Quantification of constituents in areal and intimate binary mixtures of particulate materials

We conducted an investigation of accuracy and applicability of linear and non-linear mixture models to analyze the spectra of binary mixtures of particulates as they apply to hyperspectral remote sensing. The goal of the spectral analysis is to estimate the abundance of each constituent in the binary mixture in terms of the mass of each constituent. All of the data analyzed for this were collected under controlled laboratory conditions using particulate materials that were carefully sifted to limit the particle size distributions. Quantification of intimate mixtures may not be practical in remote sensing due to the requirement that one needs to have detailed knowledge of the materials being observed and the manner in which they are mixed. However, when the particle sizes and mass densities are similar to one another, one can get a reasonable estimate of the mass fractions of a mixture using a simple linear mixture model.