Design of an integrated 3 D information model

3D city models and traditional 3D GIS have so far neglected underground stru ctures and features both in theory and practice. Man y applications are looking for such integrated approaches, however. In this paper we define a them atic semantic framework for integrated modeling of geographic 3D data, combining man made nd natural features above and below the earth surface. Our approach extends the semanti c concept and the information model created for the exchange format CityGML by dividing feature s close to the earth surface into: above, in and below the earth surface. Further, the concept of implici t storage and terrain intersection objects is extended to connect features to a TIN base d earth surface model. All features, intersecting or touching the earth surface TIN, create a pro jected geometry that is integrated in the DTM geometries. A suggestion for a possible database im plementation of the extended information model is also addressed. lation, radio wave coverage or ground water simulat ions in one representation we argue that there is need for an integrated representation mode l describing the geometric representation of 3D physical objects in 3D space and thematic semant ics describing what geometry is representing what feature in the real world. We present a fr amework to integrate subsurface features in an existing concept and thematic semantics for the top -level objects in our information model. 2 THEMATIC SEMANTIC REPRESENTATIONS For 3D city models few thematic semantic models exi t. A common understanding is that buildings and terrain objects are the most important fea tur s to describe in a 3D city model (Billen & Zlatanova 2003) and Köninger & Bartel (1998) founde d the following top-level object classes: buildings, streets, green areas, public areas and t errain surface while Dahany (1997) suggests only three groups: terrain, vegetation and built fo rm. Within the geo information field several subdivisions of space into objects on a national le vel can be found, and also some international initiatives. The Dutch harmonized base model called NEN3610 based on the ISO/TC 211 is one example of a recent specification of top-level obje cts in urban space (Quak & de Vries 2006). Figure 1. First level of the hierarchy in NEN3610 w ithout details (Quak & deVries 2006) The NEN3610 model constitutes a basis for exchange formats within the Netherlands defining 11 base elements (NEN 2005). Within the EU project INSPIRE a set of 21 spatial data themes has been defined. The themes cover infrastructural units i.e. buildings, terrain models and transport networks as well as administrative and environ me tal themes i.e. cadastral parcels, addresses and human health and safety (INSPIRE 2007). Since the INSPIRE framework is still under development no detailed descriptions or UML n otations of the themes are available. Within the field of ontology a number of thematic s emantic models are to be found. For example, the Towntology project aims to produce taxon omy of ontologies in the Urban Civil Engineering (UCE) field. To achieve this goal an onto logy tool suite call Towntology Tool Suite and a set of ontologies has been developed (Towntol ogy 2007). Figure 2. Example of taxonomy for the Urban System built using Towntology software. Physical, Socioeconomical, and mental objects are linked to the Ur ban System with part-of relationships. It is possib le to observe the first two levels of Physical objects ta xonomy (Caglioni 2006). The thematic semantic representations described are examples of subdivision of urban space into features, but they anyhow do not contain mappi ng to the geometric representations of 3D city models. In our opinion the semantic model requ ire a strong relation to the methods used to describe features within 3D city modeling.