Spectral unmixing methods to estimate materials on satellite surface

Abstract

The concept of using non-imaging measurements (NIM) to obtain information about satellites is discussed. We present the use of sunlight spectral measurements to estimate the composition of satellites exterior surface. We simulate the sunlight spectrum after reflection on a satellite surface by using real spectral measurements from 14 different materials commonly used on satellite surface. Several methods have been developed and tested: a linear constrained estimator and method based on the singular value decomposition (SVD) of the material database. The constrained estimator gives the best estimate regarding the presence of materials in the synthetic spectrum and regarding the relative quantity of the materials in the spectrum when both methods detect the correct materials present in the synthetic spectrum.