欢迎访问林业科学,今天是

林业科学 ›› 2020, Vol. 56 ›› Issue (1): 74-86.doi: 10.11707/j.1001-7488.20200108

• 论文与研究报告 • 上一篇    下一篇

基于无人机激光雷达的银杏人工林有效叶面积指数估测

吴项乾,曹林*,申鑫,汪贵斌,曹福亮   

  1. 南京林业大学 南方现代林业协同创新中心 南京 210037
  • 收稿日期:2018-01-10 出版日期:2020-01-25 发布日期:2020-02-24
  • 通讯作者: 曹林
  • 基金资助:
    国家重点研发计划(2017YFD0600904);国家自然科学基金项目(31770590);江苏省高校优势学科建设工程资助项目(PAPD)

Estimation of Effective Leaf Area Index Using UAV-Based LiDAR in Ginkgo Plantations

Xiangqian Wu,Lin Cao*,Xin Shen,Guibin Wang,Fuliang Cao   

  1. Co-Innovation Center for the Sustainable Forestry in Southern China Nanjing Forestry University Nangjing 210037
  • Received:2018-01-10 Online:2020-01-25 Published:2020-02-24
  • Contact: Lin Cao
  • Supported by:
    国家重点研发计划(2017YFD0600904);国家自然科学基金项目(31770590);江苏省高校优势学科建设工程资助项目(PAPD)

摘要:

目的: 精确估测银杏人工林有效叶面积指数(eLAI),以更好了解银杏人工林的生长和竞争、理解人工林生态系统的功能和生产力。方法: 基于多旋翼无人机激光雷达(LiDAR)系统获取的点云数据,结合45块地面实测样地数据,使用孔隙度模型法(通过计算点云的冠层穿透率,根据Beer-Lambert定律计算有效叶面积指数)和统计模型法(首先通过地面实测的有效叶面积指数和所提取的LiDAR特征变量建模,然后借助拟合的模型估测有效叶面积指数)对我国典型银杏人工林进行样地尺度的有效叶面积指数估测。结果: 1)使用统计模型法估测eLAI时,仅利用LiDAR高度特征变量估测精度为R2=0.38(rRMSE=54%),引入其他特征变量(冠层密度特征、冠层容积比以及强度特征变量)后精度分别达到R2=0.64(rRMSE=26%)、R2=0.61(rRMSE=28%)、R2=0.74(rRMSE=23%);2)根据Cover将样地分组建模后发现,分组建模的精度优于不分组建模的精度;3)孔隙度模型法估测有效叶面积指数的精度为R2=0.71(rRMSE=32.0%)。结论: 结合多组LiDAR特征变量估测有效叶面积指数能够充分挖掘LiDAR数据包含的冠层结构特性,从而提升估测精度;同时,使用孔隙度模型法可以有效估测银杏人工林有效叶面积指数。无人机LiDAR点云在估测银杏人工林有效叶面积指数上具有较好的潜力。

关键词: 无人机, 激光雷达, 银杏, 人工林, 有效叶面积指数

Abstract:

Objective: Ginkgo biloba is one of the most important tree species in China. Real-time, quantitative and accurate estimation of its eLAI(effective leaf area index) plays a key role in analyzing its growth and competition as well as understanding the functions and productivity of the plantation ecosystems. Method: Combined with the point cloud data acquired from the multi-rotor UAV-based LiDAR system and 45 sample plots data, this paper used the following two approaches:gap-fraction modeling(calculating the canopy gap fraction of the point cloud, then computing eLAI according to the Beer-Lambert law)and statistical modeling(modeling based on the eLAI measured from ground and the metrics derived from LiDAR first return, then the fitted models were applied to calculate the eLAI)method, to estimate the eLAI in a typical ginkgo plantation of China. Result: 1) When using the statistical model to estimate the eLAI, the accuracy of estimation was R2=0.38(rRMSE=54%) by modeling of LiDAR height metrics only. By gradually introducing multiple sets of metrics (density metrics, canopy volume metrics and intensity metrics), the accuracy of the estimation was increased to R2=0.64(rRMSE=26%), R2=0.61(rRMSE=28%)and R2=0.74(rRMSE=23%), respectively. 2) The sample plots were grouped according to Cover, the accuracy of group modeling was better than that of non-group modeling. 3) When using the gap-fraction model to estimate the eLAI, the accuracy was R2=0.71(rRMSE=32.0%). Conclusion: Combining multiple sets of LiDAR metrics to estimate the eLAI could fully excavate the canopy structure characteristics contained in LiDAR data, so as to improve the estimation accuracy. Meanwhile, the gap-fraction model approach could effectively estimate the eLAI of ginkgo plantations. This study illustrated that the metrics obtained from the UAV LiDAR might have certain potentials in estimating the eLAI of the ginkgo plantations.

Key words: UAV, LiDAR, ginkgo, plantation, effective, leaf area index

中图分类号: