Discrimination of rock types and main rock-forming components in Bulgarian territories through spectral reflectance characteristics

Abstract

Remote sensing studies on the spectral reflectance of natural formations – rocks, products of different physical and chemical processes in the Earth’s crust and its surface – are reported. Based on spectral reflectance characteristics (SRC) of selected representative rocks with different structure, mineral composition and colour characteristics – sedimentary (psephite-conglomerate) and metamorphic (kyanite schist) – and by applying statistical and deterministic methods the rock types and the main rock-forming components were discriminated. The spectral data were obtained in laboratory using a multichannel spectrometer developed in STIL-BAS, which delivers data in the spectral range 480–810 nm. The a priori mineralogical information about the specimens studied, the actual SRC data, affine and perspective transformations of the colour coordinates of SRC were all utilized to perform a primary classification of the main constituents of the specimens. Cluster analysis was applied and its results were used to design the necessary grouping variables as required for the implementation of linear discriminant analysis. The discriminant functions were constructed from the SRC values for a selected set of wavelengths and of transformed data. Classification of the main ingredients and the rock types were obtained. The proposed novel approach to the spectral discrimination of main rock-forming components is believed relevant for predictive classification of other specimens of rocks through discriminant functions.