Individual Tree Delineation in Windbreaks Using Airborne-Laser-Scanning Data and Unmanned Aerial Vehicle Stereo Images

This letter is aimed to compare the performance of canopy height models (CHMs) derived from airborne laser scanning (ALS) data and unmanned aerial vehicle (UAV) stereo images in the extraction of individual tree height and crown size. Treetops were identified using the local maximum algorithm from the Gaussian filtered CHMs. A parabola fitting was used to determine the crown size. Factors affecting the delineation results, such as point cloud density and the spatial distribution and growing status of trees, were analyzed. The results showed that the UAV stereo images, together with the ALS-derived digital elevation model (DEM), can achieve better performance than ALS data alone based on our data set. The match ratio between delineated and field-measured trees varied significantly, with the highest ratio of 66.94% obtained by UAV in the young aspen forest and the lowest ratio of 33.76% obtained by ALS in the old forest. Aside from the influence of point density, this letter also shed light on the important role that the spatial distribution and growing status of trees play in the delineation of individual trees. To conclude, integrating UAV stereo images with the ALS-derived DEM is effective in delineating individual tree attributes in small-scale windbreaks, which provides some suggestions for the future management of agriculture land.