44 
on which sensor causes i 
\^aittÊÊÊÊÊÊKÊiaÊÊÊÊÊÊÊÊÊÊtÊKÊÊÊM 
derived over Australia, for AVHRR channel 1. - This AVHRR channel 2 
study.+ Abel et al. o Kaufman and Holben x Che et al 
(CNES). * Che et al.ÇUAZ) 
Figure 4 same as figure 3.a, except over USA Figure 5 same as figure 3,a, except over Libya 
Figure 6.a Calibration factor ratio as derived over: 
- China and -- Libya, for AVHRR channel 1 
channel 2 
different in latitude, altitude, and climate. 
The results obtained with METEOSAT show smoother evolution (Figure 7). In addition, they do not show any cyclical 
behavior as could be observed cm the AVHRR calibration factor. It has to be underlined at this point that the two data sets 
widely differ in time resolution and that the GVI product suffers a subsampling, based on the maximization of vegetation 
index, particularly inadequate over desertic areas. Moreover, this subsampling is »rr/vnppni,vt by a reduction in the 
angular sampling of the bidirectional reflectances that grieves the retrieval of consistent parameters. On the other hand, 
the angular sampling available in METEOSAT measurements is restricted in another way. Actually it presents rally one 
view zenith angle. 
Nonetheless, we tried to intercalibrate the two sensors using equation 3. The obtained values, plotted in figure 8, are 
laiger than any calibration factor obtained elsewhere (Kriebel and Amann, 1993) but, at this time, we have no indication 
Figure 7 Evolution c 
over Libya, for ME! 
5. CONCLUSION 
In this study we develop 
relies on the analysis of 
this work has outlined tl 
from the reflectances va 
reflectance due to this 
(atmosphere, directiona 
between all these effec 
atmospheric effects usir 
then by correcting diret 
these various correction 
This has been done ove: 
are always of the same 
differences appear on c 
reflectances: e.g. the Au 
atmospheric scattering i 
Still, the point that app< 
directional behavior of 
targets seem more apprc 
such areas. But the mor 
overcome misregistratic 
that it can hamper bad! 
calibration factor. Still, 
sensitive to atmospheric 
After applying the de 
intercalibration of the t 
channel about 30% higl 
to the selected target, w< 
Despite all these uncer 
process. It is definite tl 
AVHRR data should lea 
field of view sensors. 
ACKNOWLEDGEME 
This work is supported 1 
Ministère de l’Enseigne