The Landsat images used in this investi 
gation are: 
date path/row 
MSS Oct.6,1972 203/25 
May 29,1973 
June 26, 1976 
TM April 20,1984 193/25 
July 9, 1984 
Sept.30, 1985 
April 29, 1987 
The interpretation of the false color in 
frared composite of the bands 4,5 and 7 
of Landsat-MSS and bands 2,3 and 4 cf 
Landsat-TM displayed in blue, green and 
red respectively, were the basis for the 
selecting of the training areas for the 
analysis of the spectral reflectance 
properties of the forest types and damage 
levels, as well as for the supervised 
classifications. 
size). The investigation of the spectral 
signatures provided good information 
about the reflectance properties of the 
foiest damage levels on the landsat-MSS 
as well as on the landsat-TM. A healthy 
forest stand has high values of reflec 
tion in the green spectral band (TM2 arid 
MSS4) ; low reflection value in the rod 
spectral band (TM3 and MSS5); relatively 
high reflection value in the NIR spectral 
band (TM4 and MSS7) ; and low reflection 
value in the middle IR bands (TM 5 and 
TM7) . But the damaged stand has, depen 
ding on the damage level , a higher over 
all reflection in the visable spectral 
bands and lower reflection in the NTF 
band than healthy vegetation of the same 
type, but also higher reflection value 
in the middle IR due to the dryness and 
amount of exposed branches in forest 
canopy (Kadro 1984,Kadro 1986' . Fig. (1 
and 2) show the spectral reflectance 
characteristics of different training 
areas with different damage levels. The 
The false color infrared composite image 
is very useful for interpretation purpo 
ses of vegetation condition (health) and 
type, since this image product is a 
mimick of the infrared aerial photo 
graphs. Due to clear differences in the 
spectral signatures of the defined damage 
classes and vegetation type classes on 
the false color composite images, it was 
possible to successfully determine the 
training areas regarding to these various 
classes. 
3 Results 
3.1 Investigation of Spectral Reflectance 
Properties 
To investigate the spectral reflectance 
properties of different damaged classes, 
the training areas must be determined by 
the user. Generally the available CIR 
photographs and ground check data assist 
and in combination with the in 
terpretation and analysis of digital data 
as an infrared color composite for deter 
mination and delination of training 
areas. In case of the test site at St. 
Blasien the above proceedure was used. 
However for the test site at Erzgebirge, 
there was no information, only the land 
sat images, and the interpretation of 
the Landsat data in form of false color 
IR composites were the basis for the 
selecting and definition of the training 
areas concerning the damage classes. The 
use of this method was based on intensive 
experiments gaine working with a fun- 
compliment of availble data at the 
St.Blasien test site. 
This procedure to select and to define 
the training areas required a good know 
ledge about the spectral reflectance cha 
racteristics of the different vegetation 
type and especially regard the speerra) 
reflectance properties of vegetation in 
dependent of damage. This is very impor 
tant, because the textures and the struc 
tures of forest canopy can not be ob 
tained and studied directly using Landsat 
data due- to the ground t-solution (pixel 
advantages of the Landsat-TM consists of 
the additional spectral bands TM5 and TM7 
in the middle IR region, as well as the 
infrared with spatial resolution (30 ri in 
the TM,80 m in MSS). These advantage help 
to identify the forest damage conditions 
in the ratioing evaluation. 
3.2 Ratio Evaluation of Landsat Data 
The rationing between spectral bands pro- 
vid an enhancement and improvement of 
available information. The mentioned dif 
ferences of spectral signaures in 
different spectral bands provide the 
possibility of taking the advantage of 
this phenomena to make a rationing bet 
ween certain bands, for special tasks. 
In the Ratio TM5/TM4 the damage areas 
have the highest Ratio due to the higher 
reflection of damaged area in band TM5 
and lower reflection in TM4 than the 
healthy one. The hight of the Ratio 
depend on the damage level. This ratio 
makes it possible to seperate the damaged 
areas from non damaged areas in the fo 
rest area. 
Another Ratio that is investigated in 
this work is the Ratio TM4/TM3 (Ratio of 
vegetation index= RVI) that shows the 
distribution of the vegetation in one 
image. Healthy vegetation has a higher 
Ratio than the other classes like soil or 
water surface. This is due to high 
reflection of vegetation in NIR and low 
reflection in the red spectral band. 
Different Ratio images have been 
calculated from the landsat-TM data of 
the test site Erzgebirge. A part of the 
test site is shown in photo (6) part A. 
is a infra red color composit (TRCC). 
The image in photo (6) part (B) is a co 
lor composit of two ratio’s and original 
spectral band. The Ratios are 
TM5/TM4,TM4/TM3(RVI) and TM2 displayed in 
red, green and blue respectively. In this 
ratio combination damaged and severely 
damaged area turned red and the 
healthy/slightly damaged forest. area 
turned green. 
427