Optical Depth 
Rada de Haro, Jure 28 
Minutes from 0:00 UT 
403 nm 
523 nm 
668 nm 
747 nm 
860 nm 
1032 nm 
Fig. 3. Daily course of the optical depth for selected wavelengths at the Rada de Haro site on June, 28. 
3.1.3. Narrow- To Broad-band Spectral Conversion Application To The METEOSAT- AVHRR- 
Albedo Relationship. One of the corrections that must be applied to the satellite observed reflected radiance 
for the determination of surface albedo is the narrow- to broad-band conversion if narrow-band observations 
are used, as is generally the case. This correction factor has an associated uncertainty since that conversion 
depends mainly upon the variation of surface reflectance with wavelength, which is a function of surface 
vegetation and soil properties. The use of a single narrow-band spectral channel to derive broad-band albedo 
gives good results when it is applied to a homogeneous surface. The main difficulty arises in this case from 
assuming that a narrow-band albedo can represent the total broad-band albedo over the whole solar 
spectrum, in spite of the variability in the surface spectral response. The albedo can also be derived from the 
linear combination of the spectral responses of more than one channel. The problem then is to find out the 
coefficients or weights that must be assigned to every channel in such transformation. 
By means of the 5S code (Tanré et al., 1990), the University of Valencia could simulate the 
observation of different satellites in the solar spectrum in a controlled manner regarding a wide range of 
atmospheric, terrestrial and geometric observation conditions. A standard midlatitude summer atmosphere 
was assumed with continental aerosols. Their concentration was changed to obtain visibilities of 9, 13, 17 
and 25 km, in this way covering a common visibility range. Regarding geometric conditions, the most 
representative directions for each satellite observation in a NOAA-11/METEOSAT concurrent pass were 
chosen and the incident direction of illumination from the sun was assumed constant, for June 12, 1991 
(Golden Day) at about 14:30 GMT over a pixel centered at the Iberian Peninsula. The direction of 
observation for such a pixel from METEOSAT was kept constant whereas from NOAA it was changed in a 
range between -50 to 50° for the satellite zenith angle and between 105 to 82° for the satellite azimuth angle. 
Some of the most representative spectra obtained during the EFEDA field campaign (April to June, 
1991) and corresponding to different surface types characteristic of the area were used to carry out a 
multivariate regression analysis between the narrow-band AVHRR channels 1 and 2 and the broad-band 
METEOSAT VIS channel. Spectra from 20 different surfaces were incorporated to the theoretical 
simulation, bringing it nearer to natural conditions. Unfortunately, the surface models do not have sufficient 
information both on the spectral and angular dependence at the same time. Detailed angular models usually 
have low spectral information and those spectrally detailed do not have the bidirectional dependence into 
account or do not cover the whole solar spectrum. These surface data contain no angular information since 
they were obtained at nadir and it was assumed that all surfaces behaved in a Lambertian fashion. 
A data set was then derived consisting of planetary and ground level albedos in both narrow- and 
broad-band, corresponding to AVHRR-1 and -2, METEOSAT-VIS and a rectangular band from 0.3 to 2.5 
|im. The albedos were obtained from the convolution over the upwelling and downwelling fluxes against 
the filter functions given by the different sensors before calculating the ratio of both quantities. Effects due 
to the different spatial resolution of every sensor (AVHRR and METEOSAT) were not considered in the 
theoretical simulation, which assumes that every surface with its spectral signature spreads over an infinite 
area. Only spectral effects are examined in this study because the main interest was to observe the 
relatioship between different sensor signals of NOAA-11 and METEOSAT during an almost coincident 
pass over the Iberian Peninsula around 14:30 GMT. A coincident data set was available for an almost 
cloudless day on June 12, 1991 for the final test. The details of this investigation and the discussion of the 
results can be found in Valiente et al., (1993). A brief and schematic account of the main findings are 
provided now.