thods
y or by using
es of the
a planimeter
Ocedure can
all parcels
zerland. The
onsuming as
sample units.
raverses) or
sample) or
t, point and
point samples
c samples
random
manent
e updated.
s in land use
. land use
trative
wputar maps
mation, e.g.
' slopes,
e grid.
on the total
yattern. If
.S an error
.O be
lan be
]
le points
Fractional
ontage of
1 Square error
stigation
In case the percentage error referring to the phenomena under
investigation should be of interest (relative error), the
following formula can be used:
dlls pn [2]
ny is the number of sampling points within a certain
phenomena. These formulas should only be appiied if the land
use units under consideration have small surfaces compared
with the width of the sample grid and are randomly
distributed. In case adjoining sample points are likely to
fall within the same land use parcel, the error will be
overestimated. This will happen to non-scattered land use
catecories with extensive surfaces. In this case formula 2
with the exponent 2/3 instead of 1/2 will allow a better
error estimation (8, 9).
These formulas can also be used as a basis for an estimation
of the number of sample points needed for obtaining a certain
accuracy.
3. Different possibilities for sampling on aerial photographs
In case land use data have to be collected by sampling on
aerial photographs, the geometrical characteristics of the
pictures (central perspective view) have to be taken in
consideration. Variations in scale and displacements due to
elevation differences would lead to systematic errors if a
regular sample grid would be placed directly on the photos.
There are several possibilities for eliminating these errors:
- The air photos serve only as a source for land use informa-
tion but not as a basis for sampling. The land use
boundaries are first transferred to a topographic map which
is then superimposed by the sample grid.
- Instead of transferring the land use information on a map,
the sample grid can be transferred to the photos. The land
use occuring under each dot can be interpreted directly in
the photo.
- Sampling can be done with the aid of a stereoplotter. In
this case preference should be given to line sampling
(traverses) instead of point sampling.
- The air photos can be used for producing orthophotos. Since
these are free from variations in scale they can be directly
used as a basis for sampling.