1108 
CT M 
N*atm(VI) = VI* (Oj, Vis) ] da 
where VI* Hreignatpc the value of the index at the top of the turbid atmosphere (with a prescribed amount of 
aerosols specified as a horizontal visibility), and where VI R is the value that would be observed at the top of 
the same atmosphere in the abscence of any aerosols. Figure 5a and 5b exhibit the value of the S/N ratios for 
the five indices as estimators of the ct , estimated as explained above, for two ranges of fractional cover over 
each of the 10 selected soils in order of brightness, in the case of a clear atmosphere (visibility of 23 km). 
Figure 6a and 6b show the same information but above 
Figure 5a: S/N for a clear (vis.=23 km) vs. a Rayleigh 
atmosphere for a variation cover range [0-50%] 
a much more turbid atmosphere (visibility of 5 km). 
Figure 6a: S/N for a turbid (vis.=5 km) vs. a Rayleigh 
atmosphere for a variation cover range [0-50%] 
Figure 5b: S/N for a clear (vis.=23 km) vs. a Rayleigh 
atmosphere for a variation cover range [50-100%] 
•oil type 
Figure 6b: S/N for a turbid (vis.=5 km) vs. a Rayleigh 
atmosphere for a variation cover range [50-100%] 
In the case of dark soils and for both clear and turbid atmospheres, GEMI is always superior to 
NDVI, SAVI and MSAVI and NDVI is always the worst of all the indices, irrespective of the range of 
fractional cover considered. For dark soils and for a turbid atmosphere, WDV1 and GEMI have similar 
behaviour in the higher range of vegetation cover. Over the next two medium soils (6 and 7) again for both 
types of atmospheres, the NDVI is the better index, followed by GEMI for the lower values of fractional cover, 
or by MSAVI in the higher range. For the next tho medium soils (8 and 9) and for the bright soil (10), GEMI 
appears to be the best index in the lower range of vegetation cover and NDVI in the higher range. SAVI and 
MSAVI appear to have an average performance throughout the entire experiment WDVI appears slighly