SPECTRAL ANALYSIS OF ASTER, HYPERION, AND QUICKBIRD DATA FOR GEOMORPHOLOGICAL AND GEOLOGICAL RESEARCH IN EGYPT (DAKHLA OASIS, WESTERN DESERT)

摘要

This paper presents an evaluation, to which degree geological and geomorphological information can be obtained from modern remote sensing systems like the multispectral ASTER or the hyperspectral Hyperion sensor for a hyperarid desert region like the Dakhla Oasis (Western Desert, Egypt). To account for the enhanced information content these sensors provide, hyperspectral analysis methods, incorporating for example Minimum Noise Fraction-Transformation (MNF) for data quality assessment and noise reduction as well as Spectral Angle Mapper (SAM) for classification, were applied. As compensation for ground truth data high spatial resolution Quickbird data was obtained, too. The classification results show that detailed information can be extracted from the data, which were compared with the Geological Map of Egypt 1:500.000. Regarding the data quality, the analysis revealed that the Hyperion scene was strongly affected by system induced radiometric interferences. As a result, a considerable amount of bands had to be discarded to allow satisfying results. Therefore, although similar surface types were discriminated with the ASTER and the Hyperion results, the ASTER data allowed a more detailed classification of the surface composition of the study area. Additionally, ASTER elevation data was used to relate the classification results with the relief of the study area.