GVII-76 
PHOTOGRAMMETRIC ENGINEERING 
instances out of 77 indicated differences of as much as 100 feet. Thus it appears 
that boundary lines can be located with more precision than the map units can 
be designated by means of photo interpretation. This certainly indicates the 
need for more field checking of the latter than the former. 
Discussion 
It is the opinion of the authors that most studies of the earth’s surface and 
vegetation should be evaluated quantitatively by the general procedure out 
lined above. Photo interpretation has not reached, and for this type of problem 
never will reach, the point where we can expect perfect results from the inter 
preters. Field sampling provides a check on the accuracy of the photo interpreta 
tion, both type designation and boundary line location. If the field checker 
would take the time, less than a minute per sampling location, to maintain a 
record of photo interpretation classification/field classification, he can, with 
very little effort, collect the data necessary to make a quantitative evaluation 
such as is outlined in this paper. 
Once quantitative methods are established, it should become normal pro 
cedure to include estimate of error in planning so that the standards of photo 
interpretation can be established with sufficient field checking to yield results 
within established limits. It is understood that the standards will vary with 
the purpose for which the map is made, e.g., the precision of a reconnaissance 
survey will be of a lower level than that of a strip survey for highway location. 
Quantitative evaluation provides each photo interpreter with a sound basis 
for measuring his own ability, his rate of progress and a means of comparing 
his work with that of others working in the same field. 
The accumulation of the results of a number of photo interpretation studies 
that have been quantitatively evaluated will make possible comparing varia 
tions in interpretation due to such controllable variables as scale and date of 
photography. For example, other things being equal, if the over-all accuracy of 
interpretation with photos at a scale of 1:10,000 is 75 per cent, and if interpreta 
tion has an accuracy of 70 per cent at a scale of 1:20,000, then perhaps the 
additional gain of 5 per cent in accuracy does not justify the additional expense 
in procuring large scale coverage. 
The authors have heard experienced photo interpreters make the following 
statements: (a) “1:20,000 is the optimum scale for studies in my field’’ (b) 
“Winter photography for my purposes is useless as I must have photos taken in 
the spring before the leaves are out.” (c) “Photos taken in April are much better 
than those taken in June for soils interpretation.” (d) “I can get sufficient ac 
curacy from two inch to the mile photos.” (e) Infrared photos taken in mid 
summer at a scale of 1:15,840 are the very best for my purposes.” In practically 
all instances those who make such statements are voicing opinions which may 
or may not be valid. It is not a difficult matter to field check most photo inter 
pretation studies so that data will be available for quantitative evaluation to 
determine the validity of such opinions. 
Summary and Conclusions 
The widespread use of photo interpretation as the primary means of prepar 
ing forestry, soils engineering, geology, geography and military terrain analysis 
maps generates the requirement that the finished product be quantitatively 
evaluated since perfection cannot be expected. Methods of quantitatively evalu 
ating the accuracy of map unit designations, and the accuracy of boundary line 
location between map units as determined by photo interpretation, are pre 
sented. An engineering soils map prepared by photo interpretation provided 
data to illustrate these methods. It is the opinion of the authors that quantita-