首页 >  , Vol. , Issue () : -

摘要

全文摘要次数: 167 全文下载次数: 138
引用本文:

DOI:

10.11834/jrs.20210246

收稿日期:

2020-07-05

修改日期:

2021-03-17

PDF Free   EndNote   BibTeX
无人机激光雷达人工林林分平均高估测模型分析
李梅1, 刘清旺1, 冯益明2, 李增元1
1.中国林业科学研究院资源信息研究所;2.中国林业科学研究院荒漠化研究所
摘要:

我国人工林面积居世界第一,精确地对人工林结构进行监测具有重要意义。本研究以内蒙古自治区赤峰市旺业甸林场内的落叶松和油松人工林为研究对象,利用无人机激光雷达(Light Detection And Ranging, LiDAR)点云数据和样地地面调查数据来分析人工林林分平均高估测模型,通过点云特征变量与样地实测的6种林分平均高(包括:Lorey`s高、算术平均高、最大高、优势树高、中位数高和冠幅面积加权高)间的Pearson`s相关性实现变量筛选,然后利用全子集回归和多元线性回归方法构建不同林分平均高的预测模型,并采用交叉验证进行精度验证。结果表明:激光雷达点云高度百分位数与林分平均高具有较强相关性,通过一元线性回归模型可以得到高精度的林分平均高, Lorey`s高(R2为0.91 ~ 0.97,RMSE为0.47~0.48m)、优势树高(R2为0.86 ~ 0.97,RMSE为0.53~0.72m)和冠幅面积加权高(R2为0.86 ~ 0.96,RMSE为0.57~0.66m)估测精度最高,算术平均高(R2为0.85 ~ 0.94,RMSE为0.68~0.86m)和中位数高(R2为0.80 ~ 0.95,RMSE为0.64~1.07m)次之,最大高(R2为0.69 ~ 0.87,RMSE为1.30~1.40 m)最低。针对不同森林类型,区分和未区分森林类型的不同林分平均高模型的估测精度均较高,且预测精度没有明显差异(未区分R2 为0.87 ~ 0.97,RMSE为0.49~1.40m,区分R2为0.69 ~ 0.97,RMSE为0.47~1.30m)。无人机激光雷达可以用于估测北方温带针叶林的林分平均高,能够满足人工林资源调查快速、精确的要求。

Analysis of estimation models of plantation stand heights using UAV LiDAR
Abstract:

China’s plantation area is the largest in the world. It is very important to precisely monitor plantation structure. The study area is located at Wangyedian forest farm, Chifeng, Inner Mongolia. The dominated tree include Larixprincipis-rupprechtii and Pinustabuliformis. UAV (Unmanned Aerial Vehicle) LiDAR (Light Detection And Ranging) data and in situ sample plots measurements were used to analyze the estimation models of plantation stand heights in this study. The Pearson correlations between the six stand heights (Arithmetic mean height, Lorey`s height, Dominated height, Maximum height, Median height and Crown area weighted height) and LiDAR canopy metrics were applied to identify the significant independent variables. Then, a branch-and-bound search for the best subset and multiple linear regression were conducted to fit stand heights. Finally, the accuracy of the predictions was validated using cross-validation. The results showed that the height metrics had strong relationship with the six stand heights. The simple linear regression with one independent variable was the optimal model to estimate stand heights with high accuracy. The Lorey`s height (R2 was 0.91 ~ 0.97, RMSE was 0.47~0.48m), dominated height (R2 was 0.86 ~ 0.97, RMSE was 0.53~0.72m) and Crown area weighted height (R2 was 0.86 ~ 0.96, RMSE was 0.57~0.66m) had the highest accuracy, while arithmetic mean height (R2 was 0.85 ~ 0.94, RMSE was 0.68~0.86m) and median height (R2 was 0.80 ~ 0.95, RMSE 0.64~1.07m) had a lower accuracy, maximum height (R2 was 0.69 ~ 0.87, RMSE was 1.30~1.40 m) was lower. There were no significant differences between the overall estimations and the stratified estimations of plantation stand heights considering different forest type. As for the all plots, R2 was 0.85 ~ 0.97, RMSE was 0.49~1.40m. While, as for the pure plots, R2 was 0.69 ~ 0.97, RMSE was 0.47~1.30m. UAV LiDAR has the ability to estimate stand heights of northern temperate coniferous forest, and could meet the requirements of rapid and accurate survey of plantation resources.

本文暂时没有被引用!

欢迎关注学报微信

遥感学报交流群 分享按钮