|
Africa |
1 Department of Geography, University of Reading, Whiteknights, Reading RG6 2AB, UK
2 Department of Geography, King’s College London, Strand, London WC2R 2LS, UK
3 School of Geography, University of Oxford, Mansfield Road, Oxford OX1 3TB, UK
Palaeosurfaces are often characterized by distinctive surficial material compositions; for example, iron oxide in lateritic surfaces in India or calcium carbonate in calcreted surfaces in Africa. The chemical constituents of interest often play a significant part in the induration, and hence the preservation, of palaeosurfaces. Furthermore, these components often have distinct spectral reflectance characteristics, enabling them to be detected from multispectral remote sensing systems. However, traditional image processing techniques, such as classification, are inappropriate for mapping such phenomena, which usually exhibit a gradually-varying distribution. In studies of palaeosurface development, the requirement is for data that provide quantitative estimates of the surficial material composition (i.e. the proportions of a given component, such as iron oxides, within an image pixel). Algorithms based on compositional analytical techniques can be applied to unmix the pixel spectral response into estimates of the proportions of the image components (such as vegetation, iron oxides, etc.) within each pixel of an image. The resulting fraction maps provide a powerful technique for mapping palaeosurfaces and analysing their development. This paper demonstrates how these techniques have been applied to studies of Quaternary piedmont surfaces in the Tunisian Southern Atlas.
