234
SYMPOSIUM PHOTO INTERPRETATION, DELFT 1962
Die Ergebnisse zeigten, dass bezeichnenderweise mehr Baumarten im Farbfilm richtig be
stimmt wurden, als im panchromatischen Film. Wie erwartet, lieferte der grösste Masstab die
grösste Genauigkeit, aber nur wenig Verbesserung konnte zwischen den Masstäben 1 : 1584
und 1 : 1188 erreicht werden.
Bei der Untersuchung der Farbbilder ergaben sich keine Genauigkeits-Unterschiede zwischen
den fünf Auswertern, aber wohl war dies der Fall bei der Betrachtung der panchromatischen
Abzüge. Die Genauigkeit der Interpretation belief sich auf 81% (Mittelwert) auf den Farb
bildern des grössten Masstabes. Diese vielversprechenden Ergebnisse scheinen weitere Nach
forschungen genügend zu rechtfertigen.
The U.S. Forest Service does research in the techniques of aerial photo
graphy and photo interpretation in order to increase the efficiency of a variety
of projects, including a nationwide forest resource survey and timber inven
tories for management planning on national forests. One rather fundamental
research study is discussed here, the purpose of which was to improve the
recognition of individual tree species on aerial photography. Results of this
study should be applicable not only to inventories of timber but also to other
projects as well - such as those to assess damage to forests by destructive
enemies, to appraise forest wildlife habitat and to help expedite forest admini
stration.
Procedure
Area and species involved
This study was conducted in northern Minnesota near the Canadian border
to determine the best film and scale combination for identifying tree species.
The terrain is rolling and was previously glaciated. Forests cover about 85
percent of the land area.
Fourteen important tree species were selected on 29 separate locations.
Nineteen replicates were taken on eight of the most important species and fewer
replicates on the other six species. All locations were picked in such a way that
physiographic features would not aid the interpreter in identifying individual
trees.
Collection of ground data
Trees were identified in the field on large scale (1 : 1000) black and white
stereoscopic prints made from color transparencies. (Since this study, a portable
battery-powered light table was developed for field use to locate objects on
the color transparencies.)
A detailed description was made of each tree. This included species, D.B.H.,
height, crown class, site class, and other associated features. Species were
located randomly over as many sites as possible. Only dominant, co-dominant,
and intermediate trees were included since overtopped trees are not visible
from the air.
Collection of air data
A Hulcher 70 mm camera having a 150 mm focal length was used. It can