int. j. remote sensing, 2000 , vol. 21, no. 10, 2101–2111 The study of ERS-1 SAR and Landsat TM synergism for land use classi cation T. M. KUPLICH Department of Geography, University of Southampton, High eld, Southampton, SO17 1BJ, England, UK; e-mail: tmk@soton.ac.uk C. C. FREITAS and J. V. SOARES Instituto Nacional de Pesquisas Espaciais (INPE), Caixa Postal 515, 12227-010, Sa˜o Jose´ dos Campos, SP, Brazil (Received 9 October 1997; in nal form 17 May 1999 ) Abstract. This work presents a comparative analysis of ERS-1 Synthetic Aperture Radar (SAR) and Landsat-5 Thematic Mapper (TM) images used for land use classi cation. The study area of 361 km2 is located in the City of Campinas, Sa˜o Paulo State, Brazil, and contains several classes of land use, including urban, agricultural and forests. The TM and SAR images were registered and transformed using the principal components transformation. SAR images were also ltered using an average lter. The principal components derived from SAR ltered, SAR, TM and coregistered TM/SAR and TM/SAR ltered images were classi ed using the maximum likelihood approach. Tests of ‘goodness of t’ were also made to assess the statistical properties of the images. The results, con rmed by Kappa statistics, show a signi cant improvement when classifying the principal components of ltered SAR and TM images for urban, pasture and forest classes. 1. Introduction Remote sensing images have become an important tool for land use classi cation and mapping. Examples of successful utilization of optical images for land use and land cover mapping have been reported in several works, e.g. Khorram et al. (1991), Treitz et al. (1992 ), Defries and Townshend (1994 ) and Homer et al. (1997). One of the limitations for the use of optical data alone for land use mapping is the separation between urban and non-urban land covers, as reported by Henderson et al. (1998). The potential of Synthetic Aperture Radar (SAR) images for discrimination of land covers has also been assessed (Bryan 1983, Wu and Sader 1987, Dobson et al 1992, 1995, Nieuwenhuis and Kramer 1995, Henderson et al. 1998 ). The availability of data from orbital SAR systems —Japanese Earth Resources Satellite (JERS-1), Canadian RADARSAT and European Remote Sensing Satellites ( ERS) 1 and 2—allowed the observation of the land cover regardless of the sometimes limiting weather conditions. The study of the synergism between optical and radar images also became more feasible (Pohl and Van Genderen 1998 ). Multisource data had already been proved to oŒer better potential for the digital classi cation of land cover (Wu 1984, Jiyuan et al. 1986, Nezry et al. 1993, Hoekman and Quin˜ones 1998, International Journal of Remote Sensing ISSN 0143-1161 print/ISSN 1366-5901 online © 2000 Taylor & Francis Ltd http://www.tandf.co.uk/journals