基于UAV-LiDAR的人工林林分郁闭度估测

Estimation on Canopy Closure for Plantation Forests Based on UAV-LiDAR

【目的】提取无人机激光雷达(UAV-LiDAR)点云数据特征变量,结合样地实测数据构建林分郁闭度估测模型,为快速准确估测人工林林分郁闭度提供基础数据和技术参考。【方法】以东北林业大学城市林业示范基地为研究区,基于多旋翼无人机激光雷达获取的点云数据进行人工针叶林和阔叶林林分郁闭度反演,根据点云的三维坐标和能量值计算高度、强度、冠层特征变量并采用主成分分析法降维,处理后的变量与利用植物冠层分析仪获取的郁闭度进行逐步回归分析,建立人工针叶林和阔叶林林分郁闭度估测模型,在ArcGIS平台上应用估测模型和反距离权重插值法进行林分郁闭度反演制图。【结果】冠层特征变量对针叶林郁闭度的估测精度影响最显著,强度特征变量对阔叶林郁闭度的估测精度影响最显著。人工阔叶林郁闭度的估测精度(Adj R2=0.725,RMSE=0.005)优于人工针叶林(Adj R2=0.722,RMSE=0.007)。应用估测模型和反距离权重插值法估测整个样地的郁闭度范围在0.81~0.87之间,筛选10个检验点的郁闭度与实测郁闭度显示出较高相关性(r=0.859)。【结论】结合多组LiDAR特征变量估测林分郁闭度能够充分挖掘LiDAR数据包含的冠层结构特性,提升估测精度。

【Objective】This study aims to construct stand canopy closure estimation models based on the extracted feature variables from the UAV-LiDAR point cloud data and the measured data of the sample plots,which would provide basic data and technical reference for rapid and accurate estimation of artificial forest canopy closure.【Method】Taking the Urban Forestry Demonstration Base of Northeast Forestry University as the study area,the inversion of the forest closure for the artificial coniferous forests and broad-leaved forests was carried out based on the point cloud data obtained by multi-rotor UAV LiDAR.The height,intensity and canopy characteristic variables were calculated according to the three-dimensional coordinates and energy values of the point cloud,and the principal component analysis method was used to reduce the data dimension.Then,a stepwise regression procedure was conducted on the canopy closure obtained by the plant canopy analyzer and the processed variables,and the canopy closure estimation models for artificial coniferous forests and broad-leaved forests were established,respectively.Finally,the estimation models and inverse distance weight interpolation method were applied on the ArcGIS platform to perfom canopy closure inversion mapping.【Result】Canopy characteristic variables had the most significant effect on the canopy closure estimation accuracy of coniferous forests,while intensity characteristic variables had the most significant effect on the estimation accuracy of broad-leaved forests.The canopy closure estimation accuracy of artificial broad-leaved forests(Adj R2=0.725,RMSE=0.005)was superior to that of artificial coniferous forests(Adj R2=0.722,RMSE=0.007).The canopy closure of the entire plot estimated by the estimation models and inverse distance weighted interpolation method ranged between 0.81 and 0.87,and the canopy closure of 10 testing points showed a high correlation with the measured canopy closure(r=0.859).【Conclusion】Combining multiple sets of LiDAR feature variables to estimate forest canopy closure can fully exploit the canopy structure characteristics contained in LiDAR data and improve the estimation accuracy.