|
|
|
Abstract
Mineral Abundance Mapping using EO-1 Hyperion Data – A Processing Methodology for Detecting Iron and Manganese Mines in Pre-Cambrian Rocks
Iswar Chandra Das
Scientist SE
IIRS,
Email: icdas@iirs.gov.in
The mineral deposits of economic value are neither distributed uniformly all over the surface of the globe nor were they formed all through the geological history of the earth. On the contrary, they are concentrated only in certain parts of some of the continents and were produced occasionally during a few distinct and well-defined periods of mineralization. Mineral exploration is becoming increasingly difficult, especially in obtaining ground access to sensitive or remote areas (Mukherjee, 1997).
Hyperspectral remote sensing has the potential to provide the detailed physico-chemistry (mineralogy, chemistry, morphology) of the earth’s surface. This information is useful for mapping potential host rocks, alteration assemblages and regolith characteristics. In contrast to the older generation of low spectral resolution systems, such as the Landsat Thematic Mapper with only six “reflected” bands, the new generation of hyperspectral systems enable the identification and mapping of detailed surface mineralogy using “laboratory-grade” spectroscopic principles (Clark.et.al., 1990).
This study is an attempt to use the most advanced hyperspectral sensor “Hyperion” for mineral abundance and lithological mapping in parts of Keonjhar, Orissa. The Pre-Cambrian rocks of the Singhbhum Iron deposits in Orissa, India are being chosen for the present study. There are 220 unique spectral channels collected with a complete spectrum covering from 357-2576nm. The level 1 radiometric product has a total of 242 bands but only 198 bands are calibrated. Because of an overlap between the VNIR and SWIR focal planes, only 196 unique channels can be considered for further processing. The first step involved is the atmospheric correction of the Hyperion scene. Atmospheric correction is an important procedure for spectral analysis based mapping methods. In this study, FLAASH software has been explored. FLAASH is a calibration program used to convert radiance into corresponding reflectance. FLAASH is a new radiative transfer model for atmospheric calibration and was developed by Spectral Sciences, Inc, under the sponsorship of the US Air Force Research Laboratory. It incorporates the new development of MODTRAN-based radiation transfer model (MODTRAN4) into the code for modeling radiation transfer properties and includes a correction for pixel mixing due to scattering of reflected radiance from surroundings into the pixel. FLAASH also includes a correction for the “adjacency effect” (pixel mixing due to scattering of surface reflected radiance), provides an option to compute a scene-average visibility (i.e., aerosol/haze amount), and utilizes the most advanced techniques for handling particularly stressing atmospheric conditions (such as the presence of clouds). Other features include a cirrus and opaque cloud classification map and adjustable spectral “polishing” for artifact suppression (http://www.rsinc.com/envi/flaash.asp)
Although the current atmospheric correction serves only a kind of approximation, and the software used in this study is a less rigorous type, it is found that the spectra after atmospheric correction looks much like the reflectance spectra. Most absorption features can be observed clearly. Due to the low reflectance level in some bands, the original 196 bands are reduced to 167 bands after the atmospheric correction. The hyperspectral imagery is capable of providing a continuous spectrum ranging from 0.4 to 2.5 microns for a given pixel, it also generates a vast amount of data required for processing and analysis. Due to the nature of hyperspectral imagery (i.e. narrow wavebands), much of the data in the 0.4-2.5 micron spectrum is redundant/repetition. A minimum noise fraction (MNF) transformation is used to reduce the dimensionality of the hyperspectral data by segregating the noise in the data. The MNF transform is a linear transformation which is essentially two cascaded Principal Components Analysis (PCA) transformations. The first transformation decorrelates and rescales the noise in the data. This results in transformed data in which the noise has unit variance and no band to band correlations. The second transformation is a standard PCA of the noise-whitened data. The Pixel Purity Index (PPI) is a processing technique designed to determine which pixels are the most spectrally unique or pure. Due to the large amount of data, PPI is usually performed on MNF data which has been reduced to coherent images. The most spectrally pure pixels occur when there is mixing of endmembers. The PPI is computed by continually projecting n-dimensional scatterplots onto a random vector. The extreme pixels for each projection are recorded and the total number of hits is stored into an image. These pixels are excellent candidates for selecting endmembers which can be used in subsequent processing. The first step to determining the abundances of materials is to select endmembers, which is the most difficult step in the unmixing process. N-Dimensional visualization techniques can be used to select endmembers within a scene. Extreme pixels which ultimately correspond to endmembers can be determined by rotating the scatter plot in n-dimensions. These endmembers are later used in the preparation of mineral abundance mapping. In the present study, the image is first classified using Linear Spectral Unmixing (LSU) technique to find the abundance of pure pixels of iron and manganese mines and then the final image is density sliced to separate various classes. The density slice is basically required to categorize the image, because the abundance image generated using LSU contains all the values between 0 (minimum) to 1 (maximum) with fractions in it. In the final density sliced image not only mines are clearly identified but also the various stages of mining activity such as old excavated zones, freshly excavated areas are also clearly distinguished.
References:
1. Clark, R.N., King, T., Klejwa, M. and Swayze, G.A. 1990. High spectral resolution reflectance spectroscopy of minerals. Journal of Geophysical research 95 (B8): 653-680
2. Das, I.C. & Nizamiddin, M. 2002. Spectral signatures and spectral mixture modelling as a tool for targeting laterite and bauxite ore deposits, Koraput, Orissa”. Presented in Map Asia-2002, Bangkok
3. Das, I.C. 1999. Spectral characterisation for lithological discrimination in early pre-cambrian rocks, Balaghat, Central India”, in ISRS symposium, Bhubaneswar
4. Mukherjee, P.K., Text Book of Geology. The World Press Private Limited edition III
5. Baliarsingh, R.P. & Das I.C. 2006. Remote Sensing Application for targeting chromite ore deposits, Sukinda Valley, Orissa, India, National seminar on Geoscience education and Mineral Deveopment, Bhubaneswar, Orissa.
6. Sarangi, S.K., & Singh, P. 1995. Manganese and Iron formation relationship in the sedimentary milieu- A case study from Iron ore Basin North Orissa, Vistas in Geological research, Utkal University special publication in geology, Bhubaneswar.
7. Van der meer, F. 1997. Mineral Mapping and Landsat Thematic Mapper Image Classification Using Spectral Unmixing, Geocarto International, 12(3), 1-4.
URL:
http://www.rsinc.com/envi/flaash.asp
http://www.gisdevelopment.net/aars/acrs/2004/b_hyper/acrs2004_b2010.asp
http://www.cstars.ucdavis.edu/classes/mexusenvi/tut11.htm
|
|
|