遥感陆面温度 (LST)是目前遥感界面临的重要课题 ,组分温度反演则为LST反演的主攻目标。如果混合像元组分间的温度及发射率存在明显差别 ,就可以从多角度热辐射亮度变化中分离出组分温度的信息。在热辐射矩阵模型的基础上 ,讨论了确定可反演参数和反演组分温度的方法 ,并以AMTIS多角度热红外图像为数据源 ,在进行几何和大气纠正后 ,尝试利用矩阵反演方法分解混合像元中的植被和土壤温度。误差分析表明土壤温度反演结果较好 ,但植被温度的反演精度不高

Now the remote sensing of LST is an important task of us.Components temperature retrieval is the main object of LST retrieval.If the emissitivity or temperature of components is obviously different,the information of component’temperature can be extracted from the multi-anglr radiance data.In this paper,we construct the matrix model of infrared radiance,analyze the correlation of WKｘJ，choose the optimal viewing angles and select the parameters that can be retrieved.Then based on data of the airborne sensor AMTIS developed in China,through geometrical and atmospherical correction,the plant and soil temperature in mixed pixels are separated using matrix retrieval method.The results and the error analysis show that the accuracy of retrieved soil temperature is good,but the result of retrieved plant temperature is not good enough.