t of
rence
ciples
Earth
ted in
eation
nages,
d out
ted on
ew of
ple of
visual
9 the
luding
emote
reters,
having
ing up
quality
mages,
0.000,
RS-1C
ipment
carried
he full
d by 6
rencies
onitor,
etation
5 were
)n was
arified.
etation
s been
nto the
ted. As
STERS
both in
ent of
quality
> Zuglô
merged
) 2 m
resolution) panchromatic satellite images in order to improve
the geometrical information content. For delineating census
enumeration districts the same images were used.
FOMI provided continuous consultancy during the project
period for city planning and area development experts in order
to make them acquainted with the new technology, i.e. with
information possibilities supplied by satellite images. At the
same time, this helped FOMI to be informed on the demands of
a new circle of users, the enlargement of the practical
application of remote sensing procedures and, last but not least,
by acquiring useful information.
2.3 General description of the satellite imagery used
During the project, delineation of the urban agglomeration and
the test of delineating enumeration districts defined two
requirements against geometric resolution of the satellite
images. At the very beginning of the project original resolution
of about 20 rheters of the satellite images seemed to be
satisfactory, but later a need for a higher resolution (of some
meters) raised. At first, application of the images made by the
French satellite SPOT was planned. Results of a test for photo
interpretation showed a non-ambiguous separation of the
vegetation. As the reason of this phenomenon the lack of the
middle infrared band was established.
The aim of the next investigation was to find the type of
satellite images containing the middle infrared band necessary
to make the appropriate photo interpretation. The Landsat TM
and the Indian IRS-1C LISS III sensors have the required
spectral resolution; the difference between them is in their
spatial resolution. Size of pixels of all Landsat TM bands is 30
x 30 meters. Spatial resolution of the first 3 bands of the IRS-
IC is 23.8x 23.8 meters and that of band 4, middle infrared
band is 70 x 70 meters.
For the IV level interpretation of Zugló we used at first a
combination of IRS-1C multispectral and PAN images with 24
m and 6 m resolution, but it didn't give the required geometric
information at the scale used (1:10 000), therefore the necessary
resolution was achieved using the KVR-1000 2 m resolution
image, merged with the IRS-1C multispectral image.
e land use classification for the whole area of Budapest
according to CLUSTERS level II by photo-interpretation of
colour photo hardcopy of the IRS-C LISS satellite data on a
1:50.000 scale, scanning the transparencies and
vectorization of the result
e land use classification for the whole area of the Zugló
district according to CLUSTERS level IV by photo-
interpretation of paper colour hardcopy of the IRS-C LISS
in conjunction with KVR-1000 satellite imagery on a
1:10.000 scale, scanning the transparencies and
vectorization of the result
2.4 Pre-processing
During the transformation performed in the first phase of the
project it was found that transformation of band 4 should be
performed separately as it was not fitted to the bands 1, 2 and 3.
To improve the geometric transformation we looked for a
procedure which doesn't change the spectral information or
changes it only in a minimal degree. After due consideration the
HPF (High Pass Filter) method was selected (Chavez 1991). In
case of the IRS-1C, gradient of the Band 2 with a spatial
resolution of 23.8 meters was added to the Band 4 with the
resolution of 70 meters..
2.5 Supplementary data used for processing and
classification procedures
1. Map of the Budapest agglomeration providing
information on the area
2. List of 78 localities belonging to the Budapest
agglomeration in conformity with the most recent
relevant decision. FOMI obtained the digitized
administrative boundaries of these localities from the Pest
county cadastral office, in order to get an accurate
delimitation. Pest county cadastral office provided this
database containing administrative boundaries of the
localities digitized on the highly accurate cadastral base
maps in the scale 1:4000. This database consists of about
14 000 co-ordinates of boundary points. Therefore the
delimitation of the agglomeration on the satellite images
was carried out with high accuracy.
3. FOMI's Department of Spatial Information provided us
with the highly precise digitized database containing
district boundaries for the total area of Budapest.
4. Topographic maps at scales 1:10.000, 1:50.000 and
1:100.000 were also used.
5. 1:12 000 to 1:30 000 scale colour infrared and B&W
aerial photos (images taken in 1990-1996)
6. FOMI's Department of Spatial Information also
provided us with digitized topographic data prepared on
the basis of topographic maps at a scale 1:100.000, in
EOV co-ordinate system.
7. FOMI's Department of Spatial Information also
provided us with digitized topographic data prepared on
the basis of topographic maps at a scale 1:100.000, in
EOV co-ordinate system.
8. City planning maps published by Zugló Municipality.
Zugló municipality provided us with the town-planning
scheme of the district at a scale 1:10.000, also available in
digitized form, and with the digital cadastral co-ordinates of
the district. This co-ordinates were transformed into the
merged satellite photomap with high accuracy
Table 1
Satellite data used and geometric correction parameters
Type of satellite data Date Number of Spatial Number of RMS
bands Resolution GCPs
IRS-1C LISS-III 10.08. 1996 4 1,2,3:23.8m 109 20.8 m
4:70m
IRS-1C PAN 30.05.1996 1 5.6m 108 2.7m
SPOT XS 10.08.1996 3 20 m 77 11.6m
SPOT-PAN 18.04.1996 1 10 m 77 6.8 m
KVR-1000 05.1992 1 2.7m 23 24m
Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 79
