Computer Time 
It is estimated that a full frame classification (185 
x I 85 kms) will use on the average about three hours of computer 
time. This will vary with the complexity of the classification in 
terms of both numbers of units and any modifications that might be 
required, such as changes in lookup tables. Measured against the 
use of conventional air photographs, this is also a real cost to 
the user which would not enter into the latter approach. 
User Time 
Disregarding the vagaries of computer systems, it is 
felt that the total time taken to generate a full frame ERTS map 
should be about three to four days. Added to this would be a 
variable time factor of up to two weeks for processing the image 
on the EBR system, depending on how readily it could be accessed. 
If an area to be mapped is covered by parts of four ERTS frames, 
equalling about 32,000 kms , then three to four weeks should be 
allowed to produce the terrain maps. Comparing this with the more 
traditional airphoto technique, coverage of an equivalent area 
should take three to four man-months, in order to classify the ter 
rain and transfer the data to a suitable base map. The only real 
advantage in using the latter technique is that, with the addition 
of the third dimension, greater detail and more precise definition 
of the nature of the surface materials can be achieved. 
It should be noted, finally, that the Bend.ix-PDP 10 
system used on this project is not economically feasible as an 
operational tool, largely because of its slowness, and time 
sharing costs. Further research on this project will use the 
Image 100 Series system which not only uses a dedicated computer 
but is also fully interactive, reducing both time and operational 
costs o 
CONCLUSIONS 
Programs of small scale terrain mapping in the north 
may be enhanced by the use of automated techniques of multispectral 
analysis of ERTS-1 data. However, it must be stressed that the 
product is useful only as the starting point for other, more con 
ventional mapping activities. It is felt that human experience 
and expertise alone determines the validity of the classification, 
and that this can only be verified by substantial ground checking. 
While the use of ERTS saves time and effort in the early 
stages of a mapping program, there is a continuing need for con 
ventional photography to complement the ERTS data, and to identify 
areas requiring more detailed attention. 
Further testing of the technique is required. There is 
a need to establish new training sets for each succeeding ERTS 
frame. We need to test the validity of our assumptions regarding 
the internal consistency of spectral classes within an ERTS frame.