Figure 2: Reflectance for vegetation, p„ (after Tanre 
et id. (1986)) and road dust, pi (after Frey (1992)). 
SPOT data for reflectance of tundra vegetation (*) 
and dust (O). Reflectance of band XS2 (X and □), 
corrected for an erroneous calibration of that band 
(see text for details) 
Dust Load after Lamprecht and Gräber (1993) 
30 
Figure 3: Characteristic radius r c (z, y) 
Satellite-based Dust Load 
Figure 4: Dust load u, t0 eh after Lamprecht and Graber (1993) (left) and satellite-based dust load u (right). 
See text for details 
In Fig. 5 nine west-east dust load profiles of u and u <t0C A averaged over 11 image rows are plotted. The 
approximate road positions are indicated as well. These profiles are sections across the south-north oriented 
branch of the highway. In the same figure south-north profiles of the west-east oriented branch are given as 
well. In both figures, the coincidence between the road position and the location of maximum dust load is 
evident. Moreover, a remarkable agreement between the SPOT-based and the stochastic profiles is observed. 
This is particularly true in the west-east profiles, i.e. across the prevailing wind direction. The agreement is 
less pronounced for the south-north profiles oriented along the main wind. It is worth to be recalled that the 
values on the road axis were not included when calculating the rescaling factor. Hence, these values should 
be overlooked when comparing the profile data. 
The model of deriving the dust load from the near-infrared satellite channel is rather simple and may 
not take into account all effects occurring in reality. For instance, the dust is assumed to be deposited on a 
fiat plane. In reality, dust may drop from the leaves to ground and be hidden by the canopy. Moreover, for 
dust loads covering a substantial proportion of the ground area, the assumption of spherical particles which, 
regarding reflectance, behave like individual disks, is probably not valid. Neglecting multiple reflexions at dust 
and vegetation surfaces may introduce additional errors. Finally, the non-uniform nature of the vegetation