South-Central Section - 39th Annual Meeting (April 1–2, 2005)

Paper No. 7
Presentation Time: 8:30 AM-12:00 PM

HYEPR-SPECTRAL ANALYSIS OF THE ADVANCED SPACEBORNE THERMAL EMISSION AND REFLECTION RADIOMETER (ASTER) DATA: EXAMPLES FROM THE NEOPEROTEROZOIC UM NAR BANDED IRON FORMATION, EGYPT


EL-NAGDY, Sherif and ABDELSALAM, Mohamed G., Geoscience Department, Univ of Texas at Dallas, Richardson, TX 75080-830688, sme023000@utdallas.edu

We present remote sensing results from hyper-spectral digital image processing of multi-spectral Advanced Thermal Emission and Reflection Radiometer (ASTER) data from the Neoproterozoic Um Nar Banded Iron Formation (BIF) pertaining to lithological and structural mapping. The Um Nar BIF is located in the Central Eastern Desert of Egypt and is made up of dominantly metasedimentary rocks, serpentinites, and granites deformed by N-verging thrusts and folds and NNW-trending dextral strike-slip faults. We have used the nine visible and near infrared (VNIR, 15 meters spatial resolution) and shortwave infrared (SWIR, 30 meters spatial resolution) bands of ASTER data to perform: (1) Minimum Noise fraction (MNF) transform; (2) Pixel Purity Index (PPI) mapping; (3) N-dimensional visualization and end-members extraction; (4) Whole pixel classification including spectral angle mapper (SAM) and Spectral Feature Fitting (SFF); and (5) Sub-pixel analysis including Linear Un-mixing and Matched Filtering. In addition, selected red-green-blue (RGB) ASTER color combination images have been draped over Digital Elevation Models (DEMs) extracted from the near infra-red bands (bands 3N and 3B) of ASTER data. Our analysis indicates that: (1) Regions covered by serpentinites and granites have higher PPI compared to metasedimentary rocks because granites and serpentinites are relatively homogenous lithologies compared to metasedimentary rocks. (2) Using the N-Dimensional Visualizer, ten end-members were extracted. SAM gave better results (as verified by in situ information) than SSF since the extracted end-members have weakly-defined spectral features. Matched Filtering gave better results and it is preferred over Linear-Un-mixing which might be because of the difficulty to identify all the end-members in the study area. (4) Drapping selected RGB color combination images over ASTER DEMs allowed us to examine the three-dimensional distribution of the lithological units as well as examining the morphological expression of the geological structures. This allowed us to refine previous geological interpretations including identifying new structures such as NW-trending sinistral faults and a serpentinites thrust klippe associated with the N-verging thrusts.