基于激光雷达数据的热带森林冠高模型生成及平均树高估计

海南省林业科学研究所,海南海口 571100

冠高模型;点云数据; LiDAR;树木平均高度

Crown height model generation and average height estimation from LiDAR data
CHEN Zongzhu, YANG Qi, LEI Jinrui, CHEN Xiaohua, LI Yuanling

Hainan Provincial Forestry Science Research Institute, Haikou 571100, Hainan, China

crown height model; point cloud data; LiDAR; average height of trees

DOI: 10.14067/j.cnki.1673-923x.2018.07.001

备注

激光雷达 (LiDAR)是一种适用于大范围快速获取森林三维数据的新手段,但是基于 LiDAR的林业参数估计算法和提取精度仍有待进一步提高。以海南省三亚市热带区域森林为研究对象,基于其扫描获取的 LiDAR点云数据提出了改进的动态移动窗口搜索算法,进行冠高模型建模与生成,并提出了基于此模型进行林木平均高度估计的方法。通过改进的算法,利用 C#编程、 LasTools、ArcGIS等相关工具对点云数据进行网格化,从而进行 DEM和 DSM提取、 CHM生成、平均树高等参数估计。结果 表明:经过算法的改进,大面积森林的数字地面模型 (DEM)、数字表面模型 (DSM)和冠高模型 (CHM)可以从 LiDAR数据中快速提取,但 LiDAR点云获得的树高值比实际值低。这是由于 LiDAR数据树顶的错失及点云的低密度所导致的,然而通过最低树高值的不断增加,其实测值与估计值的差值也在逐渐缩小。

LiDAR is a new method suitable for large-scale and rapid acquisition of forest three-dimensional data, but the estimation algorithm and extraction precision of forest parameters based on LiDAR still need to be improved. This paper presents an improved dynamic mobile window search algorithm based on the LIDAR point cloud data obtained from the Hainan Province Sanya Tropical area forest, for modeling and generation of crown height model, and proposes a method for average height of trees estimation based on this model. Through the improved algorithm, used C # programming, Lastools, ArcGIS and other related tools to grid the point cloud data, thus, DEM and DSM extraction, CHM generation and Average height of trees parameter estimation are carried out. The results show that the Digital Terrain model (DEM), Digital Surface model (DSM) and Crown Height model (CHM) of large area forest can be quickly extracted from LiDAR data, but the tree high value obtained by LiDAR point Cloud is lower than the actual value. Due to the LIDAR data of the top of tree and the low density of the point cloud, however, the difference between the measured value and the estimated value is gradually reduced by increasing the lowest tree height value.

·