基于混合像元分解模型的TM6/ETM+热红外波段地表比辐射率估算

Estimation of TM6/ETM+ Thermal Band Emissivity Based on Spectral Unmixing Model

地表比辐射率是热红外遥感中的关键参数,在TM6/ETM+热红外波段地表比辐射率估算过程中亟待解决的是混合像元问题。针对该问题选取湖南省张家界永定区TM6影像,基于混合像元分解模型,综合采用散点图和PPI方法提取端元,得出研究区平均比辐射率估算值为0.988。经验证发现纯净像元比辐射率估算精度较高,估算值与实际值的偏差均小于0.003,4种不同土地利用类型的估算精度由高到低依次为水域、林地、耕地和建设用地。

The emissivity of land surface is a key parameter of thermal remote sensing,the problem of spectral mixing urgently need to solve when emissivity of the TM6/ETM＋ thermal data is estimated.An algorithm based on spectral unmixing model was proposed to solve this problem,and it is used to evaluate the emissivity of Yongding County of Zhangjiajie in Hunan Province from TM6 image.The key step of this algorithm is end member selection.In order to improve the precision of end member selection,the method integrating plotting graph method and PPI method is used,and it is validated that this method can get accurate result,the average RMSE of spectral unmixing image is 0.01423.The validation of emissivity evaluating result showed that the algorithm of emissivity evaluation based on spectral unmixing model can get precise result,the estimation precision of pure pixels is high,the bias of estimated value to real value is not larger than 0.003.The emissivity precision of four land using types was validated and the result showed that the precision of water area was best and the precision of building area was worst,and the ascend order of precision is water area,wood land,farm land and building area.