Classification results of spectra obtained by video spectral system on board the ISS during the Uragan space experiment

Abstract

The paper presents the results of test activations of a new modification of the spectrometric scientific equipment Videospectral System (SE VSS) on the International Space Station (ISS) as part of the Uragan space experiment. A method for processing spectra of high spectral resolution is proposed, based on dimensionality reduction and data clustering, which makes it possible to classify the spectra of similar objects. Using the principal component method, a three-dimensional coordinate system is created in which the spectral sample under study is displayed. The data in the resulting axes are classified without training using the hierarchical clustering method. Homogeneous predominantly sandy territories of various deserts (Namib, Libyan, Arabian) were analyzed. As a result of two-stage clustering, clear clusters and corresponding spectra of desert sands from different regions were obtained and then entered into the database. They can now be used for targeted search and thematic classification of such objects in satellite data