关键词: 高光谱, 有机质, 含水量, 方差分析, 交互作用

Abstract: 【Objective】The objective of this study is to reveal the rule of soil moisture and organic matter on the high spectrum quantitatively and to provide a basis for improving the accuracy of spectral estimation of soil moisture and organic matter.【Method】Ninety brown soil samples collected from the Tai'an city, Shandong province were used as research materials, and their outdoor spectrum, soil moisture and organic matter were obtained. At the same time, smoothing denoising pretreatment of the spectral curves were carried out by Savitzky-Golay filter. Soil samples were divided into 9 groups according to water and organic matter contents. By using comparative method, 9 groups of original spectral data were analyzed, and the effect of soil water and organic matter on the spectrum was discussed. Then, correlation analysis was used to analyze the correlation between water, organic matter and raw spectral reflectance (R), first order differential reflectance (D (R)) and grouping spectra. Under the basic assumption that other factors were identical, two factor variance interaction analysis was used to analyze the degree of water, organic matter and their interaction to the soil spectral reflectance and spectral derivative quantitatively. According to the law of interaction between soil water and organic matter, characteristic factor was selected on the basis of the principle that the correlation coefficient is relatively large and the interaction is small. Finally, partial least squares regression model was established to predict the soil organic matter content of hyperspectra, and the effectiveness of the model which established with the factors selected according to the interaction was analyzed.【Result】 The results showed that water had the main effect on the reflectance of soil spectral reflectance in the range of field capacity, and the interaction between water and organic matter was exist objectively. When the soil moisture content was less than 10%, the original spectra of 600-1 800 nm could better reflect the effect of organic matter. And when the soil moisture content was more than 15%, the role of organic matter was almost covered by the action of water. The role and interaction of soil water and organic matter reached significant level in 360-1 800 nm, 410-1 800nm and 509-1 800 nm, And all of them reached significant level in 1 951-2 450 nm (α=0.05). The effect on soil spectrum from large to small was water, organic matter and interaction. The effect of water on soil spectrum was about 5 to 8 times in 425-1 800 nm and 8 to 12 times in 1 950-2 300 nm than that of organic matter. The effect of organic matter on soil spectrum was about 2 times as much as interaction in 350-2 500 nm. After the first-order differential transformation, the effect of soil water at 450-530 nm, 600-790 nm, 1 019-1 027 nm, 2 000-2 020 nm and 2 045-2 075 nm was enhanced, but weakened in other bands. The effect of soil organic matter was enhanced at 471-824 nm, 851-949 nm, 967-1 140nm, 1 172-1 340nm, 1 379-1 428 nm, 1 450-1 770 nm, 1 953-2 122 nm, 2 174-2 199 nm, and 2 271-2 342 nm, and weakened at other bands. The interaction between water and organic matter also changed at different wavelengths, but change was relatively weakened than that of the soil water and organic matter. Based on the characteristics selected by the interaction between soil water and organic matter, the accuracy of the hyperspectral model of soil organic matter was improved. The determination coefficient R2 of the 16 test samples increased from 0.6764 (without considering the interaction) to 0.7934.【Conclusion】The researches showed that the effect of organic matter on the spectral reflectance may not be considered in the inversion of soil water content. In the inversion of organic matter content, not only the influence of water on the reflectance should be eliminated, but the interaction between water and organic matter on the spectrum should be considered. The accuracy of spectral estimation of soil organic matter can be effectively improved when considering the interaction of water and organic matter on soil spectrum.

Key words: hyper-spectrum, soil organic matter, water content, variance analysis, interaction