-17- 
For both types of satellite images the scan angle is very narrow and the altitude of the platform 
very high. Consequently the earth may be considerd as a sperical surface. If tested over non - moun- 
taineous regions the relief causes negligible displacements. The use of geographic coordinates and 
spherical trigonometry, as for Nimbus evaluations also eliminates the need for reference and 
projection corrections, as they are required for Landsat data. A polynomial of 2nd or 3rd order 
will however incorporate the earth curvature or projection effects. 
4.2. Accuracy Tests for Aircraft Scanner Imagery 
  
Aircraft scanner imagery tests have sofar been performed by Konecny /42, 45/, Liebig /68/, Baker & 
Mikhail /7/ and Deremyi /25/. Unlike satellite scanners, aircraft Scanning systems have a wide 
Scan angle and a relatively low flying hight. Therefore terrain altitude variations cannot be 
neglected. A fair accuracy test should therefore either only be conducted over flat terrain or 
the terrain displacements should either be determinable by a stereo restitution of laterally over- 
lapping flight strips or by a known digital terrain model. 
A further difficulty of comparision exist because of different scanners used (scan angle, nominal 
resolution of a pixel in mrad) and because of different flight altitudes resulting in different 
image scales. It is likewise not useful to describe accuracy at image scale, since digital technigues 
offer a free choice of scale for display. It seems best to compare accuracy to pixel size for a 
nominal angular resolution of the scanner. For Landsat this nominal resolution is 0,086 mrad; 
for a Landsat altitude of 914 km this amounts to 79 m on the ground. Accuracies of half pixel size 
may thus be reached with satellite scanners. 
Table 2 gives a comparsion of results obtained sofar for aircraft scanners. It shows that accu- 
racy referred to pixel size is considerably less.