Spectral characterization of igneous rocks in the 8‐ to 12‐μm region

Abstract

Variation in the spectral response of silicates in the thermal infrared (TIR) from 8 to 12 μm is the foundation for potentially important applications in the lithologic characterization of terrestrial and extraterrestrial surfaces. The purpose of this study is to elucidate the crystallo‐chemical basis for the variation in spectral behavior in terms which may be useful in remote sensing. This has led to the definition of a new parameter, SCFM, which is defined as the ratio SiO 2 /(SiO 2 + CaO + FeO + MgO) in minerals and rocks. This parameter reflects the degree of depolymerization of the silica tetrahedra and is therefore highly correlated with the structure of TIR reflectance spectra of major silicate minerals in both fine‐ and coarse‐grained igneous rocks. It is also a good numerical descriptor of igneous rock composition. Thus SCFM is proposed as a very useful determinant for spectral identification of igneous rocks. The SCFM parameter was used to assess the effect of variations in location, number, and width of bands in discriminating rock composition. (An unfortunate ambiguity exists in the use of the term “band” which in spectroscopy indicates a feature in a spectrum and in remote sensing signifies a specific wavelength increment or channel. The latter definition is used in this paper.) Spectra of coarse‐particulate mineral and solid rock samples were acquired in the laboratory. A regression analysis of bands varying in width from 0.2 to 1.4 μm versus SCFM results in coefficients of determination (correlation coefficients), R 2 , ranging from 0.88 (for two 1.2‐μm‐wide bands) to 0.97 (for seven 0.2‐μm‐wide bands). Deletion of bands in the ozone absorption region (9.4–10.2 μm) results in a decrease in R 2 by only 0.01. Stimulation of system noise by addition of ±5% random counts to the bands decreases R 2 by only 0.03. Because spectra structure is modified by weathering, alteration, and texture of the samples, the validity of these results is limited to fresh igneous rocks. With this constraint, it is concluded that accurate spectral identification of igneous rock composition is not greatly affected by bandwidth or system noise. It is the overall spectral envelope rather than the fine structure in the TIR which contains most of the information diagnostic of igneous rock composition. This permits design of remote sensing systems with broad bands and corresponding higher signal‐to‐noise, which can be traded off against the important requirement for higher spatial resolution.