Hyperspectral remote sensing for minesite characterisation

  • Kirrilly Sue Pfitzner

    Student thesis: Doctor of Philosophy (PhD) - CDU

    Abstract

    One of the major issues facing the mining industry is to develop methods to assess the success of minesite rehabilitation. Hyperspectral data, namely from Modular Airborne Imaging Spectrometer (MATS), Compact Airborne Spectrographic Imager (CAST) and Hyperspectral Mapper (HyMap) were assessed for their usefulness for minesite characterisation. Two minesites in Australia were used to test such hyperspectral data. These mines were similar in their small size and environmental issues, particularly in their production of acid drainage.

    Hyperspectral data were used in conjunction with hyperspectral mapping techniques to analyse potential of such data for minesite characterisation. Endmember spectra were derived from data reduction techniques, selected United States Geological Survey (USGS) spectra, Analytical Spectral Device (ASD) field-based spectra and the hyperspectral data itself. Endmembers included iron hydroxide and sulphate mineral assemblages, as well as floe in water (of iron oxides and hydroxides). Spectra were then subjected to a variety of mapping methods. Band ratios were used to map vegetation cover also. The spectral match for each endmember was assessed (using reflectance spectra, Continuum Removed profiles, image and field spatial associations, data statistics and Spectral Analyst). Separable classes of land cover were exported to a GIS environment to produce image maps.

    A number of contributions have been made. This research has shown that CAST and llyMap data have high potential for minesite characterisation. In particular, minerals indicative of acid drainage processes, including jarosite, goethite, hematite, gypsum and floe were identified and mapped both on and off site. Vegetation cover, an important component of rehabilitated surfaces was identified with all three sensors. Field-based spectra provided additional information to that extracted from the hyperspectral data alone or with public domain spectral libraries. The results would be difficult to obtain by methods other than hyperspectral analysis and thus demonstrate the suitability of hyperspectral data for minesite characterisation.


    Note: Permission to digitise thesis not granted.
    Date of AwardMay 2004
    Original languageEnglish
    SupervisorWaqar Ahmad (Supervisor)

    Cite this

    Hyperspectral remote sensing for minesite characterisation
    Pfitzner, K. S. (Author). May 2004

    Student thesis: Doctor of Philosophy (PhD) - CDU