A 3D city model is a digital representation of the spatial features in an urban
environment. Buildings, terrain, vegetation, water bodies, etc. all form an integral part of a 3D city model. The possibility to enrich these city models with additional application-specific information, whether new semantics or geometry, further increases their usability. However, in practice, the applications of 3D city models are mainly focused on buildings. The majority of standards available for representing 3D city models, such as IFC and CityGML, have well-defined specifications for modelling buildings, but often none for other city features. In addition, there are several other issues associated with the development and use of 3D city models of large cities, such as massive size of 3D city models, interoperability issues for 3D data from heterogeneous sources, harmonisation of different 3D standards, etc.
In this thesis, I investigate how to better model these massive and semantically
enriched 3D city models, and I focus on their use in different applications. I make five contributions. First, I explain how CityGML, the international standard for semantic 3D city modelling, is not efficient for storing massive TIN terrains, and present an improved solution to compactly store massive terrains in CityGML. Second, I describe how to model terrains at different LODs in CityGML, since the current CityGML data model lacks the specifications for modelling different terrain LODs at geometric and semantic level. Third, I explain how CityGML lacks precise specifications for modelling metadata of 3D city models and present an ISO 19115 compliant solution to add metadata. Fourth, I describe in this thesis how the development of the new standards LandInfra and InfraGML and their integration with the existing popular standards (IFC and CityGML) can
contribute to the BIM and 3D GIS interoperability and bring the two domains to a common footing. Fifth, I demonstrate my approach for the development of a harmonised semantic 3D city model based on CityGML for use in urban noise simulations. In addition, I have developed open source prototypes to help practitioners with the use of 3D city models. In this way, I also contribute to the open source community for 3D city modelling.
The thesis proposes additional research for future work. For example, since this research focuses specifically on the LODs of terrain models, it would be worthwhile to extend the research to explore the LOD concept for other urban features such as vegetation and landuse. Furthermore, LandInfra is a relatively young standard with low community support. This too requires more attention. Tools such as parsers, validators, visualisers, DBMS support, APIs, and so on are still lacking for LandInfra (and InfraGML). It would be interesting to see how the standards evolve and whether it can be applied in practice when such support is available. Interoperability of LandInfra with IFC and other standards is also an area that requires further investigation.