ALTERATION MINERAL MAPPING IN THE CENTRAL ANDES USING HYPERION, ALI AND ASTER

Airborne hyperspectral (high spectral resolution) sensors such as AVIRIS have provided a wealth of information on the composition of most terrestrial materials. Because these instruments are flown on aircraft, they provide limited temporal resolution for some applications and sparse geographic coverage for others.

This study evaluates hyperspectral data from sensors on earth orbiting satellites. We utilize and compare datasets acquired over the same geographic area by two-multispectral and one-hyperspectral sensors aboard the Terra and EO-1 satellites, respectively. All three instruments have similar spectral ranges in the VNIR (400-1000nm)-SWIR (900-2500nm), but vary in spatial and spectral resolution. Hyperion has 70 channels in the VNIR plus 172 channels in the SWIR and 21 channels in the region of overlap. ALI and ASTER each have ten channels variously spaced throughout the VNIR-SWIR. ASTER also has five additional channels in the TIR (8100-11000nm).

Preliminary mapping results were based on scenes covering a roughly N-S oriented area of the Central Andes between Volcan Socompa and Llullaillaco, Chile. This area is a hyper-arid plateau with numerous salt basins, alluvial fans, small volcanic constructs, ignimbrite sheets, composite volcanoes, breached cones and a large debris avalanche deposit from Socompa. Areas of exposed altered rocks on several of the composite and breached cones were mapped using a combination of spectral shape-fitting and partial spectral unmixing algorithms. Analysis of Hyperion pixels confirms the presence of kaolinite-alunite, montmorillonite and hematite-goethite mineral assemblages on several of the volcanic edifices. Extensive areas containing hydrous silica minerals were mapped on the northeastern flanks of Socompa. Partial unmixing suggests the presence of gypsum in salt flat areas, gypsum and jarosite in altered bedrock areas, and calcite in parts of the Socompa debris avalanche deposit known to have incorporated sedimentary rocks from the basement underlying Socompa.

These results are compatible with earlier studies using aircraft hyperspectral data. Continuing work will apply other spectral processing methods to broader-band ALI and ASTER data to determine their utility for mapping similar suites of minerals with overlapping absorption features.