ıbul 2004 
ic 
ualTo 
E 
as) 
5] 
es] 
  
  
  
is(coor 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
  
  
   
  
   
   
   
  
  
luokka « 12300 
| 
     
liikenneverkot_viiva 
- luokka 
Road 
Vf iuokka e 12111 - intendedUse = 
: TA ——--—p primaryRoute 
| —————————E-—y secondaryRoute 
otherwise 
       
       
      
   
     
   
— unknown 
   
     
  
  
  
luokka 7 14111 ... 
  
Railway 
luokka = 1213 ... rr 
[rail 
  
  
  
  
  
  
<xsl:template match="nls:likenneverkot_viiva"> 
<xsl:choose> 
«xsl:when testz"nls:luokka < 12300"» 
<gmd:Road fid="Road.NLS {substring-after(@fid, '.")}"> 
<xsl:apply-templates select="nls:the_geom"/> 
<xsl.choose> 
<xsl:when test="nls:luokka = 12111 or nls:luokka = 12112"> 
<gmd:intendedUse>primaryRoute</gmd:intendedUse> 
</xsl:when> 
<xsl:otherwise> 
<gmd:intendedUse>unknown</gmd:intendedUse> 
«Ixsl:otherwise» 
</xsl:choose> 
</gmd:Road> 
</xsl:when> 
<xsl:when test="nis:luokka = 14111 or nis:luokka = 14112"> 
<gmd:Railway fid="Railway.NLS.{substring-after(@fid, '.")}"> 
</gmd:Railway> 
</xsl:when> 
<xsl:when test="nls:luokka = 12313 or nls:luokka = 12314"> 
<gmd:Trail fid="Trail. NLS.{substring-after(@fid, .")}"> 
</gmd:Trail> 
</xsl:when> 
«Ixsl:choose» 
</xsl:template> 
Figure 4. A Sample GML data transformation declaration 
As can be seen from the above examples, the same base 
technology (XSLT) can be used to transform both the initial 
data query and the resulting dataset. The approach makes it 
straightforward and logical to define the transformations. As the 
transformation declarations, represented as XML text-files, are 
read into the integration process during the query processing, 
the procedure is easy to debug and fine-tune. 
The Figure 5 shows a sample map view, received as a WMS 
response from the GiMoDig Portal Service. The map includes 
data coming from the Swedish (the left-hand side of the image) 
and from the Finnish GiMoDig WFS node (the right-hand side 
of the image). The datasets were transformed to the common 
coordinate system (ERTS89) and to the common data model 
(GiMoDig Global Schema) by the GiMoDig Integration 
Service. Subsequently, the integrated dataset has been 
transformed into an SVG image by the Portal Service, and 
finally displayed in the client device. 
181 
  
  
  
  
  
Figure 5. A Sample map display, provided by the GiMoDig 
Portal Service 
6. CONCLUSION 
The provision of adequate information services is becoming a 
challenge in the increasingly heterogeneous European Union. 
Accession countries are about to add a new dimension also to 
the puzzle of spatial data service provision in the Union. Local 
data management procedures and data models differ 
significantly from each other. It is essential that the spatial data 
may be managed according to the primary, local requirements, 
but at the same time offered for international use in compliance 
with a harmonized data model. 
Advanced, spatially sophisticated services cannot be created on 
basis of — perhaps visually harmonized — map series; they 
require availability of rich source datasets in a common data 
model. The huge amount of data involved, together with the 
active maintenance routines carried out by local organizations, 
effectively prohibits approaches based on centralized data base 
management. Online data integration, based on real-time 
transformations, seems to be the only solution for advanced 
information services with Pan-European coverage. 
Syntactic harmonization is rapidly being achieved with the 
general acceptance of the XML as the common data encoding 
mechanism. In the geospatial domain the GML vocabulary is 
equally widely recognized as the language for spatial data 
encoding on the Web. However, due to the differences in the 
national data models, the local GML Application Schemas are 
different and thus require schema transformations to be carried 
out before datasets can be integrated. 
XSLT, the generic transformation tool for XML-encoded data, 
can be applied to perform schema transformations also on 
geospatial data. The approach presented in the paper enables 
real-time schema and coordinate transformations to be carried 
out both on spatial data queries, and on the resulting GML 
datasets alike. Development of integration services becomes 
straightforward, as the same declarative mechanism can be used 
on both ways.