Ground and image based hyperspectral sensing to characterise carbonate mineral resources of salem namakkal and ariyalur districts south India

Abstract

Mapping of the distribution of fresh and highly altered dunite and newlinethe endproduct magnesite by conventional techniques is time consuming and newlinerequires detailed sampling and laboratory analysis Similarly the presence of newlineimpurities of different forms clay iron oxide etc and great variation in the newlinegrade of limestone resources makes their mapping a difficult task Remote newlinesensing has the potential to provide detailed information on mineralogy newlinechemistry and morphology of dunite magnesite and limestone formations and newlinedeposits newlineThis thesis presents results of experiments carried out to generate newlineground and imagebased hyperspectral sensing approaches for mapping and newlinecharacterising the magnesite crystalline and ferruginouslimestone resources newlinein the Salem Namakkal and Ariyalur Districts of Tamilnadu south India The newlinemethodology adopted includes field visit for sample collection geochemistry newlinespectroradiometry hyperspectral imagebased studies and statistical analysis newlineCarbonate spectra exhibits a number of diagnostic absorption bands newlinein the 3002600nm region It is seen that the samples collected from Salem newlinedunitemagnesite region exhibit the typical reflectance peak at 550nm region newlinefor pure dunite Typical absorption troughs are seen in the NIR region for newlineiron containing dunite in the 1400nm and 1900nm regions for the hydroxyl newlineion content of altered dunites and serpentine It is clearly evident that dunite newlinesamples show almost no absorption in the 2350 nm region due to the newlinecomplete absence of CO newlinewhile pure magnesite samples show appreciable newlineabsorption due to abundance of CO newlineIn the image ASTER based study as a result of spectral newlineunmixing two fraction images for each class namely dunite and magnesite newlineare generated There is an appreciable R 09113 relationship between the newlinegrade of magnesite estimated by spectral unmixing and the grade estimated by newlinefield based sampling and geochemical analysis This image based study has newlineproven that we can not only map unaltered partially altered fully altered newlinedunite but also estimate the grade of magnes