3D city modeling and rendering has undergone a rapid technical development during the last 10 years. The two most representative and successful modeling techniques for the time being are CityGML and GoogleEarth KML. CityGML can model four essential aspects of 3D cities: geometry, appearance, topology and semantics at five Levels of Detail. The current version of CityGML, however, is not yet able to model the temporal information describing the individual objects at various LODs. The newest version of GoogleEarth KML has included the specification of time spans for a region within which individual 3D objects are constructed or destructed at different time points within the span. Thus, the chronological sequence of events can be visualized. However, these 3D objects switch only between two temporal states – appear or disappear. Moreover, each 3D object, regardless of its complexity, is modeled as an inseparable unit without texture, possibly for the purpose of keeping the fast rendering speed. Since GoogleEarch KML does not offer the possibility to model 3D objects at different LODs, object components such as building roofs, walls and grounds are inaccessible to the user.
Numerous graphics formats, CAD platforms or viewers have been developed and constantly refined to support the visualization and exploration of virtual 3D scenes modeled by CityGML or GoogleEarth KML in a seamless and nearly real-time manner. Some examples are VRML, Aristoteles, LandXplorer, GoogleEarth viewer as well as their PlugIns. However, the interactive functions and query possibilities of these systems are largely limited by the restrictions in the underlying data models.
Furthermore, depending on their nature and their surroundings, different objects may change at differ-ent speeds following different patterns. An event, i.e., something of significance that happens during a given time span, may occur to different objects at different time points. For this reason, using time slices by freezing all objects to the same time points at predefined intervals to simulate the temporal behavior of a city model is inherently conjectural and uncertain. Likewise, averaging the temporal in-formation to a very coarse level does not allow the users to gain an insight into the interrelationships among objects and object components.
The objective of this research project is the development of general concepts and methods to enable a precise description, querying, and visualization of the temporal behavior of a 3D city model. In the scope of the project the following aspects should be addressed:
- Data structures to handle temporal information associated with 3D city objects at fine granularity levels, and to support spatial queries and queries of events, processes as well as topological tem-poral relationships
- Mechanisms for compressing spatio-temporal information for fast rendering
- Mechanisms for user interactions that go beyond virtual walks or flights through the static 3D models, and allow the immersive access to and modification of dynamic behavior of complex objects, simple objects, and object components
- Augmentations of video streams with computer-generated imagery of spatio-temporal information.
The necessary extensions will be conceptualized on the basis of the currently available Open-GIS standards CityGML and the newest release of the dominating GoogleEarth KML. Meaningful user interactions and queries of temporal information will be anticipated for the selected application scenarios of the extended 3D city model. Furthermore, data structures and compression methods for the purposes of efficient retrieval and fast rendering of spatiotemporal city objects will be conceptualized and tested.