As was the case with Drivdalen, the spectra measured with a longer 
point-spacing Ax yield a somewhat steeper inclination and a smaller E. 
Compared with Drivdalen, the standard deviation due to surface undula- 
tions is of only half the size. For smell sample spacings no significant 
differences between the predicted values are observed. In contrast to the 
results of Drivdalen, the values for A and B increase. more rapidly than Eh 
is the case for D and E. a ; 1 
However, it only makes sense to state that for Ax - 5o meters the contri- ith 
bution from the terrain undulations to the resulting accuracy of a DEM 
is approximately 1.0 meter. 
Uppland, Sweden. 
Profiles from the test area Uppland have been measured by 5 groups: A,B, 
C,D and E. In figure 5, spectra estimated from profiles measured by C and 
by D are given. The profiles from C were sampled with a point distance of IM 
30 meters, while the profiles from D were sampled with a spacing of 5 me- | 
ters. The profiles from A,B,C and D were measured at scale 1:30.000, while 
the profiles from E were measured at scale 1:6.000. 
  
LaC? (M3) 
  
  
10% a 
10? Y \ 
“10! A 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
193. 108 10 10° 10° Io gt 20 
WAVELENGTH (M) WAVELENGTH (M) 
Measured by D Measured by C 
Fig. 8. Spectra ‘of Uppland, Sweden. 
  
  
  
  
Participant Sample spacing Coeff. for straight line approx. 
Ax meter of terrain spectrum 
a E 
4 30 2.21 2208 00. 
B 30 2.35 7.35°10_, 
C 30 2:28 1.10719 + 
D S 2.14 1.37 10 i 
E 2.8 2.20 9.77*10 
  
  
  
  
  
Table 7. Uppland, Sweden.