International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
Germany: complete, if 'permanently water leading, 
else at a length > 500 m 
Denmark: Length > 50 m, or if they are important for 
the continuity of the hydrography network. 
Finland: Width more than 2 m, exception in forest or 
If they are important for the continuity of the 
hydrography network. 
Sweden: Length > 250 m 
Global Schema: the attributes 'permanently water 
leading" and 'important for the continuity of the 
hydrography network' are not included into the 
selection. Concerning the length and width of the 
watercourse the least common denominator is defined. 
So all watercourses with more than 250 m length or 
more than 2 m width are recorded. 
Attribute types and values of ERM and/or the FACC are only 
included to the GiMoDig Global Schema if they are supported 
by the majority of the countries, or if they appear to be of great 
importance to the GiMoDig service. 
S. VERIFICATION OF THE GLOBAL SCHEMA BY 
MEANS OF TEST DATA 
The process steps above yielded the preliminary Global 
Schema, preliminary in the sense that it is result of a desktop 
study but not yet tested on sample data from the NMAs. 
For a robust definition, the preliminary Global Schema is tested 
on sample data. Two test areas across the borders are defined 
where the differences between national data could be examined. 
One of the test areas is located in the Tornio/Haparanda area (1) 
at the Swedish-Finnish border with a size of 10 x 10 km, and 
the other at the Danish-German border from Hojer/Sylt to 
Augustenborg/Kappeln with a size of about 100 x 30 km (2). 
Figure 1 shows the location of the test areas. 
  
  
  
  
  
  
  
Figure 1: Location of the test areas 
The analysis of test data showed that some elements of the 
Global Schema should be adapted to the situation, and that the 
content of the national core topographic databases is not 
sufficient in some cases and therefore data has to be acquired 
from additional sources. In the following some of these issues 
are discussed. 
5.1 Missing feature types 
As already mentioned in chapter 4.3 the initial goal of using the 
least common denominator of national databases for the Global 
Schema could not be realized because important feature types 
would have been excluded. The actual Global Schema contains 
some elements that have to be acquired from sources other than 
the core topographic datasets. 
In Germany the ATKIS specification considers buildings to be 
derived from cadastre databases by means of automatic 
generalisation, but these topographic objects will not be 
available before finalisation of ATKIS in the year 2006. For 
GiMoDig they have to be acquired from the cadastre databases. 
The 'Contour Line' is an optional feature type in ATKIS, it is 
provided by some but not all German Landesvermessungsámter. 
For the GiMoDig application they are computed from the 
Digital Terrain Model (DTM). 
The Danish core topographic database does not include the 
feature type 'Ocean'. It has to be created from 'Coastline' data 
and/or 'Administrative Units'. 'Grassland' and 'Cropland' are 
both not available in the TOPIODK but are added from the 
Danish Ministry of the Environment. 
Figure 2 shows the major deficits of the Danish and the German 
databases in an example of the border area near the German city 
Flensburg. It illustrates the lack of individual buildings and 
contour lines in the German ATKIS dataset and the lack of 
grassland and cropland in Danish TOPIODK dataset. Other 
elements of land cover such as built-up area, forest and lake are 
present in both datasets. 
  
  
  
  
  
  
© KMS/BKG 0 500 1.000 1.500 2.000 2500 m 
LLL N 1 Li} 
  
  
  
Figure 2: Differences between Germany and Denmark 
In the case of Sweden, only the major buildings are recorded. 
So a complete record of buildings from the Swedish 
municipalities is required for the GiMoDig prototype. 
Furthermore, 'Grassland' and 'Forest' are so far not included in 
the Swedish dataset. Finland lacks data for 'Built-up Area' and 
polygon feature type 'Forest'. 
Inter 
2 
Un 
In Fi 
large 
does 
the r 
land 
wate! 
(whit 
distir 
inter] 
meth 
syste 
mL 
  
  
  
  
83] 
Some 
or on! 
of the 
to we 
classi 
Ocear 
'Ditch 
from 
instan 
type 
appro: 
(Wate 
visual 
since 
GIS aj 
to diff 
Denm 
"Railw 
record 
Anoth 
This f 
'Amus 
attribu 
type o 
The ci 
or loc 
officia 
by the