2.3.5 Geodetic systems. The system is based on the GRS 1980
ellipsoid with parameter values almost similar to WGS 84, the
ellipsoid used for GPS positioning. The Soviet Union system
uses the Krassovsky ellipsoid. First order points in the net are
connected and adjusted together with points from Finland,
Poland and Germany.
2.3.6 Map projections. The Baltic projection parameters are
rather similar to the parameters used in the standardized
Transverse Mercator Projection, Universal Transverse
Projection, UTM. Because of the longitude location of the
countries it was not convenient to use the standard meridian
suggested in UTM. In order to minimize distortions in the
central parts it was decided that the central meridian should be
centered in the Baltic states. All other parameters are similar to
UTM parameters.
2.3.7 Transformation. Since the old Soviet Union geodetic
datum is unknown, some of the points in the old geodetical net
have been measured by GPS. Transformation parameters are
then calculated between the two systems using well defined
positions in the Soviet system and the same points measured
and transformed to the Baltic system using a 7-parameter
transform, including 3-dimensional translation, 3-dimensional
rotations and a scale factor.
X AX, WAREZ =ryf.X
Y =| AY, [+m*{-rz 1 rmx |Y (1)
Z Baltic AZ, ry TX 1 SovietUnion
Calculation of these transformation parameters has made it
possible to use information from the old Soviet Union
topographic maps as reference material in the projects.
Hydrographic information is an example of a feature that has
been selected and digitized from the old maps.
2.3.8 Image Processing. By using image enhancement filtering
the interpretation results can be optimized. In this case two
different methods are combined and included in the production
process. The standard corrected images are DEHAZE filtered
which strengthens the edges between light and dark features in
the images. DODGING adjusts global contrast differences. In
practice this means that dark areas, like forests, will become
somewhat lighter, and light areas will appear somewhat darker.
Finally, when combining two or more scenes in a mosaic, it is
important to correlate them spectrally, to increase the
interpretability and to present a more homogenous product for
the image background without seam strips between images.
2.4 Products delivered by SSC
For a combination of technical and cost reasons it was decided
to produce the panchromatic information on a scale 1:50,000
and the multispectral on 1:100,000. According to SSC
Satellitbild’s production routines the corresponding spatial
resolution of the digital data is 10 and 20 metres respectively.
The panchromatic information was delivered as map sheets
related to the new Baltic map sheet index system and the
multispectral images as precision corrected scenes.
2.4.1 Digital Products. The digital data was delivered on CD-
ROM with a storage capacity of 600 Mbytes. As the image data
includes merely image information, the data describing map
projection, ellipsoid, resolution, etc. has to be specified. SSC -
Satellitbild has developed a GIS format including description
files on the size and layout of the image raster file, on the map
projection, map sheet and on the different sources used for
producing the image. They are presented separately from the
image data, in an ASCII file, because different image
processing software use these descriptive data in different ways,
making it possible to use the information independent of
software.
2.4.2 Photographic products. Satellite data was presented in
two types of photographic products, paper prints and
transparent films. Film is a more stable material than
photographic paper with regard to shrinkage and expansion
from variations in humidity and temperature. On the other hand,
unlike photo-paper transparent films require light-tables for
their interpretation and cannot, therefore, be used in the field.
Panchromatic (Black and White data) on a scale of 1:50,000
was delivered as both paper prints and transparent films and
multispectral data on a scale of 1:100,000 as paper prints, with
a plastic base giving a stability similar to that of transparent
films.
Furthermore the shrinkage and expansion errors, of about 2-3
mm on a print of 50 cm, could easily be minimized with a affin
transformation in 2 dimensions.
2.5 Interpretation
Information possible to interpretate from remotely sensed
registrations is in general of the "land cover" category, which is
an interpretation of vegetation, water, open areas etc,
Traditional maps usually present "land use" features implying
that some evaluation has been added during the interpretation of
the images to convert from land cover to land use information.
2.5.1 Information sources. Existing digital data bases,
digitized from topographic maps on a scale of 1:50,000 contain
coastlines
lakes
watercourses
wetlands
build-up areas
road network (three classes)
railways
administrative boundaries
peat extraction fields
contour lines with 20 m equidistance
name data base (being developed)
All these elements have to be transformed from the old datum
to the new Baltic geodetic datum. The data from the old maps
serves mainly as a complement to the satellite image data
during the interpretation.
Satellite images. The area covered by each panchromatic
Satellite Ortho Photo Map is 25*25 km. Most of the SOMs
contain information from two or more satellite scenes,
optimally mosaicked for the base mapping purpose. Due to the
oblique registrations of the satellite data it has been necessary to
use a Digital Terrain Model to eliminate displacement errors
caused by terrain variations. The multispectral data
geometrically processed in the same way is particularly suitable
for area interpretations, such as land cover information.
Other sources. Available existing Maps on a scale of 1:10,000 -
1:100,000, aerial photographs and field studies in combination
with local knowledge and competence concerning the land and
its resources is in general the best ground truth data.
Investigations of name data was a highly time consuming part
of the mapping, due to the 50 year mapping period with
Russian names spelled with Cyrillic letters.
2.5.2 Legend. Before starting the actual interpretation work, a
legend has to be established containing all types of objects to be
interpreted. Forest for example have to be precisely defined by
crown cover density before the representation in the satellite
imagery was investigated.
Since areas, that are not always homogeneous it is important to
clearly define the boundaries and priorities between different
classes to minimize possible ambiguity during the
interpretation. This was done between all pairs of classes.
2.5.3 Methods. Three alternative production methods have t0
be considered depending on the quality of the sources.
456
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
Are
dim
othe
for |
—
All:
Line
info
achi
coul
of t
inve
Sym
sour
the s
2.5.4
sepa
a ro:
two
wheı
Two
CONV
e
Lus d obe . I
Thes
powe
digit.
satel]
visua
2.61
One
digit:
not p
tradit
create
custo
intere
The
"fittir
it po
origir
black
2.6.1
infort
impo:
soluti