ERS SAR interferometry for land cover mapping in a savanna area in Africa
Abstract:This paper is a contribution to studies on European Remote Sensing (ERS) Synthetic Aperture Radar (SAR) interferometric applications, especially in tropical countries. It investigates the information content of ERS SAR tandem images for land cover mapping in a savanna environment in Africa, by interpreting and analysing an Interferometric Landuse (ILU) image. Using this image produced by an RGB combination of coherence, the average backscatter intensity and backscatter intensity change images, woodland (i.e. dense canopy vegetation areas), clear cut areas and grass/shrub land (i.e. sparsely vegetated or low canopy vegetation areas) and bare/degraded (non-vegetated or exposed areas) were identified. A Landsat 7 ETM image was classified and compared with the ILU. Ground control points were collected for processing and verification of the remotely sensed images used. This study agrees with the findings of other authors, that SAR interferometry (InSAR) products, especially coherence images, can optimize the use of ERS SAR tandem images for land use and land cover mapping. Thus, confirming that vegetated areas have low coherence, while non-vegetated areas (exposed soils and grass/shrub lands) have high coherence, even in tropical savanna areas in Africa. In addition to this, an accuracy assessment of the two images when compared with ground data shows that the ILU image provided a more acceptable land cover map of the study area with an overall accuracy of 89% and a kappa of 75%. These results show that the ILU image could be used to complement optical data in these areas, by exploiting the all weather capability of SAR.
Document Type: Research Article
Affiliations: Department of Geography Federal University of Technology Minna, Niger State Nigeria, Email: Aimiosino@yahoo.com
Publication date: September 1, 2003