Geospatial assessment of ultramafic rocks and ore minerals of Salem, India

Abstract

The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) image interpretation methods such as colour composite images (false colour composite, true colour composite) were adapted to capture the image for interpreting the visible and shortwave infrared raw bands and thus generating the mapping for the ultramafic terrain. ASTER colour composite image generated from shortwave infrared (SWIR) bands 8 and 4 and NIR band 3 shows contrast signature for the presence of rock types in the ultramafic terrain. The low and high silica percentages to be interpreted through the absorption features of the spectral range from 8.1 to 12 μm. The combinations used to map silica index are of bands 10, 11 and 7 as RGB. PCA was applied to SWIR and visible and near-infrared spectral bands and from PCA output PC 6, 2 and 1 used in the generation of RGB colour composite. Hence, the results of PCA processed image highlights the magnesite mining area by dark red colour. The minimum noise fraction transform (B1, B2, B4) in RGB applied to ASTER image to focus on the overlapped rock types inclusive of magnesite and the same achieved as sharp image output. ASTER band ratios (R3/1, G4/5 and B6/8) in RGB calculated in which the numerator represents shoulders of absorption, and the denominator represents nearest absorption band. Thus, this combination indicates clearly the magnesite mining area in blue colour. The relative band depth in thermal infrared (TIR) bands (B13 and B14) correlated to find the carbonate index is high, medium and low through the pixel ratio output of the ultramafic terrain. Preprocessing of ASTER data involved atmospheric correction using FLAASH algorithm. The minimum noise fraction transform reduces the spectral dimensionality of data in a linear combination of bands. Spectral angle mapper and support vector machines are supervised classification technique adopted that are key learning models with the aid of analyse and categorization of data. The classification defends user, producer, overall accuracy and kappa coefficient. The magnesite mining area and adjust rock samples were field observed, and the mapping of the ultramafic terrain supports the output achieved through chosen ASTER band colour composite, band ratios, relative band depth, silica index, PCA, minimum noise fraction and pixel classification. Thus, a more informative lithological map for the ultramafic terrain generated to discriminate magnesite-mining region.