基于Landsat 8数据反演地表发射率的几种不同算法对比分析

(1.长安大学 地质工程与测绘学院,陕西 西安 710054; 2.国土资源部退化及未利用土地整治工程重点实验室,陕西 西安 710075; 3.长安大学 地球科学与资源学院,陕西 西安 710054; 4.长安大学 环境科学与工程学院,陕西 西安 710054)

地表发射率; 遥感反演; 归一化植被指数; Landsat 8数据

Comparison of several different algorithms to retrieve land surface emissivity using Landsat 8 data
WANG Li-xia1,2,SUN Jin-hua3,LIU Zhao4,ZHANG Shuang-cheng1,YANG Yun1

(1.School of Geology Engineering and Surveying,Chang'an University,Xi'an 710054,China; 2.Key Laboratory of Degraded and Unused Land Consolidation Engineering,The Ministry of Land and Resource,Xi'an 710075,China; 3.School of Earth Science and Resources,Chang'an University,Xi'an 710054,China; 4.School of Environmental Science and Engineering,Chang'an University,Xi'an 710054,China)

surface emissivity; remote sensing inversion; normalized vegetation index; Landsat 8 data

DOI: 10.13800/j.cnki.xakjdxxb.2019.0220

备注

为了分析研究不同地表发射率反演算法的精度和适用性,文中选取西安市的遥感影像Landsat 8为基础数据,运用ENVI,ArcGIS等软件,首先对Landsat 8数据进行预处理,提取西安市的NDVI影像; 然后,建立决策树模型得到西安市地表分类影像,并基于像元二分模型反演得到植被覆盖度,基于NDVI得到4种不同算法的地表发射率; 最后,以精度0.01的MODIS LSE产品为标准数据,从像元尺度上对比分析了4种算法的精度,并依据回归决策树方法的分类结果,对比分析了不同算法在各类地表覆盖类型上的发射率反演差异。结果 表明:在像元尺度上,植被指数混合模型法与NDVITEM改进算法精度较高且较为接近; 从不同下垫面的反演差异来看,在植被区域4种算法之间的差异较小,而对于水体区域,4种算法之间的差异较大; 从反演方法的适用性而言,植被指数混合模型法与NDVITEM改进算法较为适合本研究区。

In order to analyze the accuracy and applicability of different surface emissivity inversion algorithms,the Landsat 8 data of remote sensing imagery in Xi'an was selected,and software such as ENVI and ArcGIS was used.First,the Landsat8 data is preprocessed to extract NDVI images from Xi'an.Then,the decision tree model is established to obtain the surface classification image of the city,the vegetation coverage is obtained based on the inversion of the pixel binary model,and the surface emissivity of four different algorithms is obtained based on NDVI.Finally,with the accuracy of 0.01 MODIS LSE products as the standard data,the accuracy of the four algorithms is compared and analyzed from the pixel scale,and based on the classification results of the regression decision tree method,the differences of emissivity inversion of different algorithms on different types of surface coverage are compared and analyzed.The results indicate that the vegetation index hybrid model method and the NDVITEM improved algorithm are more accurate and close to each other on the pixel scale.From the inversion difference of different underlying surfaces,the difference between the four algorithms in the vegetation region is small.For the water body region,the difference between the four algorithms is great; from the applicability of the inversion method,the vegetation index hybrid model method and the NDVITEM improved algorithm are more suitable for the study area.