AVHRR-NDVI已广泛应用于宏观土地覆盖研究,但在区域研究中应用比较少。论文根据1999年锡林郭勒草原4种草地NDVI的动态过程,结合对应AVHRR提供的额外信息,对4种草地进行了分类。首先,根据全年NDVI低平的特点检测出荒漠;第二,使用6月Ch3>3180和Ch5>3100区分出沙地草地;第三,使用8月Ch2>200或6～8月NDVI_6～8月>1.2检测出长势良好的草甸。与基于TM的草地图比较,AVHRR的分类结果失去了许多细节,但对土地覆盖长期变化研究仍有价值,特别是对沙漠化过程的监测。需要进一步的研究才能确定AVHRR在区分长势差的草甸和典型草原方面的潜力或局限。

AVHRR-NDVI has been widely used for land cover characterization in global and national scales,yet there are few reports on regional scale,and fewer investigations about the potentials of the original bands of NDVI for grassland classification.This paper explores the seasonal dynamics of four grasslands(meadow-steppe,steppe,desert-steppe,and sand-steppe)in Xilingol,Inner Mongolia using1999AVHRR and its NDVI .First,desert could be easily detected due to its low-flat NDVI pattern.Second,Ch3>3180and Ch5>3100in June could be applied to detect sand-steppe from other two grasslands.Third,Ch2>200in August or NDVI_6～8 >1.2is used to differentiate good meadow-steppe from other land covers.The results have less details compared with TM-based classification,but there are still informations for long-term change detection,espe-cially for desertification.More works are needed to verify the potentials and limitations of AVHRR to detect poor meadow-steppe from normal steppe.