【摘要】 随着人工智能的发展，无人机机载激光雷达技术和机器学习算法在森林蓄积量估测中得到了广泛的应用。选取山西省左云县云西林场的油松和新疆杨各3块样地，样地内油松和新疆杨共计318株，根据获取的无人机激光雷达点云数据进行单木分割，提取各树种的特征变量，并筛选出30个特征变量作为建模变量，采用随机森林和支持向量机算法构建单木蓄积量估测模型。结果表明：随机森林模型估测油松和新疆杨单木蓄积量的决定系数(R~2)分别达到0.941和0.934,其精度均高于支持向量机模型；在高度变量、密度变量和强度变量中，高度变量对建模具有较大的贡献；针叶树种和阔叶树种的两种机器学习模型均能获得较好的估测精度，树种的类型对模型精度影响较小。因此，随机森林模型可作为单木蓄积量精准估测的优选模型。更多还原

【Abstract】 With the development of artificial intelligence, unmanned aerial vehicle(UAV) mounted LiDAR technology and machine learning algorithms have been widely applied in forest stock volume estimation. This study selected three sample plots of Pinus tabuliformis and Populus albe in Yunxi Forest Farm, Zuoyun County, Shanxi Province. A total of 318 trees of the two species were included in the sampling. UAV LiDAR point cloud data were acquired and used to extract feature variables of each tree species after individual tree segmentation. Thirty key feature variables were selected as modeling variables, and random forest and support vector machine algorithms were employed to construct models for estimating individual tree volumes. The results showed that the random forest model achieved R~2 values of 0.941 and 0.934 for P. tabuliformis and P. albe, respectively, outperforming the support vector machine model in accuracy. Height variables contributed significantly to the modeling among height, density, and intensity variables. Both machine learning models demonstrated high estimating accuracy for coniferous and broadleaf species, with tree species type having minimal impact on model accuracy. Therefore, the random forest model is recommended as the preferred model for precise estimation of individual tree volumes.更多还原