23. Satellite data 
498 
23.1. NOAA-ll/AVHRR-3-4-5 data. After proper navigation and calibration, the Li and Becker (1993) method 
for surface temperature and channels 3, 4 and 5 emissivities retrieval was implemented. It was checked that the 
noise of channel 3 was low during the studied period (less than 0.5 K). Both daytime and nighttime data are used 
(0200 +/- 60' and 1400 +/- 60' UTC, respectively). The data are first corrected for atmospheric effects by using 
LOWTRAN-7 (Kneizys et al. 1983). As a doubt exists on the nighttime cloud-mask, differences may exist 
between channels 3, 4 and 5 derived temperatures. It is shown that a simple thresholding on these differences 
enables to filter nighttime clouds. 
23.2. METEOSAT data. After navigation and calibration, atmospheric corrections are performed with 
LOWTRAN-7. The NOAA-derived surface temperature serves as a reference to estimate the surface emissivity 
for METEOSAT. At NOAA time, the Planck law for the METEOSAT window is first inversed for each 
cloudless AVHRR pixel. Provided the cloud coverage is reasonably low, the obtained radiances at AVHRR pixel 
size (about 1.2x1.2 km) are then averaged over the corresponding METEOSAT resampled pixel (about 6x6 km). 
Given this black-body radiance, the measured METEOSAT radiance (corrected for atmospheric effects) and the 
computed atmospheric downwelling emission, the METEOSAT surface-emissivity can be obtained easily. 
Depending on the surface type, it happens that daytime AVHRR-derived METEOSAT emissivity is significantly 
different from the nighttime value. In order to extrapolate emissivity, it is supposed that emissivity variations are 
due to vegetation properties, and that these properties depend on the direct incident solar radiation. Therefore, 
daytime emissivity is written linearly as a function of (M0.7 ^ u ) , where u = cos@q , and @q is the solar zenith 
angle. Once emissivity is known, surface temperature can be computed directly from METEOSAT data with a 
high temporal resolution (every 30'). An example of METEOSAT-derived surface emissivity and temperature 
over the Tomelloso test-site (vineyard) is shown by Figures 1 and 2. 
TIME (UTC) 
Figure 1 - NOAA/AVHRR-derived and extrapolated emissivity for METEOSAT/IR at Tomelloso 
(vineyard). June 23, 1991. 
Figure 2 - Comparison between the diurnal surface temperature cycle as determined from 
METEOSAT/IR and measured separately for soil and vegetation at Tomelloso. June 23,1991.