488 
Data Reduction and Analyses 
The photointerpretation test was designed to determine: (1) if gopher 
sign could be interpreted from the aerial photographs, (2) the relationships 
between photo and ground counts by photo scale, and (3) the effect of herbaceous 
and shrubby vegetation height on gopher sign-counts by photointerpretation. Three 
scene we] 
from sim: 
of old g< 
gopher m< 
photointerpreters were trained to learn the apparent photo-image differences 
between new and old gopher sign, and then asked to count the number of new 
signs in each subplot. These results, compared to ground truth, provided the 
relationship between actual and interpreted counts, omission errors (mounds 
present but not identified) and commission errors (objects mistaken for mounds). 
In addition, correlation and analysis of variance were used to compare 
similarity among interpreters, effects of kind of vegetation on photointerpreted 
sign-counts, and effects of treatment on photointerpreted sign-count. 
RESULTS AND DISCUSSION 
interprel 
differen< 
of fresh 
Although 
mound coi 
lack of i 
photo te< 
degree oJ 
as well i 
the phot< 
The 1:1200 scale photographs were unsuitable for the objectives of 
this work. The earth mounds could not be identified with any confidence. 
Earth plugs could be identified with confidence in the 1:600 scale 
photographs only when the coded ground markers were interpreted in the 
photographs. Their small size, the absence of relief in relation to the soil 
surface, and vegetation shadows contributed to the difficulty in identifying 
results < 
treatmenl 
may have 
in the ii 
were obvj 
to be ex< 
the earth plugs. Therefore all relationships developed between ground-based 
estimates of gopher populations and photointerpreted estimates were based on 
earth mound counts on 1:600 scale photographs. 
The earth mounds were discernible in the photographs as variously 
shaped patches of bare soil. New mounds were distinguishable from old mounds, 
or other bare soil, by using color, textural appearance, shape, and relative 
height. The mounds, viewed in the photographs, were mostly concentric, although 
were neg] 
photointi 
done for 
and the i 
equation 
some were horseshoe-shaped, and occurred mostly singularly or sometimes in 
aggregates of two, three, or four. New mounds in the color infrared film 
appeared as a very light greenish-blue; the old mounds were a medium bluish-green. 
A 
where: Y 
R 
New mounds in the normal color film were pale yellow; the old mounds were light 
gray-yellow brown. In addition to color discrimination between new and old 
mounds, the new mounds had a characteristic fluffy, rough-appearing texture, 
M 
while the older mounds had a more compact, smooth looking textural image. 
Omission and commission errors were made by the three interpreters 
evaluating the photographs for counting earth mounds (Table 1). Omission errors 
were probably due to a variety of factors. Small mounds creating little surface 
relief were more difficult to discriminate than larger mounds. Some mounds 
were partially obscured by plant foliage or even shadows of nearby plants although 
subsequent analyses showed that apparent plant height had no effect on interpretation 
(1966) gi 
populatic 
infrared 
as the es 
developec 
of earth 
among plots. Larger mounds or mound aggregates were sometimes overlooked because 
they appeared to be old signs,other bare soil, or very lightly foliated low 
growing shrubs with numerous exposed stems. 
among inc 
comparing 
increasec 
photointc 
populatic 
between 1