Seasonal Impacts on Individual Tree Detection and Height Extraction Using UAV-LiDAR: Preliminary Study of Planted Deciduous Stand

Light Detection and Ranging (LiDAR) has proved to be an effective technology for accurately extracting forest structural parameters. Unmanned Aerial Vehicles (UAVs) are characterized by its flexibility and low cost. Combining the advantages of both technologies, UAV-LiDAR exhibits great potential in the accurate surveying of large forests. However, for forests dominated by deciduous tree species, the accuracy of individual tree detection and height extraction is inevitably impacted by the leaf-on and leaf-off seasons when UAV-LiDAR scans point clouds. In this study, a planted forest of dawn redwood (Metasequoia glyptostroboides Hu & W. C. Cheng) in Ma’anxi Wetland Park of Chongqing, China, was chosen as the study object. The UAV-LiDAR was first leveraged to capture the point clouds of summer and winter seasons. Then, the canopy height models (CHMs) with different spatial resolutions were generated, based on which the tree quantity and individual heights were extracted. The achieved outcomes included the following: (1) The CHMs of the two seasons could be used to obtain the tree quantity, and the accuracy of individual tree detection from the point cloud scanned in the winter was relatively higher than that in the summer. (2) The spatial resolution of CHM impacted the accuracy of individual tree segmentation and height extraction, and the optimum spatial resolution was 0.3 m (approximately 1/10 of the average canopy diameter of the dawn redwoods). Therefore, to obtain more accurate individual tree heights of the deciduous forest, it is better to scan the point cloud using UAV-LiDAR in the leaf-off season and choose the appropriate spatial resolution of the CHM.