City GML, providing an open standard for 3D city models

Services such as the very popular Google Earth and Virtual Earth (both covered in this blog), while pushing the 3D urban modelling field to its limits in more ways than one also carry a lot of limitations. One of the most important ones is that these web services largely provide only graphic or geometric models, neglecting the semantic and topological aspects of the buildings and terrain being modeled.

These 3D models can only be used for visualization purposes but not, in most cases, for thematic queries, analytical tasks or spatial data mining. Another problem with these and other city 3D modeling efforts is lack of interoperability. Web-aligned open standard geometry models for computer graphics (e.g., X3D and geospatial technologies (Geography Markup Language 3 (GML3)) are available but still relatively new and not yet widely adopted. Though their common XML foundation provides the elements needed for convergence, this has not been addressed (not yet anyway). Moreover, in those cases where semantic and topological aspects are considered, there are no widely used standard application schemas, though such schemas are under development, drawing on a considerable body of previous data modeling work. Without a common definition of the basic entities, attributes and relations that can be shared over different applications, those who develop and own the models find it difficult to share and maintain them.

Therefore, for this post, I will provide with a few details on an information model that tackles all these issues and more, called City GML. CityGML, began as a research project and has been developed since 2002 by the members of the Special Interest Group 3D (SIG 3D) of the initiative Geodata Infrastructure North-Rhine Westphalia (GDI NRW) in Germany, and is a common information model for representing 3D urban objects. It defines classes and relations for the most relevant topographic objects in cities and regional models with respect to their geometric, topological, semantic and appearance properties. It should be noted that "City" is broadly defined to include not just built structures, but also elevation, vegetation, water bodies, sidewalk furniture and more. Included are generalization hierarchies between thematic classes, aggregations, relationships between objects and spatial properties.These thematic information types go beyond graphic exchange formats and allow users to employ virtual 3D city models for sophisticated analysis tasks in different application domains such as simulation, urban data mining, facilities management, decision support and thematic inquiries.

So, in short, some of City GML's features include:

- Geospatial information model for urban landscapes based on the ISO 191xx family
- GML3 representation of 3D Geometries, based on the ISO 19107 model
- Texture and material representation of object surfaces
- Taxonomies and aggregations
- Digital Terrain Models as a combination of (including nested) triangulated irregular networks (TINs), regular rasters, break and skeleton lines, mass points
- Sites (currently buildings; bridges and tunnels tbd. in the future)
- Vegetation (areas, volumes, and solitary objects with vegetation classification)
- Water bodies (volumes, surfaces)
- Transportation facilities (both graph structures and 3D surface data)
- City furniture
- Generic City objects and attributes
- User definable (recursive) grouping
- Multiscale model with 5 well-defined consecutive Levels of Detail (LOD):
a) LOD 0 : Regional, landscape
b) LOD 1 : City, region
c) LOD 2 : City districts, projects
d) LOD 3 : Architectural models (outside), landmarks
e) LOD 4 : Architectural models (interior)
- Multiple representations in different LODs simultaneously; generalization relations
- Optional topological connections between feature (sub)geometries
- Application Domain Extensions (ADE): specific "hooks" in the CityGML schema allow to define application specific extensions, for example for noise pollution simulation, or to augment CityGML by properties of the new National Building Information Model Standard (NBIMS) in the US

To provide (only some) names of bodies working on and supporting CityGML: municipalities of Berlin, Hamburg, Cologne, Düsseldorf, Recklinghausen, Leverkusen; British Ordnance Survey, State Mapping Agency of North-Rhine Westphalia; companies T-Mobile, Rheinmetall Defence Electronics, Snowflake, CPA Geo-Information, GIStec, 3D Geo; research institutions: Universities of Bonn, Potsdam, Dortmund, Applied Sciences Stuttgart; Research Centre Karlsruhe and Fraunhofer Institute for Graphics Research.

For anyone with even a passing interest in urban modelling and all associated challenges arising from producing more than just 3D geometry for urban spaces, more details on City GML can be found at http://www.citygml.org/