欢迎访问林草资源研究

林业资源管理 ›› 2023›› Issue (2): 96-103.doi: 10.13466/j.cnki.lyzygl.2023.02.013

• 技术应用 • 上一篇    下一篇

天然林保护工程区森林植被类型遥感监测

王晓慧1,2(), 张会儒1,2(), 庞勇1,2, 覃先林1,2, 李海奎1,2, 蒙诗栎1,2, 余涛1,2   

  1. 1.中国林业科学研究院资源信息研究所,北京 100091
    2.国家林业和草原局 林业遥感与信息技术重点实验室,北京 100091
  • 收稿日期:2023-01-10 修回日期:2023-04-21 出版日期:2023-04-28 发布日期:2023-06-26
  • 通讯作者: 张会儒(1964-),男,甘肃合水人,博士,研究员,主要从事森林可持续经营研究工作。Email:huiru@ifrit.ac.cn
  • 作者简介:王晓慧(1974-),女,山西忻州人,博士,副研究员,主要从事森林资源遥感监测研究工作。Email:wangxh@ifrit.ac.cn
  • 基金资助:
    中央级公益性科研院所基本科研业务费专项资金(CAFYBB2020ZD002);国家重点研发计划(2022YFF0711602)

Forest Vegetation Type Monitoring in the Natural Forest Protection Project Area

WANG Xiaohui1,2(), ZHANG Huiru1,2(), PANG Yong1,2, QIN Xianlin1,2, LI Haikui1,2, MENG Shili1,2, YU Tao1,2   

  1. 1. Institute of Forest Resource Information Techniques,Chinese Academy of Forestry,Beijing 100091,China
    2. Key Laboratory of Forestry Remote Sensing and Information System,National Forestry and Grassland Administration,Beijing 100091,China
  • Received:2023-01-10 Revised:2023-04-21 Online:2023-04-28 Published:2023-06-26

摘要:

应用中等分辨率遥感影像开展森林植被类型遥感监测,为天然林保护工程区成效监测以及森林植被制图提供技术支撑。以汪清林业局为例,基于Landsat 8 OLI的生长季和非生长季影像,应用随机森林分类法,以及基于时间维的分类结果校正,开展森林植被类型信息提取研究,根据混淆矩阵和查全率分析引起森林植被类型混分的因素。结果表明:1)研究区森林植被类型遥感监测的总体精度为86.41%,Kappa系数为0.82,体现了较好的分类效果。2)在各森林植被类型中,落叶阔叶林地分类精度高,生产者精度和用户精度在90%以上;落叶针叶林地生产者精度较高,为86.96%;常绿针叶林地和针阔混交林地生产者精度和用户精度较低,平均为75.19%。常绿针叶林地、落叶针叶林地、落叶阔叶林地与针阔混交林地分类混分的影响因素为混交比例、郁闭度和林龄。3)汪清林业局森林覆盖率为96.64%,落叶阔叶林地所占面积比例最大,针阔混交林地次之,落叶针叶林地、常绿针叶林地、灌木林地和其他林地分布比较少。通过分析可知:基于中等分辨率影像的多时相和物候信息,有利于提取天然林保护工程区森林植被类型。

关键词: 天然林保护工程区, 森林植被类型, 多时相和物候信息, 随机森林, 时间维校正, 汪清林业局

Abstract:

Based on moderate resolution remote sensing images,forest vegetation type monitoring was performed to provide technical support for achievement monitoring of the natural forest protection project area and forest vegetation mapping.Based on Landsat8 OLI images in growing and non-growing seasons,random forest and time dimension correction methods were applied to forest vegetation type monitoring in Wangqing Forestry Bureau.Based on confusion matrix and recall ratio,factors leading to classification confusion were analyzed.The results showed that:1) Overall accuracy of forest vegetation type monitoring was 86.41%,and kappa coefficient was 0.82,illustrating a better classification effect.2) Among various forest vegetation types,deciduous broadleaved forest land had high classification accuracy,with producer's and user's accuracy of over 90%,respectively.Deciduous coniferous forest land had relatively high producer's accuracy of 86.96%.Evergreen coniferous forest land and coniferous and broadleaved mixed forest land had relatively low producer's and user's accuracies of average 75.19%.Classification confusion between evergreen coniferous forest,deciduous coniferous forest,deciduous broadleaved forest,and coniferous and broadleaved mixed forest frequently occurred as the result of mixed proportion,forest canopy closure and forest age.3) Forest coverage of Wangqing Forestry Bureau was 96.64%.The area proportion of deciduous broadleaved forest land was the largest,that of coniferous and broadleaved mixed forest land was the secondly largest,and that of deciduous coniferous forest land,evergreen coniferous forest land,shrubland and other forest land was small.The analysis showed that multitemporal and phenological information of moderate resolution remote sensing images was effective in obtaining forest vegetation types of the natural forest protection project area.

Key words: natural forest protection project area, forest vegetation type, multitemporal and phenological information, random forest, time dimension correction, Wangqing Forestry Bureau

中图分类号: