基于光谱特征与PLSR结合的叶面积指数拟合方法的无人机画幅高光谱遥感应用

以冬小麦LAI为研究对象,利用孕穗期、开花期和灌浆期获取的无人机UHD185高光谱影像以及同步测定的地面数据(冬小麦冠层ASD反射率和冬小麦LAI),论证光谱特征(红边参数或植被指数)与偏最小二乘回归算法结合的改进型LAI拟合方法在无人机画幅高光谱遥感LAI探测方面的应用价值。首先,从光谱反射率相关性和植被指数相关性两方面比较UHD185与ASD,验证UHD185数据精度;结果表明,第3~第96波段(458~830 nm)的无人机UHD185高光谱数据具有较好的光谱质量,适宜探测冬小麦LAI。其次,分析光谱特征(6种植被指数和4种红边参数)与LAI的相关性,并通过独立验证和交叉验证方法,依次对基于红边参数或植被指数的传统LAI拟合方法和改进型LAI拟合方法的冬小麦LAI预测精度进行评价,相比于传统LAI拟合方法,改进型LAI拟合方法能大幅度提高冬小麦LAI的预测精度,特别是PLSR+REP。研究结果证实,改进型LAI拟合方法能更加充分地利用无人机UHD185高光谱数据预测冬小麦LAI,可望为无人机高光谱遥感的作物理化参数探测提供几点可借鉴的思路。

Application of an Improved Method in Retrieving Leaf Area Index Combined Spectral Index with PLSR in Hyperspectral Data Generated by Unmanned Aerial Vehicle Snapshot Camera

The objective of this study was to demonstrate the value of an improved method of retrieved leaf area index (LAI) based on unmanned aerial vehicle (UAV) hyperspectral data combined spectral characteristics, as red edge parameters (REPs) and vegetation indices, with partial least squares regression (PLSR). We got UAV UHD185 hyperspectral images at booting, anthesis, and filling stages in winter wheat. And synchronously measured ASD hyperspectral data and winter wheat LAI. We compared UHD185 data with ASD data in terms of the correlation between reflectivity and vegetation indices to verify the UAV hyperspectral data accuracy. The band 3 to 96 (458-830 nm) of UHD185 hyperspectral data had better spectral quality and was suitable for detecting winter wheat LAI. We did correlation analysis between spectral characteristics, six kinds of vegetation indices and four kinds of red edge parameters, and LAI, and used two kinds of validation methods, independent validation and cross validation, to analyze the prediction accuracy of winter wheat LAI. Compared with traditional LAI fitting method, the improved LAI fitting method especially PLSR+REPs, greatly improved the prediction accuracy of winter wheat LAI. The above results confirmed that the improved LAI fitting method is able to better utilize UAV UHD185 hyperspectral data to predict LAI of winter wheat. Moreover, it is expected to provide a few new ideas for retrieving crop physical and chemical parameters based on UAV hyperspectral data.