The International Archives of thq Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B6b. Beijing 2008 
166 
Industry Foundation Classes (IFC): is a standard for 
representing and exchanging Building Information Models 
(BIM), which is supported by most CAD software. IFC (or 
ifcXML, based on XML structure) standard developed by the 
International Alliance for Interoperability (IAI) and became an 
ISO standard. IFC objects model provides rich semantic 
elements in the AEC/FM domain, for example, walls, roofs, 
windows and stairs. Nevertheless, it does not provide enough 
city objects like streets, water bodies and vegetations, so it is 
not appropriate for building complex city models. Moreover the 
spatial and semantic coherence can not be automatically 
ensured in some IFC models (Stadler and Kolbe, 2007) 
CityGML: will probably soon become the OGC standard for 
representing 3D city objects in a spatio-semantic coherence data 
model (Kolbe et al., 2005; OGC, 2007; Stadler and Kolbe, 
2007). That means “geometrical objects “know” what they are. 
Semantic entities “know” where they are and what their spatial 
extents are”. 
The semantic model of CityGML based on the IFC model 
consists of class definitions for the most important features 
within 3D city models, including buildings, DTMs, water 
bodies, transportation, vegetation, and city furniture. It does not 
contain a whole set of IFC objects, but it can be easily extended 
according to users’ own necessary 3D semantic objects. All 
CityGML classes are derived from the basic class ‘Feature’, 
defined in ISO 19109 and GML3 for the representation of 
spatial objects and their aggregations. 
The geometric model of CityGML is equivalent to a subset of 
GML3 geometry packages, which is based on the standard ISO 
19107 “Spatial Schema”, representing 3D geometry according 
to the well-known Boundary Representation. Therewith, 
CityGML is regarded as a specific GML application schema. 
Moreover, CityGML supports the concept of Levels of Detail 
(LoD) up to five discrete levels (LoD0-LoD4), ranging from 
outdoor wide areas to indoor architectural models like the 
furniture and interior installations. 
Thus, users can define 3D city models at various degrees of 
complexity with respect to geometry as well as semantics. For 
example, for merely outdoor LBS applications, lower LoDs are 
adopted; otherwise higher LoDs are employed to represent 
detailed indoor 3D objects. 
Summary: Comparison of other existing 3D data formats is 
described in (Kolbe et al., 2005; Stadler and Kolbe, 2007). 
CityGML as a special case of GML format, representing virtual 
3D city objects with semantic and spatial information in 
different LoDs, is a very good candidate for 3D LBS 
applications. 
2.5 Standardized Geospatial Web Services 
Geospatial web services are indispensable parts of LBS. The 
mobile user sends its current location to a particular geospatial 
web service, and the server is able to response the geospatial 
data in many formats, such as raster maps or vector XML-based 
documents. We list some notable geospatial web services that 
often used in LBS applications in the following. 
Web Map Server (WMS): is an OGC standard for producing 
maps of spatially referenced data dynamically from geographic 
information over HTTR The WMS responses can be pictorial 
formats like JPEG PNQ GIF or vector based graphic elements 
like SVG. 
Web Feature Server (WFS): is an OGC standard for 
requesting and serving vector geospatial data over HTTP. The 
WFS responses contain the data in the GML format. 
Furthermore, the transactional WFS (WFS-T) standard supports 
transactions (such as insert, update and delete) which allow the 
client to modify data on the remote server. 
Web Terrain Service (WTS): is an OGC standard similar to 
(and builds on) WMS but provides a static 3D rendered image 
of a dataset instead of the data itself to the client. 
Web 3D Service (W3DS): is developed as an extension of 
WMS/WTS by OGC and offers additionally the possibility to 
visualize 3D scene graphs. In contrast WMS/WTS only 
provides the representation of static views as bitmaps. The 
W3DS merges different types (layers) of 3D data in one scene 
graph and outputs it as the default VRML97 format, GeoVRML 
and X3D are also suggested. Then the scene graphs are 
rendered and interacted on the client side. 
Building Information Models Web Perspective View Service 
(BIM-WPVS): is proposed to integrate and visualize both BIM 
and GIS data by (Hagedom and Dollner, 2007). It is one of the 
high level high-level geoinformation services having several 
distinct features: enhancing geoinformation, provision of 
business functionality, integration of complex geoinformation, 
provision of high-quality geovisualization, and support of user 
interaction and context-awareness. 
Summary: In fact, BIM-WPVS method can be regarded as a 
special case of mixing other different web services, because it 
can be built on top of the WFS when data-oriented, the W3DS 
when Scenegraph-oriented or WMS/WTS/WPVS when 
visualization-oriented. WMS and WTS only sever 2D bitmaps 
to the client, and thus are mainly limited to 2D LBS. The W3DS 
is able to offer real 3D scene graphs to the client for different 
renderings, which is reasonable to 3D LBS. However, a scene 
graph like the VRML format only contains geometric 
information but little semantic characteristics and relations of 
geospatial objects. Thus the W3DS could be an appropriate 
solution for the 3D visualisation in many 3D LBS, but is not 
satisfying for highly required semantic applications. 
The WFS delivers the geospatial data itself to the client as the 
GML format, so it severs necessary semantic contents. For 
example, users can request geospatial objects within a bounding 
box, or with a specific identifier, or users even can perform 
complex spatial operators on the server such as getting buffer 
areas within a certain area. CityGML being one of the GML 
application schemas therewith can be used as the output format 
of the WFS in order to provide spatio-semantic coherence 
information for more complex 3D LBS applications, such as a 
disaster management system in built-up areas demanding not 
only 3D graph scenes but also 3D spatial queries. 
3. A PROTOTYPE OF 3D LBS 
This paragraph discusses a method about how to set up spatio- 
semantic 3D LBS with BIM and GIS data. A 3D LBS prototype 
emphasizing on data standards and geospatial web services is 
explained. Other components of LBS are not studied in this 
paper. The method used in this prototype can be considered as