Main content area

Predicting particle-size distribution using thermal infrared spectroscopy from reclaimed mine land in the semi-arid grassland of North China

Bao, Nisha, Liu, Shanjun, Zhou, Yongchun
Catena 2019 pp. 104190
arid lands, calibration, desertification, ecosystems, grasslands, infrared spectroscopy, land use, least squares, mined soils, models, monitoring, particle size, particle size distribution, prediction, reclaimed soils, remote sensing, sand fraction, spectral analysis, support vector machines, China
Particle-size distribution is one of the vital properties of semi-arid and reclaimed soils that closely correlates with desertification. The thermal infrared (TIR) spectrum between 8 and 14 μm has the potential for particle-size monitoring in remote sensing. This study aims to investigate the characteristic TIR spectra of reclaimed mine soils and establish a particle-size prediction model. The characteristics of the reststrahlen band (8–9.5 μm) could be used to distinguish the coarse soil content that occurs in various land use and reclamation areas. A significant negative correlation was noted between the coarse sand content and the TIR spectrum (8–14 μm). The support vector machine calibration model exhibited a higher prediction accuracy for estimating the coarse sand content, with a cross-validated R2 of 0.95 and root mean square error (RMSE) of 3.01%, than the partial least squares regression model. These outcomes provide a theoretical basis and technical support for particle-size distribution estimations using TIR spectroscopy in semi-arid and reclamation areas. Hence, this study proposes that the spectral characteristics and model undergo further testing and optimization before wider application for the observation of semi-arid and reclaimed mine land ecosystems.