摘要
【目的】定量揭示土壤水分与有机质对高光谱的作用规律,为提高土壤水分、有机质的光谱估测精度提供基础。【方法】以山东省泰安市岱岳区90个棕壤土样为研究对象,进行室外光谱采集、室内土壤水分和有机质测定,运用Savitzky-Golay filter对光谱曲线进行平滑去噪预处理。根据含水量、有机质含量的高低将土壤样本分为9组,运用比较法对9组原始光谱数据进行分析,初步探究土壤水、有机质对光谱的作用规律。然后采用相关分析法,分析水、有机质与土壤原始光谱反射率(raw spectral reflectance,R)、光谱一阶微分变换(first order differential reflectance,D(R))以及分组光谱的相关性。在假定其他影响因素基本相同的条件下,利用有交互作用的双因素方差分析法,定量分析水、有机质对土壤光谱反射率、光谱一阶微分的作用程度及其交互作用。根据土壤水与有机质的交互作用规律,按相关系数较大而交互作用小的原则选取特征因子,采取偏最小二乘回归模型建立土壤有机质含量的高光谱估测模型,分析依据两者交互作用规律选取的因子对提高光谱估测模型精度的有效性。【结果】在田间持水量范围内,水对土壤光谱反射率影响起主要作用;水与有机质对土壤光谱客观存在交互作用,当土壤含水量小于10%时,600—1 800 nm的原始光谱能较好反映有机质的作用,而当土壤含水量大于15%时,有机质的作用几乎被水的作用所掩盖。水、有机质对土壤原始光谱的作用及其交互作用分别在360—1 800,410—1 800,509—1 800 nm达到显著水平,且三者均在1 951—2 450 nm达到显著水平(α=0.05);对土壤光谱的作用程度由大到小依次为:水、有机质、交互作用;在425—1 800 nm水对土壤光谱的作用大约是有机质的5—8倍,在1 950—2 300 nm为8—12倍;在350—2 500 nm有机质对土壤光谱的作用大约是水与有机质交互作用的2倍。光谱经一阶微分变换之后,在450—530、600—790、1 019—1 027、2 000—2 020以及2 045—2 075 nm土壤水的作用增强,而在其他波段处减弱;土壤有机质的作用在471—824、851—949、967—1 140、1 172—1 340、1 379—1 428、1 450—1 770、1 953—2 122、2 174—2 199以及2 271—2 342 nm处得到增强,而在其他波段处减弱。水与有机质的交互作用也在不同波段处有所变化,但相对于土壤水与有机质的作用变化幅度而言是相对减弱的。基于土壤水与有机质的交互作用规律选取的特征因子,所建立的土壤有机质高光谱估测模型精度有所提高,其中16个检验样本的决定系数R2由不考虑交互作用的0.6764提高到0.7934。【结论】研究表明,在反演土壤含水量时,可以不考虑有机质对光谱的影响;而在反演有机质含量时,必须要剔除水对反射率的影响,还要考虑水与有机质对光谱的交互作用。考虑水与有机质对土壤光谱的交互作用,可有效提高土壤有机质的光谱估测精度。
[ 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 （or=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.
出处
《中国农业科学》
CAS
CSCD
北大核心
2017年第8期1465-1475,共11页
Scientia Agricultura Sinica
基金
国家自然科学基金(41271235)
山东省自然科学基金(ZR2016MD03)
关键词
高光谱
有机质
含水量
方差分析
交互作用
hyper-spectrum
soil organic matter
water content
variance analysis
interaction
