Spectral analysis of ASTER data covering part of the Neoproterozoic Allaqi-Heiani suture, Southern Egypt

This paper examines the potential of ASTER data in mapping the geology of an arid area in Africa using spectral analysis techniques. Equipped with a relatively broad spectral range, fine spatial resolution, and a large number of bands, ASTER data are especially attractive to many geological researchers in Africa, due to their open availability and associated minimal costs for academic use. The spectral analysis of an ASTER dataset covering part of the Neoproterozoic Allaqi-Heiani suture, Southern Egypt has demonstrated it to be more effective in detecting lithological units than traditional multispectral analysis procedures, such as the maximum likelihood classifier. The effectiveness of these spectral analysis techniques used in this study lies in their ability to compare a pixel spectrum with the spectra of known pure materials, extracted from the spectral endmember selection procedures, including minimum noise factor (MNF), pixel purity index (PPI) and n-dimensional visualization. Among the three spectral analysis algorithms employed, spectral angle mapping (SAM) and spectral feature fitting (SFF), produced accurate classifications that were close to the ground reference data. Linear spectral unmixing (LSU), although failing to achieve a satisfactory overall classification accuracy, demonstrated potential in depicting intercalation of layered rocks. It is concluded that these spectral analysis techniques have proven to be useful tools in geological mapping of the arid regions in Africa, with their abilities to capitalize on the enhanced spectral information of ASTER data.