ial Park
teria of
re and
/ zones
| these
(predo-
ng and
ouristic
erns of
) of the
rtô T6 -
ntation
Il as of
nd the
of the
ed and
egrated
ntation,
] thus
gement
on with
Iti-scale
spatial
)gnomy
igratory
)92).
spatial
solution
jraphy -
iciency.
aps for
Jetation
rived at
e digital
of the
ynamics
d image
and the
|, scan
) shows
| due to
nt film
the BW-
e in red-
7 10nm),
^ aerial
such as
mits are
A-3000-
ith the
rease in
ring of
concrete
over and
A-3000-
to some
arable to
gure 2).
indcover
IR-aerial
photography is still of unbroken efficiency for multi-
thematic mapping on scales > 1:10000. Especially when
monitoring landcover and landcover change in protected
regions threatended by various forms of human impact
mapping accuracies have to meet requirements of
(cadastral) land information systems. Plots of land,
houses, roads and paths have to be detected. Detailed
change analysis of spatial as well as thematic patterns is
thus used for updating the Austrian orthophotomap
1:10000, but also for supporting updates of cadastral
maps. Environmental impact is for example documented
by the evidence of agricultural use inside primary zones
of the national park.
Map-conform, land register-accurate large scale
monitoring of confrontation zones between conservation
strategies and agricultural landuse is thus still dominated
by aerial photointerpretation, though comparative tests
using the 2-m resolution space photographs of the KVR-
1000-camera have shown positive results. It is thus
evident, that forthcoming high-resolution space imagery
like the 1m-panchromatic lkonos-1 data or the O.8m-
panchromatic data of QuickBird of Spacelmaging
(announced for 1999) combined with the corresponding
multispectral 4m-resolution bands will sufficiently meet
the requirements for environmental monitoring at a
cadastral level. Geographic information systems easily
allow the analysis of multitemporal dynamics of landuse
for compiling change detection maps (Csaplovics et
Senftner, 1991, Csaplovics et al., 1996).
High accuracies of space photointerpretation are
restricted to certain photo-object groups, which are
characterized by easily detectable spatial and/or spectral
textures or patterns and by significant contrast
differences of adjacent features. Nevertheless data
fusion of panchromatic KVR-1000- and bispectral (R,nIR)
KFA-1000-data apparently optimizes both the geometric
and spectral resolution. Multi-sensor digital spacephoto-
maps based on 2m-panchromatic and 8m-R + nIR-data
can meet the requirements for thematic interpretation of
orthophotomaps 1:10000 to a great extent. Focussing on
the separability of different landuse and landcover
patterns they even seem to be more efficient than BW-
orthophotos based on panchromatic films with limited
spectral resolution. On the other hand planimetric
accuracy is decisively decreased by the 8m-KFA-1000-
data.
Multitemporal analysis of landcover change is of urgent
need for updating spatial and non-spatial informations for
maximizing reliabilities of decisions for regional planning
and protection. High-resolution digital KVR-1000-
orthophotomaps can be merged with digitized landcover
maps derived from stereoscopic analysis of historical
aerial photography. /n concreto comparative analysis of
time series based on aerial photography of the mid-fifties
to the early eighties and KVR-1000-data of the mid-
nineties covers a period of landcover change of more
than 35 years.
4. CONCLUSIO
Landcover change detection based on remotely sensed
data analysis depends on carefully selected parameters
of spatial, spectral, radiometric and temporal resolution
(Atkinson et Curran,1997, Quattrochi et Goodchild,
1997). Multi-level monitoring of heterogeneous landscape
patterns of either urban, rural or semi-natural vegetation
needs well-defined interpretation methods (Csaplovics,
1996).
For ecological purposes the hierarchical structure of
regions implements regionalization of data based upon
the holistic understanding of ecosystem analysis
(Townshend, 1981).
Multispectral classification of raster data of satellite
sensor systems is sometimes interpreted as an objective
criterium of object detection and areal subdivision.
Nevertheless accuracies of in praxi research prove the
subjectiveness of results depending on limitations of
automation of landcover pattern recognition and texture
analysis or on the various reference levels of analysts
when choosing training samples for various steps of
mapping and classification (Edwards et al., 1998).
Updating orthophotomaps needs surveys in time intervals
of 2 years. Costs for aerial survey missions are high.
Updating of landcover and landuse classes in regions
threatened by a clash of interests of different pressure
groups requires high-resolution space orthophotomaps
based on 1m-to-2m-panchromatic imagery and 4m-to-8m-
multispectral imagery. These requirements are at the
moment (April 1998) only provided by the Russian
spaceborne photography of the KVR-1000 and KFA-
3000-cameras and - with restrictions - by the bispectral
data of the Russian KFA-1000-camera.
Costs for data acquisition are low compared to aerial
survey missions - a factor of 1:3 is realistic even when
calculating costs for a multi-sensor KFA-1000/KFA-3000-
data set.
Detecting dynamics of urban growth, cultivation and
environmental impact is bound to a temporal resolution of
topographic and thematic map information of less or
equal to a year. Traditional methods of map production
are based on revision intervals of about 10 years. Digital
space-orthophotomaps are therefore the "missing link" to
actualize topographic and thematic maps without high
financial and/or organizational outlays.
The future has to face the availability of operational
space-borne CCD-systems with spatial resolution of
0.8m to Im (P) and 3.3m to 4m. {B.GR,nR)
respectively, which have been announced by
U.S.agencies for 1997. It has to be critically remarked,
that the forthcoming new generation of high-resolution
EO-data will on the one hand definitely increase the
power of geometric and thematic environmental
monitoring in regional to local scales, but on the other
will also increase dependencies on vulnerable high
technology sensor systems as the actual situation proves
(failures of Lewis of NASA, ADEOS of NASDA, EarlyBird
of Spacelmaging in 1997!) Being aware of these
constraints it is evident, that the importance of
strengthening research on the methodology of in situ
field work and visual stereoscopic photointerpretation by
carefully using the unbroken power of human perception
and interpretation has not to be ignored during the next
years (Philipson, 1997).
In the year 2004 a total of about 78 space missions will
include about 40 missions focussing on land and ocean
monitoring. Needs for optimizing data retrieval from vast
data pools, for establishing efficient meta-data base
management systems and for decreasing the relation
between costs and specific information demands by
operationalization of thematic information extraction are
evident. Misuse of data by merging heterogeneous
Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 213
