LiDAR采样形状及采样尺度对林分均高估测的影响  

Effects of LiDAR sampling shape and scale on stand mean height estimation

作　　者：彭涛[1] 邢艳秋[1] 尤号田[1] 丁建华[1] 戚大伟[2] PENG Tao;XING Yanqiu;YOU Haotian;DING Jianhua;QI Dawei(Center for Forest Operations and Environment;College of Science, Northeast Forestry University, Harbin 150040, Heilongjiang, China)

机构地区：[1]东北林业大学森林作业与环境研究中心,黑龙江哈尔滨150040 [2]东北林业大学理学院,黑龙江哈尔滨150040

出　　处：《中南林业科技大学学报》2018年第4期27-32,共6页Journal of Central South University of Forestry & Technology

基　　金：林业公益性行业科研专项(201504319)

摘　　要：采样形状及采样尺度的合理选择对于节约人工成本、提高计算效率以及进一步提高森林结构参数估测精度均具有重要意义。以长春净月潭国家森林公园激光雷达(LiDAR)数据为基础,通过对LiDAR数据进行圆形及方形采样,并在方形采样的基础上对LiDAR数据进行不同空间尺度采样,从中提取一系列点云分位数高用于估测樟子松及落叶松的林分均高,以此量化LiDAR采样形状及采样尺度对不同林分均高估测的影响。结果表明:对樟子松而言,无论是圆形还是方形采样,林分均高估测结果最优时所对应的参数均为HP55,且方形采样林分均高估测结果(R^2=0.896,R_(mse)=0.853 m)高于圆形采样估测结果(R^2=0.892,R_(mse)=0.868 m);对落叶松而言,圆形和方形采样均是在HP99时林分均高估测结果最优,且圆形采样林分均高估测结果(R^2=0.741,R_(mse)=1.161 m)高于方形采样估测结果(R^2=0.705,R_(mse)=1.238 m)。在方行采样条件下,樟子松林分均高估测精度最高(R^2=0.904,R_(mse)=0.820 m)时所对应的采样尺度为35 m;落叶松估测精度最高(R^2=0.720,R_(mse)=1.206 m)时所对应的采样尺度为15 m。结果表明:不同LiDAR数据采样形状对不同林分类型均高估测的影响不同,且不同林分类型均高估测结果精度达到最高时所对应的采样尺度不同。The reasonable selection of sampling shape and scale is of great significance for saving the labor cost, improving the calculation efficiency and further improving the accuracy of forest structure parameter estimation. In this paper, the LiDAR data from Jingyuetan National Forest Park located in Changchun were used and sampled with circular and square samples, separately. Additionally, the LiDAR data were resampled with different sampling scale based on square sampling. Then a series of quantification height metrics were extracted and used to estimate the stand mean height of Scotch pine and Larch pine. Finally, the effects of LiDAR sampling shape and scale on different stand mean height estimation were quantified. For Scotch pine, the results of stand mean height estimation achieved the best when the metric was HP55 both for circular and square LiDAR sampling. And the best result（R-2=0.896,Rmse=0.853 m） from square LiDAR sampling was better than the best result（R-2=0.892, Rmse=0.868 m） from circular LiDAR sampling. For Larch pine, the results of stand mean height estimation achieved the best when the metric was HP99 both for circular and square LiDAR sampling. And the best result（R-2=0.741, Rmse=1.161 m） from circular LiDAR sampling was better than the best result（R-2=0.705, Rmse=1.238 m） from square LiDAR sampling. For square LiDAR sampling scale, the accuracy of Scotch pine mean height estimation achieved the highest（R-2=0.904, Rmse=0.820 m） when the sampling scale was 35 m. For Larch pine, the result achieved the best（R-2= 0.720, Rmse=1.206 m） when the sampling scale was 15 m. The results demonstrate that the effects of LiDAR sampling shape on stand mean height estimation are related to forest stand types. And the sampling scale is different when the estimation accuracy of forest different stand types achieves the highest.

关 键 词：圆形采样 方形采样 不同空间采样尺度 樟子松 落叶松 

分 类 号：S771.8[农业科学—森林工程]