818 
2 - WATER VAPOR RETRIEVAL FROM THERMAL CHANNELS 
In channel 3 the effect of the water vapor should be accounted for. As it varies strongly with time and space, it is 
important to get the water vapor content at the time and location of the AVHRR observed radiances. Several 
published studies have used T 4 -T 5 (Split Window Technique) for the retrieval of the amount of water vapor in 
the atmosphere [Prabhakara et al., 1974; Dalu, 1986; Schluessel, 1989; Justice et al., 1991]. We tried to retrieve 
and to improve the relation between water vapor content and T 4 -T 5 both over ocean and land. 
2.1. Retrieval of water vapor over sea. 
Over sea, we used one year of SSM/I and AVHRR data over the Pacific Ocean, to establish the relation. We 
assume that water vapor content is linearly proportional with T 4 -T 5 [Dalu, 1986]: 
Uh 2 o — A(0). (T 4 -T 5 ) (01) 
A plot of SSM/I water vapor content versus nadir measurements of T 4 -T 5 is given Figure la. The coefficient A 
can be deduced: A(0°)=1.98 ± 0.5 g/cm 2 /°k. For a viewing angle 0=60°, this coefficient becomes, Figure 1-b, 
A(60°)=1.53 ± 0.5 g/cm 2 /°K. 
Figures 1 : Plot of Uh 2 0 from SSM/I versus (T4-T5) for 2 zenith angles (a) 0° and (b) 60°. 
The angular dependency for some values of cos(0) is shown in Figure 2. The retrieved results are then compared 
to theoretical computations. Defining 36 different atmospheres with a water vapor content ranging from 0 to 6.5 
g/cm 2 , we computed with Lowtran7 the coefficient A for different angles. These computations confirm the 
angular dependency of coefficient A that does not follow a cos(0) law [Dalu, 1986]. 
Figure 2: Angular dependency of A(0) determined over sea both by the measurement 
and by the computations. Also reported, the law in cos(0) 
2.2. Retrieval of water vapor over land. 
Over Land, to retrieve the relation we used AVHRR data and the sunphotometer measurements performed 
during the SCAR-A experiment which has been conducted in the Eastern US coast on July 1993. Figure 3 shows