The study of mineral mapping in black soil using TASI thermal infrared data, taking the Baiquan area of China as an example
Jiang Chen,Zhizhong Li,Xiaoyan Chen,Huimin Dai
DOI: https://doi.org/10.1007/s12145-022-00765-z
2022-01-20
Earth Science Informatics
Abstract:Soil is composed of many minerals. Because minerals have different spectral features of emission and reflection, the spectral characteristics of soils are different. Although there are mineral spectral libraries for the spectral analysis of pure minerals (mainly primary minerals), it is difficult to quantitatively determine the composition and content of minerals in soils with secondary minerals and clay minerals. Thermal infrared (8 ~ 14 μm) hyperspectral remote sensing has the ability to identify minerals, but there are few studies on mineral identification in soil. This study takes the Baiquan area of China as the research area, and hyperspectral infrared sensors, known as the Thermal Airborne Spectrographic Imager (TASI), are used for image analysis and mineral mapping. The emissivity data are formed by the emissivity normalization method. After interference information such as humidity is removed by masking, the spectral endmembers in the soil are classified by the sequential maximum angle convex cone (SMACC) method; thus, an endmember spectrum is formed. Based on a comparison of the endmember spectra from the TASI images and the field-measured spectra with the mineral spectra in the ASU and USGS spectral libraries, the main mineral types in the soil are determined. Finally, a map of mineral abundance in soil is formed by matched filtering. The results show that the main mineral types in the soil of the Baiquan area include palygorskite, anorthite, halloysite, montmorillonite and carolinite. Additionally, in the visible shortwave infrared (VIS-SWIR) spectral ranges, hydroxyl minerals can be distinguished, and the absorption characteristics of clay minerals are based mainly on their absorption depth. However, if noise is generated in the case of soil with humidity, it is not easy to identify hydroxyl minerals, but the thermal infrared band can have a greater ability to distinguish minerals in this situation. In this study, the quantitative inversion of soil mineral composition based on thermal infrared hyperspectral data and the method of drawing a soil mineral abundance map are proposed. Compared with the traditional laboratory analysis of soil samples, the proposed strategy has the advantages of being nondestructive, rapid, and widely applicable, and compared with previous mineral mapping based only on the VIS-SWIR range, the proposed strategy improves the ability of fine identification of minerals in the thermal infrared range. These results provide a basis for the study of fine soil mapping and mineral distribution in soil based on full-spectrum hyperspectral images.
geosciences, multidisciplinary,computer science, interdisciplinary applications