The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008 
1045 
Figure 1. comparison among feature spaces from various vegetation indexes above mentioned 
Figure 2. comparison among feature spaces from various vegetation indexes above mentioned 
According to the figure 1 and 2, the vegetation NDVI may 
exaggerate the information of low coverage vegetation and 
inhibite that of high coverage vegetation such as field farm, but 
the vegetation index RVI is inverse to that of NDVI. As to 
MSAVI, it is between NDVI and RVI. So the vegetation index 
NDVI is effective to research the total areas of vegetation in 
Tarim river basin. 3 
3. MODEL, VERIFICATION AND RESULT ANALYSIS 
The remote sensing vegetation index always changes with life 
process and growth periodic time of vegetation from bud 
bursting, growth to wilting. So we thought that remote sensing 
vegetation index of life process or growth of green vegetation 
would change, but the soil and litter would not cause spectral 
changes. In this paper, based on previous models and methods, 
we compared the remote sensing Vegetation index, NDVI, RVI 
and MSAVI (250m), which are most efficient parameters and 
are used widespread in arid and semiarid area, and selected RVI 
as basic data of information extraction through their comparison. 
According to above analysis, we presented standard deviation 
model using time serious MODIS vegetation index RVI. The 
time series MODIS RVI products (every 16 day) was selected 
from April to October. The methods were as following: 
standard deviation model : 
1 
n 
1=1 
n — 1 
n 
L 
.