the 21 year period from 1984 until 2005 with a multiscale object-oriented method. The image object-oriented classification approach we used consists of a two step method : A segmentation technique has been performed prior to classification. Preliminary results show an important urban expansion (+ 62%) during the 21-year investigated period, with a large increase since the early nineties. The object-oriented method used in this research has proven to be reliable to discriminate pixels that are truly urban from pixels that have similar spectral properties, such as clear cuts or bare soils, from moderate/high resolution images, on the contrary of per-pixel classifications.