308 
visibility dominates the clear sky cases (see 
dashed line in Fig. 3.1). The mean relative error 
is only 9 X (refering to the mean daily sum). 
In areas, where ground-based measurements are 
available, these errors can be reduced. 
(2) satellite data sampling, e.g.: 
- errors caused by inaccurate location of each 
pixel. This is important in those areas with 
variabel surface albedoes (e.g. coastlines, 
mountains). 
- errors caused by the temporal sampling rate 
(or the number of measurements which are 
available for a given location). These errors 
occur when determining the daily sum using 
Eq. 2.7. 
If less than four measurements per daylight period 
are available it might be not possible to sample 
the diurnal variation of the global radiation 
sufficiently. In those cases it is required to 
check the model results for systematic deviations. 
Fig. 3.1: mean error for a daily sum of global 
radiation, calculated with AL mln = AL™.* = AL = 5X 
solid line: AM qo< i = OX 
dashed line: AM BO d = 5 X 
3420 J/c 
3240 
3060 
2880 
2700 
2520 
2340 
2160 
1980 
1800 
1620 
1440 
1260 
1080 
o o 
O CM 
O 
Fig. 4.1: Maps of monthly averages of daily sums of 
global radiation for Meteosat (June 1983) and GMS 
(July 1983) 
Fig. 4.2: The same as in Fig. 4.1, but here all 
values normalized with the daily sum for a 
cloudfree atmosphere