336 
sediment in selected water bodies. Because of problems in repro 
ducing the map and units at a suitable scale, only the northwest 
quarter of the map is shown in figure 6. 
Comparing the ground data map with the ¿CRTS map showed 
good agreement between unit bounaaries. The boundaries were most 
precise where there was minimal vegetation cover on at least one of 
the compared units, for example, between unvegetated marine silts 
and other units, between unvegetated sands and other units, and 
generally Detween bedrock and other units. 
In comparing the materials, the unvegetated marine silts 
were correctly identified, the original gravel class was, in fact, 
sand, and the original bedrock class was bedrock, with or without 
a sparse vegetation cover. However, as differences in vegetation 
types in the original classification were not taken into account, 
the nature of materials underlying the vegetated areas proved dif 
ficult to determine. Most of the area classified as till was veget 
ated marine silts, some was vegetated marine sands, and a very 
small part, vegetated till. The unclassified land areas proved to 
be mostly sparsely vegetated marine sands. To the north of the 
Arrowsmith River, close to the coast, sparsely vegetated marine 
sands showed up consistently as bedrock, although "this is not re 
peated elsewhere. It is thought that this may be caused not only 
by lithologic similarities but also by similarities of surface 
texture and vegetation cover. The problem has still to be resolved 
Despite these problems, it was felt that, for the 
purposes of the evaluation, the original ERTS map was still accep 
table, and that closer attention to detail would permit reclassifi 
cation of the vegetated units in terms of dominant plant communi 
ties and unit boundaries. Since the map was intended to be used 
only in the field checking program, correct identification of the 
boundaries rather than the materials, was considered to be the 
prime requisite. 
Using the ground data from the summer, the area was re 
classified to determine whether differences in the vegetation 
supported by marine sands and marine silts would produce discrete 
classes. Using the cursor on full screen, nine units were estab 
lished: high, medium and low concentrations of sediment in water 
bodies, unvegetated. sands, unvegetated silts, sparsely vegetated 
sands, vegetated sands, vegetated silts, and bedrock. Class 
statistics showed that, with the exception of the sparsely veget 
ated sands, there was an 8$% probability that the units were 
identified correctly. 
COSTS AND BENEFITS 
Concern with evaluating a technique which might replace 
much of our traditional use of air photographs requires some com 
ments on the comparative costs and benefits of the two approaches. 
It would be unrealistic, however, to view these in terms of dollars 
and cents because of the current developmental nature of the res 
earch; rather, time will be discussed, both as it relates to 
computer use, and in terms of user time applied to the job.