中国沙漠 ›› 2006, Vol. 26 ›› Issue (2): 243-248.

• 研究论文 • 上一篇    下一篇


高志海1,2, 李增元1, 魏怀东2, 丁 锋2, 丁国栋3   

  1. 1.中国林业科学研究院 资源信息研究所, 北京100091; 2.甘肃省治沙研究所, 甘肃 武威733000; 3.北京林业大学 水土保持学院, 北京100083
  • 收稿日期:2005-04-20 修回日期:2005-06-22 出版日期:2006-03-20 发布日期:2006-03-20

Study on the Suitability of Vegetation Indices (VI) in Arid Area

GAO Zhi-hai1,2, LI Zeng-yuan1, WEI Huai-dong2, DING Feng2, DING Guo-dong3   

  1. 1.Research Institute of Forest Resources Information Techniques, Chinese Academy of Forestry, Beijing 100091,China; 2.Gansu Desert Control Research Institute, Wuwei 733000, Gansu, China; 3.College of Water and Soil Conservation, Beijing Forestry University, Beijing 100083, China
  • Received:2005-04-20 Revised:2005-06-22 Online:2006-03-20 Published:2006-03-20

摘要: 以位于典型干旱区的甘肃河西地区石羊河下游的民勤绿洲为例,对NDVI、SAVI、MSAVI和GEMI等4种VI(植被指数)受土壤背景光谱影响的程度和探测低盖度植被的能力进行了对比研究。通过分析VI提取植被信息时植被土壤噪音比的变化发现,植被覆盖较低条件下VI提取植被信息总体受土壤背景光谱影响程度较低,相比而言,GEMI提取植被信息中受土壤背景影响最小,其他3种VI的区别不太明显。通过分析不同VI随植被覆盖度增加的反映速率变化及不同覆盖条件下不同VI的取值范围的变化发现,NDVI探测低盖度植被的能力最强,GEMI次之。GEMI可能是干旱地区植被探测较适宜的植被指数。

关键词: 干旱地区, 植被探测, 植被指数(VI), 植被土壤噪音比, VI的反映速率

Abstract: Determining a suitable vegetation index(VI) is the key technique for change detection of vegetation by remote sensing in arid and semiarid regions where the vegetation is sparse and land surface is highly heterogeneous. This paper provides comparisons of the Normalized Difference Vegetation Index(NDVI), the Soil-Adjusted Vegetation Index(SAVI), the Modified Soil-Adjusted Vegetation Index(MSAVI) and the Global Environment Monitoring Index(GEMI) in capabilities for eliminating soil background noise and the capabilities for detecting sparse canopies of vegetation by using TM images in Minqin Oasis locating at the lower reaches of Shiyanghe River, a representative arid region in Gansu Province. The main outcomes of the study are that (1) the influences of soil spectral on extraction of vegetation information in the case of sparse canopies are found to be generally slight by analyzing the changes of ratio of vegetation signal to soil noise(S/N), comparatively, GEMI is more slightly influenced from soil noise than other three VIs; (2) NDVI and GEMI takes first and second positions respectively in capabilities for detecting lower-cover vegetation among total four indices by analyzing the change of reacting rate of different VIs along with the increase of vegetation coverage, and the change of VIs valuing scopes under different land cover conditions. Therefore, GEMI should be relatively suitable VI for application in arid regions.

Key words: arid region, vegetation detection, Vegetation Index (VI), ratio of vegetation signal to soil noise(S/N), reacting rate of VI