利用高分辨率森林覆盖影像实现高山林线的自动提取  

Automatic alpine treeline extraction using high-resolution forest cover imagery

作　　者：江鑫 何心悦 王大山 邹俊宇 曾振中 JIANG Xin;HE Xinyue;WANG Dashan;ZOU Junyu;ZENG Zhenzhong(School of Environmental Science and Engineering,South University of Science and Technology,Shenzhen 518055,China;School of Earth and Environment,University of Leeds,Leeds,LS29JT,UK)

机构地区：[1]南方科技大学环境科学与工程学院,深圳518055 [2]利兹大学地球与环境学院,英国利兹LS29JT

出　　处：《遥感学报》2022年第3期456-467,共12页NATIONAL REMOTE SENSING BULLETIN

基　　金：国家自然科学基金(编号:42071022)。

摘　　要：快速准确地提取高山林线对标定全球气候变化、科学管理森林资源具有重要意义。秦岭高山林线位于高海拔的生态交错带,垂直带谱分布明显,为本文提供了理想的试验区域。遥感技术具有重访周期短、观测范围大和不受地理环境等条件限制的特点,克服了传统实地调查方法效率低、成本高等缺点。本文基于全球首套30 m空间分辨率森林覆盖数据,结合数字高程模型以及秦岭山脉分布数据,提出一种基于遥感的自动提取高山林线的算法;结合高分辨率的Google Earth影像和GPS地面站点观测数据以及NDVI数据,验证高山林线提取结果的准确性;基于高程数据系统分析研究区高山林线分布与地形特征的关系。结果表明:(1)本文结果与Google Earth影像中实际林线分布基本一致,进一步说明本文提出的林线搜索算法的优越性;(2)秦岭林线高程分布特征呈现显著的坡向差异,呈现南坡林线高于北坡,东坡林线高于西坡的特征。鉴于遥感技术的大范围对地观测的能力,以及卫星影像数据较高的数据质量和易获取性,本文提出的分块迭代搜索策略适用于寻找不同级别的高山林线分布,可进一步推广至全球高山林线制图研究,以期为全球山地生态系统监测、保护和恢复提供技术支撑。Alpine treeline is not only an important source in calibrating global climate change but also a fundamental element in scientifically managing forest resources.Furthermore,the location,area size,and change patterns of forest lines are also used as essential information in monitoring and modeling the environment.The alpine forest line of the Qinling Mountains is located in the ecological staggered zone at high altitude,with an obvious distribution of altitudinal spectrum,which is an important north-south geographical dividing line in China.Therefore,a novel approach for the rapid and accurate identification of alpine treeline in the Qinling Mountains must be developed.We propose a remote-sensor-based algorithm for extracting alpine treelines in the Qinling Mountains by combining the high-resolution global forest cover data in 2000 with a digital elevation model and mountain distribution data.Specifically,tree cover is first extracted from the forest cover data.Next,the highest point of the study area is determined from the elevation data.Finally,the 8-connected domain search algorithm is employed to find the boundary between forest and non-forest covers to determine the alpine treeline.The algorithm is validated by high-resolution Google Earth images,GPS ground-based data,and the NDVI dataset.Further,we systematically investigate the relationship between the alpine treeline distribution and geographical factors(elevation,slope,and aspect)in the study area using the elevation data.The distribution of treelines in this paper is consistent with the actual treelines distribution in Google Earth images,further demonstrating the performance of the proposed algorithm.The elevation of treelines in the Qinling Mountains ranges from 2400 m to3800 m.The treelines are concentrated in steep slope areas ranging from 15°to 55°.The distribution of alpine treeline elevation shows significant slope differences,with the treeline on the south slope being higher than those on the north slope,and the treelines on the east slope being

关 键 词：高山林线 遥感 自动提取 空间分布 全球 

分 类 号：S771.8[农业科学—森林工程]