NTORY OF TERRAUTN CLASS
ORTIONS DIRECTLY ON THE
S OF PHOTO -OR OT HER
"y qM M
Q-Hh ui
Tr IMAGERY
Dr. Simo Poso
University of Helsinki
Unioninkatu 40B
SF-OO170 Helsinki 17
Finland
Introduction
The natural resources, forests, agricultural land, etc. are
usually in good correlation with topographical features. This
correlation means that a sampling for the inventory of natural
resources must be unbias in respect of terrain classes.
The bias is not a great problem when sampling is made on
topographic maps or on other orthoprojectional medium. Problems
arise when aerial photos or other remote imagery are used for
sampling and inventory purposes. The scale of photos varies
from a point to another. This is why it has been concluded
that the photo basis is not applicable to drawing a sample
for an inventory, or it is necessary to apply an adjustment
procedure for correction of results (Loetsch and Haller 1964).
The objective of this article is to work with this problemacy
and to show that it is possible to get estimates for area-
class proportions without bias also when the sampling is made
directly on the basis of aerial photographs, or other remote
imagery.
Basic idea for sampling
For reasons of simplicity it is assumed that aerial photo-
graphy is made according to systematic sampling for inven-
tory purposes. This means that the distance from one photo
to another should be according to systematic design in respect
to field measures. This requirement is usually reached with
sufficient accuracy when an automatic timer is used in flying
at constant velocity. The systematic sampling should be re-
garded as if it had been taken with random start.