Leaf Area Density Inversion For Broadleaf Forest Based On Terrestrial And Ariborne LiDAR Data

The vertical distribution of forest canopy leaf area has an important influence on the radiation balance of forest and the exchange of mass and energy with the atmosphere.It is a key factor in the bottom-up process from the leaf to the whole ecosystem.Leaf Area Density(LAD)is an important parameter to represent the vertical distribution of leaf area.Accurate inversion of LAD profile of vegetation is of great significance to the study of the whole ecosystem.LiDAR technology has developed rapidly in recent years.It has the advantages of all-time,all-weather and active non-contact real 3D measurement of targets.It is widely used in quantitative measurement and inversion of forest parameters.In order to retrieve LAD accurately,a branch-leaf separation algorithm based on terrestrial LiDAR data is established,and the LAD of single tree and broad-leaved forest canopy based on terrestrial and airborne LiDAR data is retrieved by volume-based canopy analysis method.Through comparative analysis,a method of reconstructing LAD profile is proposed to eliminate the influence of scanning blind zones on LAD inversion.The main works and conclusions are as follows:(1)Leaf extraction algorithm: Based on the difference of normal vectors between leaf point clouds and non-photosynthetic component point clouds,a neighborhood interior point cloud normal vector difference method is proposed,and leaf point clouds are extracted from LiDAR point clouds of two Magnolia trees.The segmentation results are verified by sampling analysis and visual inspection.The leaf extraction rates of two Magnolia trees are 86.53% and 84.63%,respectively.(2)LAD inversion of single tree and broad-leaved forest: Based on LiDAR data of single Magnolia foundation,LAD inversion model of voxel-based Canopy Profiling(VCP)was studied and validated.The selection of the optimum volume element size was discussed.The results show that when the size of volume element is set to the average point spacing of the input model point clouds,the LAD inversion results are the most accurate,and the leaf area distribution curve is highly consistent with the vertical structure of canopy,and the accuracy of LAI verification can reach more than 90%.Secondly,the LAD inversion model based on VCP was successfully applied to the canopy LAD inversion of the regional birch forest sample by topographic correction and block processing.(3)Inversion of LAD from terrestrial and airborne LiDAR data: Firstly,registration of birch forest sample terrestrial and airborne LiDAR data is completed by using quadrate corner GPS coordinates combined with multi-site point cloud mosaic principle and ICP algorithm.The experimental results show that after registration,the terrestrial LiDAR data and airborne LiDAR data are basically overlapped.The average distance between the two sets of points is less than 0.2m,which is much less than the resolution of airborne LiDAR data of 0.5m.Secondly,LAD profile of terrestrial LiDAR data and airborne LiDAR data of Betula platyphylla quadrate foundation were inverted by volume element sizes of 0.05 m * 0.05 m * 0.05 m and 0.5 m * 0.5 m * 0.5 m,respectively.Through comparative analysis,the occurrence heights of scanning blind zones of terrestrial LiDAR and airborne LiDAR on each quadrate foundation are determined.On this basis,the LAD profile is reconstructed and the final LAD profile of the quadrat is obtained,which eliminates the influence of scanning blind zones on LAD inversion.