地理科学进展 ›› 2013, Vol. 32 ›› Issue (6): 913-923.doi: 10.11820/dlkxjz.2013.06.008

• 遥感与GIS模型应用 • 上一篇    下一篇

基于图像阈值法的森林雪灾损失遥感估测——以云南省为例

吴健生1, 陈莎1, 彭建2   

  1. 1. 北京大学深圳研究生院城市规划与设计学院城市人居环境科学与技术重点实验室, 深圳518055;
    2. 北京大学城市与环境学院地表过程分析与模拟教育部重点实验室, 北京100871
  • 收稿日期:2013-02-01 修回日期:2013-05-01 出版日期:2013-06-25 发布日期:2013-06-25
  • 通讯作者: 陈莎(1987-),女,硕士研究生,主要研究方向为土地利用规划、景观生态学。E-mail:chensha@pku.edu.cn E-mail:chensha@pku.edu.cn
  • 作者简介:吴健生(1965-),男,副教授,主要从事土地利用与景观生态学研究。E-mail:wujs@szpku.edu.cn
  • 基金资助:
    国家自然科学基金项目(41271195)。

Assessment of forest damage due to ice storm using image thresholding techniques: A case study of Yunnan Province

WU Jiansheng1, CHEN Sha1, PENG Jian2   

  1. 1. Key Laboratory for Environmental and Urban Sciences, School of Urban Planning and Design, Shenzhen Graduate School, Peking University, Shenzhen 518055, China;
    2. Laboratory for Earth Surface Processes, Ministry of Education, College of Urban and Environmental Sciences, Peking University, Beijing 100871, China
  • Received:2013-02-01 Revised:2013-05-01 Online:2013-06-25 Published:2013-06-25

摘要: 雪灾是森林生态系统的严重生态干扰之一, 会严重影响林木生长并降低生态系统功能。利用2000-2011年SPOT NDVI长时间序列影像数据, 基于SG滤波函数进行时序重建, 采用灾后同期影像的图像阈值法, 以常年多灾的云南省2008 年雪灾为例, 进行雪灾森林植被受损评估。评估结果与全国灾情月报中的云南省雪灾范围基本一致, 全省森林雪灾受损面积达12.09%, 主要分布在滇西北的迪庆州、怒江州。对森林雪灾受损程度分析表明, 全省重度受灾县7 个, 中度受灾县13 个, 轻度受灾县45 个。对海拔、坡度、坡向、坡位等地形条件的叠加分析表明, 海拔3300~4000 m、坡度5~15°、处于东坡或东北坡中坡位的森林植被受损最严重。本研究有助于实现森林雪灾损失的快速评估, 对有效规避次生灾害、开展灾后恢复重建规划具有重要意义。

关键词: SPOT NDVI, 森林植被, 图像阈值法, 雪灾, 云南省

Abstract: Ice storms are one of the severe disruptions to forest ecological systems, causing vegetation loss and reduction of the ecological systems' functions. For this reason it is vital to assess the damages to forests after ice storms. Using SPOT Normalized Difference Vegetation Index(NDVI) time serial images of Yunnan Province of China during 2000-2011, forest damage caused by ice storms in 2008 was assessed based on image thresholding techniques of post-storm NDVI time series after Savitzky-Golay filtering by TIMESAT software. The damage threshold was determined by the difference of standard deviation between the years with ice storms and those without, which eventually turned out to be 21%. The range of extracted forest damage is almost consistent with the ice storm extent of Yunnan in the national monthly disaster report, therefore the result is reliable. The destroyed vegetation accounted for 12.09% of the total area of forest. Forest within Diqing County and Nujiang County, in northwest Yunnan, suffered the most losses. On the whole, seven counties took the worst hit by the natural adversity, while thirteen were moderately affected and forty five slightly affected. The most severe damage of forest occurred at the elevation of 3300 m to 4000 m, the slope of 5 to 15 degree, the middle slope position and the east or northeast aspect. Even so, it had little to do with slope position because the most of vegetation is located in the middle slope position. In-situ measurement was not employed here to verify the results because of time and money limits, which compromised the overall accuracy. However, with the acceptable precision, the research method can be used as a real-time forest loss assessment, which is of great significance for taking effective measures to avoid secondary impacts and starting the process of recovery.

Key words: forest, ice storm, image thresholding techniques, SPOT NDVI, Yunnan Province