Thermal Infrared Spectral Features And Quantitative Deconvolution Of Geological Samples

A sophisticated theory for mineral identification and geological mapping technology, based on spectra band from visible to near-infrared, has been more matured with the development of hyperspectral remote sensing. However, it is difficult to identify the no hydroxyl minerals for the limited spectra region. One approach to overcome the limitation is to expand the spectra band to thermal infrared region, which can detect the Si-O bond vibration of rock-forming minerals. Thus, combining use of the two aforementioned theories will improve the ability and precision of mineral identification by remote sensing.However, mineral information extraction based on emissivity spectra has developed very slowly, compared with the reflectance spectrum, because it is difficult to obtain quantitative thermal emissivity with many influencing factors such as particle size, crystal orientation and construction. In this paper, mechanisms of emissivity of geological materials determined by particle size and the viability of using linear spectral deconvolution method to derive minerals information from rocks based on thermal emissivity spectra (7-14μm) are studied.The emissivity spectral mechanism and classification of main rock-forming minerals in the thermal infrared(7～14μm) are systematically analyzed and summarized. The results show that silicate (including no hydroxyl minerals), sulfates, carbonates, phosphates, oxides, hydroxides and other minerals can be identified in this region. The identification capabilities and precision of remote sensing can be effectively improved.Spectral contrast of mineral and rock spectra decreased with the decrease of particle size. It can be considered that two important surface properties affect the spectral contrast—olume scattering and the cavity effect. Both volume scattering and the cavity effect reduce the spectral contrast of reststrahlen bands. However, the cavity effect increases the emissivity at all wavelengths, while volume scattering increases the emissivity within the reststrahlen bands, but decreases the emissivity at other wavelengths(transparency bands). Field and laboratory measurements of thermal emissivity spectra of6mineral mixtures and6rock samples ranging from2to5end-members with particle diameters of0.4-0.71mm and0.71-1mm were obtained to test the viability of linear spectral deconvolution and predict mineral abundances from infrared data. The deconvolution results show that the assumption of linear mixing is valid and enables mineral percentage prediction to within2%on average with residual errors of less than10-3total emissivity. Extensive error analysis and model testing confirm the appropriateness of linear deconvolution as a useful and powerful tool to examine complexly mixed emissivity spectral in the laboratory and the field.The results of this study provide a foundation for remote sensing analyses of thermal infrared identification and abundance deconvolution of minerals and rocks.