摘要:森林植被高度与树木分布格局是植物群落重要结构特征，也是计算森林生物量分布的重要参数。传统的森林群落调查方法耗费大量人力物力难以进行较大尺度的群落结构测量，而一般的遥感影像也难以获得精确的地形信息及垂直结构。近年来激光雷达（Light Detection and Ranging，LiDAR）技术快速发展，能够较好的进行植被三维特征的提取并被广泛应用于森林生态系统检测模拟。且随着无人机低空摄影技术的发展催生的无人机激光雷达（UAV-Lidar）更增加了激光雷达的灵活性以及获取较大范围植被冠层信息的能力。而受限于激光的穿透性以及不同植被类型郁闭度的影响，该技术的应用多局限于在针叶林群落的垂直结构研究，而在常绿阔叶林的研究中应用较少。为探究现有无人机激光雷达设备及垂直结构提取分析技术应用于常绿阔叶林的可行性，利用无人机载激光雷达遥感技术对哀牢山中山湿性常绿阔叶林3块面积1hm²的样地进行基于数字表面模型以及数字地表高程模型做差得到树冠高度模型测量的植被冠层高度、基于局部最大值法进行单木位置提取并使用Clark-Evans最近邻体分析方法进行样地内高大乔木分布格局的计算。分析结果显示，植被高度提取精度平均大于95%，与地表实测的植被高度值拟合度较高，相关系数R²介于0.833-0.927之间；3个样地冠层高度平均值分别为18.79、19.08、17.03 m，标准差分别为8.10、7.34、7.17 m。单木探测百分比平均86.3%，用户精度以及生产者精度平均分别为75.69%和65.15%。实测得出三个样地全部高大乔木空间分布格局均为聚集分布，而激光雷达测量结果显示为随机分布或均匀分布。实验显示基于无人机激光雷达技术能够很好地提取植被冠层高度信息并能够较好地获取树木位置，但对于树木空间分布格局判定的准确性有待于进一步探索。未来研究应从多角度对激光雷达测量造成的误差原因予以分析（如环境因素），并进一步研究更为精确的单木提取以及植被高度提取方法，为通过无人机激光雷达测算森林生物量及各种生态过程提供更加精准的指标数据。



Abstract:Forest canopy height and trees distribution pattern are important indicators for studies of characterizing plant community vertical structure, which are also important parameters for calculating forest biomass distribution pattern. The traditional forest community survey methods expend a lot of manpower and resources and are also difficult to measure large-scale community structures easily. It is also difficult to obtain accurate topographic information and vertical structures using general traditional remote sensing images. In recent years, the rapid development of Light Detection and Ranging (LiDAR) technology makes it much easier to extract three-dimensional features of vegetation and is widely used in forest ecosystem detection and simulation. UAV-LiDAR has become a more flexible LiDAR technology which had ability to obtain wide range of vegetation canopy information with the development of low-altitude unmanned aerial photogrammetry and remote sensing technology. Due to the penetrability of the laser and the influence of canopy density on different vegetation types, the application of this technology is usually limited to the studies of measuring vertical structure of the coniferous forest community while is less used in the studies of evergreen broad-leaved forest. In order to explore the feasibility of applying the existing UAV-LiDAR equipment and vertical structure extraction analysis technology to evergreen broad-leaved forest vertical structure study, we used UAV-lidar to calculate canopy height based on Canopy Height Model (CHM) which is obtained from the value of Digital Surface Model (DSM) minus Digital Terrain Model (DTM). We also extracted trees location based on the local maximum value to calculate trees distribution pattern by Clark-Evans nearest neighbor analysis of large tree in three 1 hm² plots of Ailao Mountain evergreen broad-leaved foresy. The results of accuracy analysis show that the accuracy of canopy height's measurement is above 95%. There are very significantly correlations between the vegetation height measured by LiDAR and field measurement. The maximum R² is 0.927 while the minimum is 0.833. The mean values of canopy height of the three samples are 18.79, 19.08, and 17.03 m while the standard deviations are 8.10, 7.34 and 7.17 m, respectively. The average detection percentage of individual tree detection in three samples is 86.3%. The means of average user's accuracy and the producer's precision are 75.69% and 65.15% respectively. The spatial distribution patterns of the whole tall trees in three plots were regarded as aggregated distribution by the result of field measurement, while the result of measurements by LiDAR showed random or uniform distribution. The experiment shows that it is efficient for UAV-LiDAR technology to extract the vegetation canopy height information accurately and obtain the location of trees. However, the accuracy of the tree spatial distribution pattern determining needs to be further improved. Research in future should focus on analyzing of the causes of error caused by LiDAR data from various angles (like environmental factor) and develop more accurate single-wood extraction approaches and vegetation height extraction methods to provide more accurate indicator data for measuring forest biomass and more biological processes by UAV-LiDAR technology.





PDF全文下载地址:

https://www.ecologica.cn/stxb/article/pdf/stxb201809282112