基于Landsat-8遥感数据和PROSAIL辐射传输模型反演叶面积指数  

Retrieving leaf area index using PROSAIL radiative transfer model based on Landsat 8 image

作　　者：杜育璋[1] 姜小光[1] 张雨泽[1] 黄成[1] 刘朝霞[2] 刘亮[1,3] 

机构地区：[1]中国科学院大学,北京100049 [2]中国科学院新疆生态与地理研究所,新疆乌鲁木齐830011 [3]民政部国家减灾中心,北京100124

出　　处：《干旱区地理》2016年第5期1096-1103,共8页Arid Land Geography

基　　金：国家重点基础研究计划(2013CB733402);国家自然科学基金项目(41231170;41471297)

摘　　要：为了探讨Landsat 8 OIL数据在LAI大范围反演方面的应用潜力,使用Landsat 8 OIL影像,通过PROSAIL辐射传输模型,采用3种波段组合(Band2-7,Band2-5,Band3-5)建立了3个模拟冠层反射率-叶面积指数(LAI)查找表,用2种代价函数(Geman and Mc Clure代价函数,均方根误差代价函数)实现了对玉米、土豆、森林LAI的定量反演,并用LAI-2200测量数据作为相对真值对反演精度进行评价。结果表明:(1)使用Landsat 8数据,通过PROSAIL模型反演叶面积指数的精度是可以接受的,RMSE范围为在[0.892 4,1.205 0],R2范围为[0.721 3,0.873 3]。(2)Band5(近红外),Band4(红)Band3(绿)的波段组合反演效果在3种组合中精度最高,平均RMSE=0.993 1,R2=0.787 3。(3)Geman and Mc Clure代价函数比常用的均方根误差代价函数得到了更高的反演精度,平均RMSE=0.940 5,R2=0.817 5。(4)相对最优的反演策略是Band5,Band4,Band3的波段组合结合GM代价函数,RMSE=0.892 4,R2=0.873 3。(5)存在玉米土豆的反演值普遍低于测量值,而森林的反演值普遍高于测量值的问题。Landsat 8 is the latest satellite in the Landsat program launched on February 11,2013. Landsat 8＇s Operational Land Imager（OLI）improves on past Landsat sensors in radiation resolution and the scan mode. In order to discussing the potential of Landsat 8 OIL data for LAI inversion application,a leaf area index（LAI）and canopy reflectance lookup table（LUT）was established by using the PROSAIL radiative transfer model and Landsat 8 OIL image for the LAI inversion of corn,potatoes and forest. Firstly,the paper set PROSAIL model parameter values according to previous research to generate a lookup table. Secondly,the paper divided Landsat 8OIL reflectance data into three groups based on wavelength respectively,they are 6-bands group（Blue,Green,Red,NIR,Swir1,Swir2）,4-bands group（Blue,Green,Red,NIR）and 3-bands group（Green,Red,NIR）. Green,red and NIR are often used in many studies,and the paper took two other bands of Landsat 8 OIL into consideration to assess their applicability. And then,the paper found a series of records with the smallest differences in the lookup table in the corresponding band by using RMSE and Geman and Mc Clure function as a cost function（hereinafter referred to as RMSE cost function and GM cost function）. RMSE cost function is widely used.However,GM cost function is more effective when analyzing the reflectance of NIR band and red band in vegetation region,because GM cost function can decline the negative impact when the absolute value of one parameter is significantly higher than other items. Finally,the paper regarded the corresponding LAI value of the record as the inversion result. The result demonstrated as follows：（1）The retrieval accuracy was acceptable,RMSE was in the range of 0.892 4 to 1.205 0,R2 was in the range of 0.721 3 to 0.873 3;（2）The band combination of band5（near infrared）,band4（red）and band3（green）got the highest accuracy among three band combinations（RMSE=0.993 1,R2= 0.787 3）;（3）GM cost function had higher accuracy than

关 键 词：叶面积指数 PROSAIL模型 LANDSAT 8 查找表 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]