Extraction of the Leaf Area Density of Maize Using UAV-LiDAR Data

  Objectives  Leaf area density (LAD) represents the vertical distribution of canopy leaf area, and it is an important structural parameter for crop growth, crop nutritional diagnosis and breeding. Light detection and ranging (LiDAR) can detect the structural information of crop canopy by transmitting multiple pulses and receiving multiple echo signals.

  Methods  Firstly, unmanned aerial vehicle-LiDAR (UAV-LiDAR) is utilized to collect Maize point cloud data with multiple trajectories in 60 plots, which in turn is employed to estimate the leaf area density of Maize with the contact-frequency-based voxel method. Then, combined with the width of the Maize leaf, the optimal voxel size is determined by analyzing multiple voxel sizes (0.2 m), and the optimal pulse incidence angle for UAV-LiDAR to get point cloud data is obtained through the comparison between different trajectories (-30°-52°). By superimposing two non-orthogonal trajectorys, and comparing the two non-orthophoto graphic trajectorys, it is proved that the superposition of the two non-orthogonal trajectorys can reach or exceed the effect of orthographic data. In order to improve the accuracy of the estimated LAD, Maize leaf inclination and pulse incidence angle are considered, and the estimation accuracy of leaf area density is improved compared with that before correction. With the optimal trajectory and the optimal voxel size, the leaf area density distribution of different planting densities and varieties is analyzed, the growth rate, plant type characteristics and the most reasonable planting density of different varieties of Maize plant can be obtained through the analysis of the leaf area density distribution of different plant densities and different varieties.

  Conclusions  It is believed that our findings in this paper can provide guidance for estimating leaf area density using UAV-LiDAR data and provide reference for Maize breeding and scientific management.