727 
The use of blue band in ARVT, SARVI. and ASVI increased their sensitivities to soil background variations 
(Fi°. 3c). The soil sensitivities of these indices are worse at lower vegetation density (LAI < 0.5) and are the same as 
those of NDVI, SAVI, PVI, and MSAVI when LAI is greater than 0.5. The SAVI and MSAVI still remained the least 
sensitive to the soil background variations, and their resistances to soil noise were also inherited by the SARVI and 
ASVL The GEMI was very sensitive to soil background at low vegetation density, but quickly became the most 
insensitive index to soils. Consequently, for vegetated land surfaces, ASVI, GEMI, MSAVI, and SAVI can all be used, 
depending on their sensitivities to atmosphere conditions. 
The atmospheric sensitivities of all indices are plotted in Figure 3d. As can be seen, the PVI and MSAVI are 
more sensitive to atmospheric conditions than NDVI (Fig. 3d), while the ARVI, SARVI, and ASVI are the least 
sensitive indices to atmosphere at high vegetation densities (LAI > 0.6). The GEMI was almost independent on 
atmospheric conditions at low vegetation densities (0.2 < LAI < 0.6). The sensitivities to atmospheric conditions are 
also demonstrated in Figure 4 where only GEMI, ARVI, SARVI and ASVI were plotted for 12 different visibilities as 
a function of vegetation densities. For all vegetation densities, the most pronounced differences due to the atmosphere 
were found with GEMI (Fig. 4a), while the least affected indices were found with ARVI (Fig. 4b) and ASVI (Fig. 4d). 
The atmospheric effect on ARVI was consistent with vegetation density while that on ASVI changed a little. At 
vegetation densities of LAI < 1, the ASVI and SARVI were the best ones, due to their insensitivities to soils. 
In general, there is no single vegetation index that is better in all aspects: high vegetation sensitivities and low 
soil and atmosphere sensitivities. Consequently, the choice of vegetation indices to be used in practical applications 
depends on the conditions under which the remote sensing measurements were taken. General guidelines are 
demonstrated in Figure 5. When there is only one single factor to be the major concern, one should consider either 
Figure 5a, or 5b, or 5c for sensitivities to vegetation, soils, and atmosphere. When all factors co-exist, one should 
consider Figure 5d with respect to vegetation densities (seasons, crop growth stages, etc.), soil background conditions 
(uniform soil background, or large variations), and atmospheric conditions (always clear, frequent variations from day 
to day, or atmospheric corrections made). 
Vegetation Sensitivity 
Figure 3. Dyn ami c ranges (a) of the eight vegetation indices and their sensitivities to vegetation (b), 
soils (c), and atmosphere(d) using simulated data with Rayleigh correction.