Data from spectroscopic measurements of two ternary particulate mixtures with two common minerals.
This data-set contains of two ternary particulate mixtures. The first mixture is composed of the minerals bytownite, diopside and olivine (San-Carlos), while the second mixture is composed of the minerals augite, diopside and olivine (San-Carlos). The grain sizes of the augite particles lie in the range 63-108 µm, while the other mineral particles have a grain size in the 28-63 µm range. The (triangular) abundance simplices were then sampled 66 times using a regular grid, intimate mixtures were then manually prepared corresponding to each abundance vector, where each value on the vector corresponds to the fraction of a specific material.
Each sample was then indvidually placed in a sample holder and imaged using a Micro-Hyperspec ® SWIR M-Series imaging sensor in the in the Headwall Hyperspec Starter Kit [Micro-Hyperspec ® is a registered trademark of Headwall Photonics]. Each sample is imaged in the wavelength range 0.9-2.6µm with a spectral resolution of 9.6 nm. For each image the pixels corresponding to the sample are extracted and placed in the dataset (There are about 30 samples corresponding to each abundance combination).
The data is available for download here .
The .rar file linked here contains: 1) endMembers_mixY.tif : - plots of the endmember spectra for the Yth mixture. 2) mixtureY.mat : - data for the first ternary mixture, in MATLAB format, this file contains the variables: a) mixYSamp : - N X 165 matrix where each row corresponds to a spectra belonging to the Yth mixture b)mixYSamp_Abun : - N X 3 matrix where each row corresponds to the abundance vector where the weight in the first column corresponds to the abundance to the mineral plotted in blue in corresponding plot, the second to the mineral plotted in green and the third column to mineral plotted in red in item-1. c) mixYLabels : - due to difference in illumination or local geometry due to roughness the spectra measured for (pixels) for the same abundance vector have some minor variations. The dataset contains spectra for the same abundance combination but with differences in illumination or local geometry. All the spectra with the same abundance combinations are given the same label. These labels are stored in in this variables. 3) prinComp_bothMixtures.tif : - This image shows the scatter plot of the scores in the first three principal components of complete data-set made up of both mixture-1 and mixture-2. 4) wavelength_SWIR : - the wavelength range for the spectra.