联合无人机激光雷达和高光谱数据反演玉米叶面积指数

叶面积指数（Leaf Area Index， LAI）是作物长势监测及产量估算的重要指标，准确高效的LAI反演对农田经济的宏观管理具有重要作用。研究探索了联合无人机激光雷达（Light Detec-tion and Ranging， LiDAR） 和高光谱数据反演玉米叶面积指数的潜力，并分析了LiDAR数据不同采样尺寸、高度阈值、点密度对LAI反演精度的影响同时确定三者的最优值。该研究分别从重采样的LiDAR数据和高光谱影像中提取了LiDAR变量和植被指数，然后基于偏最小二乘回归（Partial Least Square Regression，PLSR）和随机森林（Random Forest， RF） 回归两种算法分别利用LiDAR变量、植被指数、联合LiDAR变量和植被指数构建预测模型，并确定反演玉米LAI的最优预测模型。结果表明：反演玉米LAI的最优采样尺寸、高度阈值、点密度分别为5.5 m、0.55 m、18 points/m²，研究发现最高的点密度（420 points/m²）并没有产生最优的玉米LAI反演精度，因此单独依靠增加点密度的方法提高LAI的反演精度并不可靠。基于LiDAR变量获得的LAI反演精度（PLSR：R²=0.874，RMSE=0.317；RF：R²=0.942，RMSE=0.222）高于基于植被指数获得的LAI反演精度（PLSR： R²=0.741，RMSE=0.454；RF：R²=0.861，RMSE=0.338），而使用组合变量构建预测模型的反演精度（PLSR：R²=0.885， RMSE=0.304；RF：R²=0.950，RMSE=0.203）优于使用单一变量建立的LAI预测模型，其中利用联合LiDAR变量和植被指数建立的随机森林回归模型为最优预测模型。因此，将两种数据源融合在提高植被LAI反演精度方面具有一定的潜力。

Combining UAV LiDAR and Hyperspectral Data for Retrieving Maize Leaf Area Index

Leaf Area Index （LAI） is an important index for crop growth monitoring and yield estimation. Accurate and efficient LAI retrieval plays an important role in the macroscopic management of farmland economy. This study explored the potential of combining UAV LiDAR and hyperspectral data to retrieve maize leaf area index， studied the effects of different sampling size， height threshold and point density of LiDAR data on LAI inversion accuracy， and determined the optimal values of the three parameters. In this study， LiDAR variables and vegetation indices were extracted from resampled LiDAR data and hyperspectral imagery respectively. Then， based on Partial Least Squares Regression （PLSR） and Random Forest （RF） regression， LiDAR variables， vegetation indices， combined LiDAR variables and vegetation indices were used to construct prediction models， and the optimal prediction model for LAI inversion of maize was determined. The results show that the optimal sampling size， height threshold and point density of maize LAI inversion are 5.5 m， 0.55 m and 18 points/m² respectively. We found that the highest point density （420 points/m²） did not obtain the optimal LAI inversion accuracy of maize. Therefore， it is not reliable to improve the inversion accuracy of LAI by increasing point density alone. The LAI inversion accuracies based on LiDAR variables （PLSR： R² = 0.874， RMSE = 0.317； RF： R² = 0.942， RMSE = 0.222） were higher than those based on vegetation indices （PLSR： R² = 0.741， RMSE = 0.454； RF： R² = 0.861， RMSE = 0.338）， and the inversion accuracies of the prediction model constructed using combination variable （PLSR： R²=0.885， RMSE=0.304； RF： R²=0.950， RMSE=0.203） were better than using single variable， in which the random forest regression model established by using combined LiDAR variables and vegetation indices is the best prediction model. Therefore， the fusion of the two data sources has a certain potential in improving the accuracy of vegetation LAI retrieval.