to east. A parameterization of this atmospheric behaviour could be helpful in forthcoming studies to achieve
a better estimation of NOAA albedos. If this could be achieved, the NOAA data would be a better system to
derive the broad-band albedo than METEOSAT data. Furthermore, if the range for the NOAA viewing
angles is not large, the transformation using eq. ( 2 ) could give estimated albedos within an acceptable error.
Fig. 4. Scatter diagram of METEOSAT vs predicted planetary albedo using eq. (lb). The albedos were
calculated from the satellite image of June 12, 1991. The diagonal indicates the position in the scatter
diagram where predicted albedos by means of eq. (lb) are coincident with actual values.
Fig. 5. Image comparison between METEOSAT albedo and predicted METEOSAT albedo using eq. (lb)
with NOAA-11 data for June 12,1991 The anisotropy of the earth-atmosphere system reflectance divided
the Iberian Peninsula into two distinct regions The NOAA-11 trajectory which crosses by Lisbon has also
been indicated.
3.1.4. Variation Of The NOAA-11 Viewing Angle. The Free University of Berlin obtained
outstanding results on the empirical determination of corrections for inclined viewing directions
(Koslowsky, 1993). Daily values of planetary albedo derived from AVHRR-1 or of NDVI show
considerable variability which partly could be due to the influence of clouds, but there is still a strong
dependence on satellite viewing and solar zenith angles (Fig. 6 ). The application of cloud clearing methods
yields better results but does not remove the influence of the sun/satellite geometry. The maximum NDVI of
a period of 7 to 10 days is taken normally as representative for that period. This leads to acceptable results
in less cloudy regions, but requires a period length that fits the geometry repetition rate of 8.3 days in 1992
for NOAA-11. This repetition rate varies with the lifetime of the satellite. Even if longer periods are used to
select the maximum NDVI, a phase shift may result in an increasing variability. Thus the annual course of
NDVI shows a great variability which docs not arise from changes in vegetation cover.
