关键词: 激光雷达, 单木分割, 古树, 修正系数

Abstract:

To explore the feasibility of applying remote sensing survey methods in the 3rd National Census of Ancient and Famous Trees,this study used field survey data from one ancient-tree-group subcompartment and unmanned aerial vehicle(UAV)-borne LiDAR individual-tree segmentation data in the Northeast Tiger and Leopard National Park.The analogy method and large-tree-based approach were estimated to be the main characteristics of ancient tree groups,including the total number of ancient trees,the mean diameter at breast height(DBH),mean age,mean height,and mean crown width.Three schemes for selecting ‘large trees’ were designed,and for each scheme,seven sets of ‘large trees’ numbers were set,thus forming 21 combinations for comparison and analysis.The results showed:1)Based on the individual tree segmentation data of LiDAR,the three core factors(the number of ancient trees,mean age,and mean DBH)were estimated directly using the analogy method,with relative errors being within 10%;2)Based on overall consideration of tree height and crown width,‘large trees’ were first selected at a quantity 4 times the number of ancient trees observed in the field,subsequent analogy-based estimation reduced the relative errors of the three core factors to within±3%.These findings demonstrate that surveying ancient tree groups is feasible based on LiDAR from UAV.Selecting the appropriate number of "large trees" firstly based on individual tree segmentation data and then using the analogy method for estimation would minimize the errors to the lowest level.

Key words: LiDAR, tree segmentation, ancient tree, coefficient of correction