To gain a higher precision of retrieval , the method of pixel information decomposition is introduced in . the method is to divide ground objects correspond to a pixel in to several basic components : soil , vegetation , moisture , water , and ice , and then , using remote sensing data , calculate the contents of every components according to the model by which the component ' s spectrum compose 为获得较高的反演精度，采取象元组分信息分解的方法，将象元对应地物分解为土壤、植被、水分、水体、雪被等基本组分，然后根据这些组分的光谱组合模型用遥感数据计算象元中各组分含量，再根据组分含量反演半球反射率和反照率。
And the exponent relation between water content and soil spectrum are demonstrated , then the function of water content and soil spectrum are put forwarded , and the curve of absorption coefficient of water content in soil measured . basing the study on the course of reflection and scattering taking place on soil surface , a function of relation between soil spectrum and the roughness of clod was put forwarded . then a mathematic model on the relation between soil spectrum and water content and surface roughness are proposed 为建立象元组分信息模型，本文对土壤光谱与土壤含水量及表面粗糙度的关系进行了深入研究，论证了含水量与土壤光谱呈指数关系，提出了土壤含水量与土壤光谱的函数关系式，并测定了土壤水吸收系数；在研究土壤光谱反射与散射过程基础上，提出了土块级土壤粗糙度与土壤光谱的函数关系式，进而提出了湿润、粗糙土壤反射率的数学模型，并通过实验证明了模型具有较高的精度。
