[目的] 无人机高光谱遥感是获取田间尺度作物生长参数的新型手段,如叶面积指数 (leaf area index, LAI) 的无人机快速观测对作物生长监测具有重要意义.[方法] 研究以河北省衡水市冬小麦为研究对象,利用以多旋翼无人机为平台搭载Cubert UHD185成像高光谱传感器获取了冬小麦乳熟期的无人机高光谱影像数据;在无人机飞行的冬小麦试验田,利用LAI 2200进行了同步观测试验.该研究利用植被辐射传输模型PROSAIL模拟小麦冠层反射率数据,进而模拟9种植被指数 (RVI, NDVI, EVI2, OSAVI, MSA-VI2, TCARI/OSAVI, RENDVI (red edge NDVI), MSI, S2REP).将模拟的植被指数与LAI进行相关性分析,分别构建LAI反演模型并通过拟合效果选择最优的反演模型.然后基于LAI最优反演模型利用无人机高光谱遥感数据反演冬小麦乳熟期的LAI.最后利用地面实测LAI数据对反演结果进行了验证.[结果] 9种植被指数中包含红边波段的RENDVI和S2REP与LAI具有高度相关性,而且消除了在小麦LAI高值区时其他植被指数对LAI饱和的问题.基于RENDVI指数模型模拟的LAI与模型模拟的LAI之间RMSE为0.51,无人机高光谱数据LAI反演结果与地面实测值高度拟合 (R2=0.83, RMSE=0.16, NRMSE=10%, n=25,P<0.001),因此RENDVI是用于估算LAI的较理想的植被指数.[结论] 无人机高光谱是获取小麦LAI的有效手段,该研究为利用无人机高光谱数据监测作物生理生态参数提供了参考.%Unmanned aerial vehicle(UVA) hyperspectral remote sensing is a new means for obtaining crop growth parameters at field scales. Rapid observation of leaf areas index(LAI) is of great significance to crop growth moni-toring. In this study,the winter wheat of Hengshui city,Hebei province was used as the research object,and the hyperspectral image data of UVA were obtained by using Cubert UHD185 hyperspectral sensor with a multi-rotor unmanned aerial vehicle as the platform. The LAI 2200 was used for simultaneous observation. The vegetation radi-ation transmission model PROSAIL was used to simulate the wheat canopy reflectance data,then nine vegetation in-dices (RVI,NDVI,EVI2,OSAVI,MSAVI2,TCARI / OSAVI,RENDVI,MSI,S2REP) were calculated. The LAI inversion model was constructed and the optimal inversion model was selected by the fitting effect. Then the LAI optimal inversion model was used to invert the LAI of the winter wheat during the milky stage using the UAV hyperspectral remote sensing data. Finally,the inversion results were verified by ground measured LAI data. The results showed that RENDVI and S2REP,which contained the red band in the nine vegetation indices,were highly correlated with LAI and eliminated the problem of LAI saturation when other LAI values were high. The RMSE of the LAI and the LAI of the model simulation based on the RENDVI index model was 0.51,and the LAI inversion results of the UAV hyperspectral data were highly fitted with the measured values of the ground(R2= 0.83,RMSE= 0.16,NRMSE=10%,n= 25,P<0.001), so RENDVI was the best vegetation index used to estimate LAI. This study provided a reference for monitoring the physiological and ecological parameters of crops using UAV hy-perspectral data.
展开▼
