欢迎访问林业科学,今天是

林业科学 ›› 2017, Vol. 53 ›› Issue (7): 94-104.doi: 10.11707/j.1001-7488.20170710

• 论文与研究报告 • 上一篇    下一篇

森林面积多阶遥感监测方法

张煜星, 王雪军, 黄国胜, 党永峰, 陈新云   

  1. 国家林业局调查规划设计院 北京 100714
  • 收稿日期:2016-01-08 修回日期:2016-06-29 出版日期:2017-07-25 发布日期:2017-08-23
  • 通讯作者: 王雪军
  • 基金资助:
    科技部对发展中国家科技援助项目(KY201402011);三峡后续工作科研项目"三峡库区生态屏障区生态效益监测技术与评价方法研究"(2016K-1)。

Forest Area Remote Sensing Monitoring Using the Multi-Level Sampling Interpretation Approach

Zhang Yuxing, Wang Xuejun, Huang Guosheng, Dang Yongfeng, Chen Xinyun   

  1. Academy of Forest Inventory and Planning, State Forestry Administration Beijing 100174
  • Received:2016-01-08 Revised:2016-06-29 Online:2017-07-25 Published:2017-08-23

摘要: [目的]从林业数据采集的迫切需求出发,采用遥感技术实现森林资源数据的快速年度出数,以及时掌握区域森林资源变化信息。[方法]建立遥感3阶抽样与地面调查技术相结合的森林资源遥感监测技术体系,并以辽宁省为例,利用2013-2014年高、中、低多源遥感数据,进行基于多阶抽样的省级森林面积监测方法实践。[结果]遥感结合地面样地实测,建立了地面样地与高分样地,高、中分样地以及中、低分样地的森林面积回归模型,其模型确定系数(R2)分别为0.99、0.91和0.70,模型精度较高;基于3阶分层抽样的遥感监测方法得到辽宁省2014年森林面积预测值为590.83×104hm2,森林覆盖率为40.54%,略高于清查结果(40.49%);从抽样精度来看,采用分层抽样方法得到的森林面积最为可靠,抽样精度达99%以上;通过精度分层控制,基于MODIS NDVI阈值法制作了辽宁省森林分布图,实现了辽宁省森林面积和空间分布的年度产出。[结论]基于遥感抽样与地面调查技术相结合的森林面积多阶遥感监测方法可行,既可提高监测成果的时效性,又能缩短成果产出周期,具有很好的应用前景,可为下一步建立全国遥感监测体系提供重要的技术支持。

关键词: 多阶抽样, 遥感监测, 判读区划, 森林面积, 年度监测

Abstract: [Objective] It is the urgent requirement for forestry that forest resources data and changing information are quickly and effectively acquired by using remote sensing.[Method] This study built forest resource monitoring approach by combining three levels sampling of remote sensing imagery and field surveying. We chose Liaoning Province as study area to assess our method by using multi-source remote sensing data with high, medium and low spatial resolution from 2013 to 2014.[Result] Firstly, we obtained the forest area estimating model by combining separately filed measured plots and high, middle and low resolution remote sensing data. The performance of these three forest area regress model were evaluated and showed high precision(R2=0.99, 0.91, 0.70,respectively). Secondly, we estimated the forest area (590.83 million hectares) and forest coverage(40.54%) of Liaoning Province,respectively. The value of forest coverage is similar to that determined by forest resources inventory(40.49%), which proved that the proposed method in this study could obtain forest area with high precision(more than 99%). In addition, the forest distribution map of Liaoning Province was produced based on MODIS NDVI threshold value method. This study proposed a forest resource monitoring approach which can obtain annual forest area and distribution map of Liaoning Province.[Conclusion] This research explored the method of remote sensing monitoring indices as an available technique. It would be helpful to monitoring forest resource because of the shortening forest monitoring result output cycle. The forest resource monitoring approach proposed in this paper should have a bright future of application and dissemination, and would provide important support for building remote sensing monitoring system of national forest resource.

Key words: multi level sampling, remote sensing monitoring, interpretation division, forest area, annual monitoring

中图分类号: