结合植被指数与作物高度反演冬小麦叶面积指数

RETRIEVAL OF WINTER WHEAT LEAF AREA INDEX BASED ON VEGETATION INDEX AND CROP HEIGHT

关键词:  叶面积指数植被指数作物高度冬小麦无人机

Key words:  leaf area index  vegetation index  crop height  winter wheat  unmanned aerial vehicle

摘要:

［目的］叶面积指数LAI（Leaf Area Index）是反映作物长势的关键参数之一。目前，基于无人机影像进行LAI反演多注重影像光谱信息的应用，但是由于高分影像存在强烈的光谱异质性以及无法区分高密闭度植被垂直方向枝叶的光谱特征等不足，在反演作物LAI时，需要探讨作物高度等结构参数对LAI反演的影响。［方法］文章以冬小麦为例，将无人机影像的光谱信息与点云数据相结合，共同构建LAI反演模型，并与单利用光谱信息的一元线性LAI回归模型进行对比，探讨作物高度信息对LAI反演精度的影响。［结果］（1）无人机影像获取的点云数据能有效反演作物高度，其决定系数R2=061，均方根误差RMSE=002； （2）基于作物高度和植被指数Ⅵ（Vegetation Index）反演LAI的二元模型（Adjust R2=038，Adjust RMSE=055）优于单用植被指数反演LAI的一元模型（Adjust R2=029，Adjust RMSE=059），［结论］研究表明作物高度和光谱信息结合的反演模型能够提高作物LAI的反演精度，同时表明作物高度因子在LAI反演中具有重要的应用价值。

Abstract:

LAI (Leaf Area Index) is one of the key parameters in reflecting crop growth. At present，LAI inversion based on UAV images mainly focuses on the application of spectral information. However，due to the lack of strong spectral heterogeneity of high resolution images and the saturation of vegetation spectral with high tightness of high resolution images，it is necessary to explore the influence of the crop height and other structural parameters on LAI inversion when retrieving crop LAI. However，few existing studies involve studies on crop LAI inversion by structural parameters such as crop height. In view of this，this study took winter wheat as an example, combined the spectral information of UAV image with point cloud data information to build the LAI inversion model，and compared it with the monocular linear LAI regression model solely using spectral information to explore the influence of crop height on LAI inversion accuracy. A variety of vegetation indexes were calculated based on spectral information of UAV image, to select the optimal vegetable index GLA (Green leaf algorithm) that was most suitable for building winter wheat LAI inversion model，and to construct a binary linear inversion model of winter wheat LAI jointly with CH(Canopy Height)，the crop height information obtained based on point cloud data inversion. The measured value of LAI and measured height were verified as the truth values. The adjusted determination coefficient Adjust R2 and adjusted root mean square error Adjust RMSE were used as the evaluation indexes of the optimal parameters to join the regression model. The determination coefficient R2 and root mean square error RMSE were used as the evaluation indexes of the optimal regression model. The results showed that: (1) The point cloud data obtained by UAV image could effectively invert crop height, with the determination coefficient R2=0.61 and root mean square error RMSE=0.02. (2) Dual model (Adjust R2=0.38, Adjust RMSE=0.55) for LAI inversion based on the plant height and Vegetation Index VI (Vegetation Index) was superior to univariate model (Adjust R2=0.29, Adjust RMSE=0.59) for LAI inversion solely based on Vegetation Index. The study shows that plant height and spectrum information combined inversion model can improve the inversion precision of the crop LAI, the addition of height information enhances the inversion accuracy of LAI by the inversion model. Meanwhile, the addition of height information improves the reliability of the inversion model of LAI. At the same time, it is verified that the remote sensing image obtained based on UAV platform is feasible to invert crop height information and the inversion result achieves an ideal result. What′s more，it shows that the crop height factor is of great significance in the LAI inversion. It extends the practice and application of UAV image data in the field of the crop LAI inversion，and it adds some new experience to the application of cost effective UAV remote sensing system in precision agriculture.