Proceedings Article | 14 May 2014
KEYWORDS: Soil science, Reflectivity, Remote sensing, Data modeling, Nitrogen, Particles, Statistical modeling, Analytical research, Absorption, Chemical analysis
Soil organic matter(SOM) is an important composition of soil. Soil mechanical composition
determines soil physical properties. Retrieving soil organic matter and mechanical components by
remote sensing is an important research content of soil remote sensing. In this study soil spectra data
are measured in Duolun county in the Inner Mongolia autonomous region, in China. Based on
statistical analysis of soil reflection spectra characteristics, 12 high spectral indexes, including soil
reflectance, reflectance ratio and normalized difference soil index（NDSI）, were constructed for
building inversion models of soil organic matter , total nitrogen (TN), soil mechanical composition
Firstly, calculating correlation coefficient between 12 high spectral indexes and soil organic matter,
total nitrogen, soil mechanical composition, then choosing high spectral indexes whose absolute
values of correlation coefficient greater than 0.5, finally by using linear regression analysis method
were biult inversion models of soil composition parameters. With the help of comprehensive analysis,
eliminating poor quality models, finally are established inversion model of part of soil composition
parameters, including SOM, TN, P(2-0.2), P(0.2-0.02), P(<0.002) etc. In view of total salt factor, P(>2)
factor and P(0.02-0.002) factor, this study failed to obtain a satisfactory inversion model. Limits on the
number of samples may be an important influence factors.
Research shows that in the study area there are 3 types of soil spectral curve. Between bow area of soil
reflection spectrum and soil organic matter, total nitrogen and soil mechanical composition there are a
deterministic relationship, and the deterministic relationship can be described. In this study, NDSI as
high spectral characteristic index is one of the most successful index for retrieving satisfactory model.
The position of 2 characteristic wavelength in NDSI is determined according to analyzing feature of
bow area of the soil reflection spectrum. The position of 2 characteristic wavelength just reflects
differences between different types of soil. Research shows that the NDSI in process of retrieving soil
compositions has great application potential. Can be expected application of similar model in
inversion model study by using hyperspectral remote sensing or multispectral remote sensing.
