基于反射率模拟模型的黑土有机质含量估测

定量分析了黑龙江省黑土室内高光谱反射率曲线特征,确定了影响反射光谱曲线的主要特征控制点,建立了黑土光谱反射率模拟模型并对其进行评价,分析比较反射率数据、模拟后的光谱数据与土壤有机质含量的关系,建立了土壤有机质含量光谱预测模型。结果如下:有机质是小于1000 nm范围黑土反射光谱特征的决定因素,随有机质含量变化,黑土光谱反射率在该范围呈现单/双吸收谷特征;黑土反射光谱曲线在450~930 nm范围内有5个主要特征控制点;黑土反射率模拟模型能较准确地描述黑土反射光谱曲线,直线模型的模拟效果更好;以反射率模拟模型系数为自变量的有机质含量预测模型优于基于反射率及其一阶微分的模型,说明反射率模拟模型的曲线控制点选择合理且有代表性,反射率曲线模拟方法能够准确描述黑土的实际光谱反射率。

Black Soil Organic Matter Content Prediction Based on Reflectance Simulation Models

The hyperspectral reflectance characteristics of black soil in Heilongjiang province were analyzed quantitatively,and then the main characteristic controlling points of reflectance were determined and used to build soil reflectance prediction models;the relationship between organic matter content and reflectance and the coefficients of simulating models were studied,Black soil organic matter content spectral prediction models were built,and the feasibility of hyperspectral reflectance simulatiib method was discussed.The results are as follows:(1) Organic matter content is the determining factor of black soil reflectance characteristics in the range less than 1 000 nm.When the content is low,the covering effect of organic matter on the black soil parent matrix reflectance characteristics is very weak,there are two absorption vales at 500 and 640 nm;when the content reaches a certain content(about 5%),the reflectance characteristics of black soil parent matrix are totally covered by organic matter,and there is only one large absorption vale in the region caused by organic matter.(2) The spectral characteristic controlling points of black soil hyperspectral reflectance in the range of 450-930 nm are located at 450,500,590,660 and 930 nm,and divide the black soil reflectance into four parts.(3) Simulation models(linear,quadratic) rightly describe the characteristics of black soil hyperspectral reflectance,and the linear piecewise model shows a better performance.(4) The organic matter content prediction models with the coefficients of reflectance simulation models as independent variables are more precise than that based on soil reflectance and its derivate,which indicates that the characteristic controlling points for reflectance simulation models are selected reasonably and representatively,and the simulation models partly solve the data redundancy problem of soil hyperspectral reflectance,and improve the precision of black soil organic matter content prediction models with remote sensing methods.Reflectance simulating method can be used for data simplification and compression,data redundancy removal,organic matter and other soil parameters' remote sensing studies.