欢迎访问林草资源研究

林业资源管理 ›› 2021›› Issue (1): 61-68.doi: 10.13466/j.cnki.lyzygl.2021.01.009

• 科学研究 • 上一篇    下一篇

结合自适应阈值与峰值的LiDAR林分平均高反演方法研究

吴思敏1,2,3(), 孙华1,2,3, 林辉1,2,3()   

  1. 1.中南林业科技大学 林业遥感信息工程研究中心,长沙 410004
    2.林业遥感大数据与生态安全湖南省重点实验室,长沙 410004
    3.南方森林资源经营与监测国家林业与草原局重点实验室,长沙 410004
  • 收稿日期:2020-11-06 修回日期:2020-12-25 出版日期:2021-02-28 发布日期:2021-03-30
  • 通讯作者: 林辉
  • 作者简介:吴思敏(1996-),男,浙江温州人,在读硕士,研究方向:林学。Email: wusimin@csuft.edu.cn
  • 基金资助:
    “十三五”国家重点研发计划项目:人工林资源监测关键技术研究(2017YFD0600900);国家自然科学基金面上项目(31971578);湖南省教育厅科学研究重点项目(17A225);湖南省普通高校青年骨干教师培养对象资助项目(90102-7070220090001)

Research on LiDAR Stand Average High Inversion Method Based on Auto-adaptive Threshold and Peak Value

WU Simin1,2,3(), SUN Hua1,2,3, LIN Hui1,2,3()   

  1. 1. Research Center of Forestry Remote Sensing & Information Engineering Central South University & Technology,Changsha 410004,Hunan,China
    2. Key Laboratory of Forestry Remote Sensing Based Big Data & Ecological Security for Hunan Province.Changsha 410004,Hunan,China
    3. Key Laboratory of State Forestry & Grassland Administration on Forest Resources Management and Monitoring in Southern Area,Changsha 410004,Hunan,China
  • Received:2020-11-06 Revised:2020-12-25 Online:2021-02-28 Published:2021-03-30
  • Contact: LIN Hui

摘要:

随着激光雷达获取的点云密度不断增加,提取样地尺度的林分平均高成为可能。但样地尺度林分平均高的提取精度与树种之间的关系尚不明确,急需一种能适应各种树种的林分平均高提取方法。以广西省国有高峰林场为例,采用机载LiDAR点云数据生成的冠层高度模型(Canopy height model,CHM),结合地面实测的201个样地数据,提出了一种结合自适应阈值与峰值的林分平均高提取算法,并分析了树种对提取精度的影响。结果表明:1)不同树种的林分平均高提取精度存在差异,杉木精度最高,而桉树和其他阔叶树种精度次之;2)自适应阈值结合峰值的算法能够较好提取林分平均高(R2=0.75,RMSE=3.11m,rRMSE=22.07%),并且对于不同的树种都有较强的稳健性;3)阔叶树种和针叶树种对不同的提取方法存在敏感性差异。研究提取的林分平均高可为森林蓄积量与生物量反演研究提供依据和参考。

关键词: LiDAR, 自适应阈值, 冠层高度, 模型, 林分平均高

Abstract:

With the increasing density of point cloud obtained by LiDAR,it is possible to extract the average height of stand at sample plot scale.However,the relation between the extraction accuracy of average height of stand at sample plot scale and tree species is still unclear,so a stand average height extraction method suitable for various tree species is urgently needed.This study tries to take the state-owned Gaofeng forest farm in Guangxi Zhuang Autonomous Region as an example and use the canopy height model(CHM) generated by airborne LiDAR point cloud data.Based on 201 sample plots data measured on the ground,the paper proposes a stand average height extraction algorithm combining auto-adaptive threshold and peak value and analyzes the influence of tree species on extraction accuracy.The results show that:1) the average high extraction accuracy of different tree species is different.The accuracy of Chinese fir is the highest,followed by Eucalyptus and other broad-leaved trees;2) The auto-adaptive threshold combined with peak value algorithm can extract the average stand height(R2=0.75,RMSE=3.11 m,rRMSE=22.07%),and has strong robustness for different tree species;3) The sensitivity of broad-leaved and coniferous species to different extraction methods is different.The average stand height extracted in this study can be used as the basis and reference for the inversion of forest volume and biomass.

Key words: LiDA, Radaptive threshold, canopy height model, average stand height

中图分类号: