ESSEN 
Fig.2: Test a 
A A 
k: k 
1.0 LCD cos i 1.0 (1-s)*s cos'e 
x. ; 
cos'e >. : | 
» 
So At AAT A, 
   
  
   
  
  
  
  
Fig. 1: Extended normalisation funktion 
As already mentioned above the correction due to the 
observation angle will not be pursued any longer in our 
current work. We recommend further investigations. The 
examples later show that the influence of the observation 
angle seems to be negligibly small. 
Before we continue with the determination of the para- 
meters of the normalisation function it is advisible to prove 
that in reality this function may serve as suitable model. For 
that reason we chose our test area in the Austrian Alps 
where high mountains and steep slopes guarantee all sorts 
of illumination effects, from brightly sun lit regions with 
incidence angles around 0°, to dark shadows where the 
(virtual) incidence angle is greater than 90°, all sorts of 
shades in between, and eventually even cast shadows. 
Figure 2 shows an overview of the some 35 km x 25 km 
large area in the so-called "Salzkammergut" with its lakes. 
We use in our investigation the six optical bands of a 
Landsat TM image (1 to 5 and 7). 
The DTM provided by the Austrian Federal Office for 
Metrology and Surveying has a grid width of 25 m x 25 m. 
The geometrically rectified satellite image has the same 
ET EEE NC CUP NUDO SORS asc 
rea located in the Austrian Salzkammergut (Landsat TM, Band 4) 
      
    
resolution. We want to emphasize that the geometric 
rectification must be performed with the help of a parametric 
model, so that displacements due to terrain heights are also 
rectified. In our case we used the bundle block adjustment 
program ORIENT [Kager, 1989] in conjunction with the 
digital orthophoto modul of SCOP DOP [Molnar et al. 1998]. 
High geometric accuracy is a prerequisite for a good quality 
of topographic normalisation in particular along terrain 
discontinuities, such as steep ridges or narrow valleys. 
  
  
Fig.3: cos i of the DTM (black = i greater 90°) 
10 Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 
First we : 
calculatin 
either be 
delivered 
known act 
sun azimu 
(90 - 9) is 
western ‘ 
detail of 
patches ¢ 
90°. 
In order to 
sun incide 
two dime 
along ver! 
the horiz 
band 4 ca 
in this sce 
is one qu 
the diagr 
test area 
Below th 
grey. It is 
horizonta 
they stre! 
the DTM 
area prc 
should th 
to avoid 
that hav 
normalis: 
The mos 
o firstly 
mean 
angle 
supp: 
appre 
beha 
simpl 
. Seco 
depei 
figure 
signif 
of th 
angle 
influe 
from 
e Third 
are r 
(see 
COS“ ( 
are tr 
not EL 
prepi 
there 
instal 
The cor 
formula 
would pr 
to achiev 
entire im