依据背包和无人机激光雷达数据对思茅松林分结构参数估测  

Estimation of Stand Structure Parameters for Pinus kesiya var.langbianensis Based on Backpack and UAV LiDAR Data

作　　者：张澜钟 岳彩荣[1] 李初蕤 李馨 李佳[1] 沈健 宗发荣 Zhang Lanzhong;Yue Cairong;Li Churui;Li Xin;Li Jia;Shen Jian;Zong Farong(Southwest Forestry University,Kunming 650224,P.R.China;Liangshan Yi Autonomous Prefecture Academy of Forestry and Grassland Sciences;Jiangcheng County Forestry and Grassland Bureau)

机构地区：[1]西南林业大学,昆明650224 [2]凉山彝族自治州林业草原科学研究院 [3]江城县林业和草原局

出　　处：《东北林业大学学报》2024年第12期92-100,共9页Journal of Northeast Forestry University

基　　金：国家自然科学基金项目(42061072);云南省教育厅项目(2018JS330);云南省重大科技专项(202002AA100007-015)。

摘　　要：以15块样方调查数据验证背包激光雷达数据提取单木参数的精度,依据270块背包激光雷达数据单木分割后计算得到的林分结构参数与无人机机载激光雷达数据特征参数相结合,通过最优估测模型对研究区的林分结构参数进行反演与预测。结果表明:(1)背包激光雷达数据提取单木参数时,单木分割F-score(F)得分的范围在95%~100%,胸径的均方根误差(E_(RMS))在0.08~1.68 cm,树高的均方根误差在0.71~2.29 m;(2)平均树高最优估测模型决定系数(R^(2))为0.87,E_(RMS)为1.09 m,平均绝对误差(E_(MA))为0.68 m;Lorey’s树高最优估测模型R^(2)为0.88,E_(RMS)为0.79 m,E_(MA)为0.54 m;算术平均胸径最优估测模型R^(2)为0.81,E_(RMS)为1.02 cm,E_(MA)为0.79 cm;蓄积量最优估测模型R^(2)为0.82,E_(RMS)为17.06 m^(3)·hm^(-2),E_(MA)为12.05 m^(3)·hm^(-2);胸高断面积最优估测模型R^(2)为0.78,E_(RMS)为3.48 m^(2)·hm^(-2),E_(MA)为2.55 m^(2)·hm^(-2);郁闭度最优估测模型R^(2)为0.85,E_(RMS)为0.03,E_(MA)为0.02;林分密度最优估测模型R^(2)为0.91,E_(RMS)为157.60株·hm^(-2),E_(MA)为97.68株·hm^(-2)。The precision of extracting individual tree parameters from backpack LiDAR data was validated using 15 sample plots.Based on the segmentation of 270 backpack LiDAR data points,forest structural parameters were calculated and combined with feature parameters from UAV airborne LiDAR data.Through the optimal estimation model,the forest structural parameters in the study area were inverted and predicted.The results showed that:(1)When extracting individual tree parameters from backpack LiDAR data,the range of F-score for single tree segmentation was between 95.00% and 100.00%,the root mean square error(E_(RMS))for diameter breast height ranged from 0.08 to 1.68 cm,and the E_(RMS) for tree height ranged from 0.71 to 2.29 m.(2)The optimal estimation model for mean tree height had a coefficient of determination(R^(2))of 0.87,E_(RMS) of 1.09 m,and mean absolute error(E_(MA))of 0.68 m;for Lorey’s height,R^(2) was 0.88,E_(RMS) was 0.79 m,and E_(MA) was 0.54 m;for arithmetic mean diameter breast height,R^(2) was 0.81,E_(RMS) was 1.02 cm,and E_(MA) was 0.79 cm;for hectare volume,R^(2) was 0.82,E_(RMS) was 17.06 m^(3)·hm^(-2),and E_(MA) was 12.05 m^(3)·hm^(-2);for hectare basal area,R^(2) was 0.78,E_(RMS) was 3.48 m^(2)·hm^(-2),and E_(MA) was 2.55 m^(2)·hm^(-2);for canopy closure,R^(2) was 0.85,E_(RMS) was 0.03,and E_(MA) was 0.02;and for stocking density,R^(2) was 0.91,E_(RMS) was 157.60 trees·hm^(-2),and E_(MA) was 97.68 trees·hm^(-2).

关 键 词：背负式激光雷达 无人机机载激光雷达 林分结构参数 机器学习 

分 类 号：S771.8[农业科学—森林工程]