利用多时相Landsat8图像提取苜蓿人工草地信息  

Extraction Artificial Alfalfa Grassland Information Using Landsat8 Remote Sensing Data

作　　者：任海娟[1] 董建军[1] 李晓媛[1] 牛建明[1,2] 张雪峰[1] 

机构地区：[1]内蒙古大学生命科学学院,内蒙古呼和浩特010021 [2]中美生态能源及可持续性科学内蒙古研究中心,内蒙古呼和浩特010021

出　　处：《中国草地学报》2015年第2期81-87,120,共8页Chinese Journal of Grassland

基　　金：现代农业产业技术体系建设专项资金;国家重点基础研究发展计划("973")项目(2012CB722201);国家科技支撑计划课题(2011BAC07B01);国家自然科学基金(31060320)

摘　　要：以内蒙古阿鲁科尔沁旗为例,利用2013年5月5日至10月21日期间的16景Landsat8OLI和1景Pleiades遥感图像,研究苜蓿人工草地遥感信息提取方法。结果表明:依据一年中苜蓿生长与生产动态变化特点,提出基于多时相Landsat8OLI遥感图像的NDVI加和法和NDVI差值加和法。非监督分类方法的提取精度最高(100%),NDVI差值加和法次之(94.55%),NDVI加和法最低(91.23%)。NDVI加和法会出现误判,不宜用于苜蓿人工草地遥感信息提取;非监督分类方法与NDVI差值加和法各有优势,只要依据多时相Landsat8OLI图像,结合野外调查数据,均能够获得满意的结果。Based on 16 scenes Landsat8 satellite remote sensing data from May 5th to October 21st in 2013, the extraction methods of artificial alfalfa grassland remote sensing information were studied. This research was conducted in Alukeerqin. The results showed that： According to dynamic change characters of the growth and production of alfalfa in one year, two approaches of summation of NDVI and summation of NDVI difference were proposed based on muhi temporal Landsat8 remote sensing images. The extrac tion accuracy of alfalfa grassland patch with the unsupervised classification method was the highest （100%）,summation of NDVI difference was followed（94.55%） and summation of NDVI was the lowest （91.23%） ;Summation of NDVI was not suitable for extracting alfalfa grassland because its result had some errors. Both unsupervised classification and summation of NDVI difference approaches had advanta- ges, and they can obtain good results if there were multi-temporal Landsat8 remote sensing images and field investigation data.

关 键 词：苜蓿 遥感 非监督分类 多时相 

分 类 号：S127[农业科学—农业基础科学]