Balsam fir occurs mainly as an understory species in the boreal 
forest of Ontario, except occasionally where the overstory has been 
decimated to the extent that the balsam fir understory is exposed. If 
it is exposed, balsam fir can be identified. However, since the 
reflectance from balsam fir appears similar to that from jack pine 
on the imagery of this study, further investigation of the discrim- 
ination of the two species using Landsat data is required. 
METHOD 
Twenty-five four-hectare plots were selected in the boreal forest 
of Ontario so as to include a wide variety of species and combinations 
of species and wide variations in crown closure. This data was used 
to create training plots for discrimination of forest types using 
digital Landsat data. The plots were half the size of the smallest 
type (8 hectares) distinguished during the standard forest inventory 
procedure. A 100% tally of the tree species within the plots was 
recorded in the field, together with measurement of diameter at breast 
height and crown closure. Table 1 shows the data from one tally. The 
supervised classification of the Landsat tapes was conducted by colour- 
coding the digital value of the reflectance from the major species 
represented by each of the training plots, using the modified G.E. 
Image 100 System of the Canada Centre for Remote Sensing. The computer 
"map" resulting from delineating areas with similar reflectance values 
was then compared with a "map" produced by detailed photo interpret- 
ation of the species composition of stands using the panchromatic 
aerial photography. 
DISCUSSION OF RESULTS 
To test the applicability of digital analysis of Landsat data for 
areas where a rapid estimation of the occurrence of specific tree 
species was required, a study area of 500 hectares in size was selected. 
Over 100 stands were delineated within the study area by interpretation 
of forest conditions using 1:15,840-scale aerial photography (Figure 1). 
This level of typing was more detailed than that used for current forest 
inventory maps but it was broader than the maximum level of detail which 
could be attained from this scale of photography. The objective of the 
detailed photo interpretation was to show the wide variations in species 
composition between stands and the wide variations in the proportional 
representation of species for stands which had the same species compos- 
ition. Since this level of forest stand mapping would be impractical 
for use in an inventory survey, which would necessarily involve a large 
area, the 100 stands delineated were grouped into three broad classes 
which were defined as follows: 
   
      
  
  
  
   
    
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
   
    
  
  
  
  
  
  
  
     
   
     
The 
bal: 
are 
pre: 
pre 
It 
Spr 
for 
the 
by 
tap 
ref 
res 
var 
can 
tha 
of 
typ 
wit 
but 
val 
giv 
con 
ren 
con 
maj 
not 
evi 
fre 
ph 
ae] 
(T: 
ty 
ge