2.2. Main features of 3DCityDB

Many (but not all) of the features referring to object modelling and representation are implied by following the CityGML standard 2.0.0 issued by the Open Geospatial Consortium.

2.2.1. CityGML 2.0.0 and 1.0.0 compliant database

The implementation defines the classes and relations for the most relevant topographic objects in cities and regional models with respect to their geometrical, topological, semantical, and appearance properties. Included are generalization hierarchies between thematic classes, aggregations, relations between objects, and spatial properties. These thematic information go beyond graphic exchange formats and allow to employ virtual 3D city models for sophisticated analysis tasks in different application domains.

For the representation of all vector and grid geometry the built-in data types provided by the spatially-enhanced relational database manage­ment systems Oracle Spatial/Locator (10G R2 or higher) or PostgreSQL (9.6 or higher) with PostGIS extension (2.3 or higher) are used. This way, special solutions are avoided and different geoinformation systems, CAD/BIM systems, and ETL software systems can directly access (read and write) the geometry objects stored in the SRDBMS.

The version and history management employs Oracle’s Workspace Manager and, hence, is only available for 3DCityDB instances running on an Oracle RDBMS. It is largely transparent to application programs that work with the database.

2.2.2. Support for CityGML Application Domain Extensions (ADEs)

Semantic 3D city models are employed for many different applications from diverse domains like energetic, environmental, driving, and traffic simulations, as-built building information modeling (as-built BIM), asset management, and urban information fusion. In order to store and exchange application specific data aligned and integrated with the 3D city objects, the CityGML datamodel can be extended by new feature types, attributes, and relations using the CityGML ADE mechanism. ADEs are specified as (partial) GML application schemas using the modeling language XML Schema. Starting from release 4.0.0 the 3DCityDB database schema can be dynamically extended by arbitrary ADEs like the Energy ADE, UtilityNetwork ADE, Dynamizer ADE, or national CityGML extensions like IMGeo3D (from The Netherlands).

Since ADEs can define an arbitrary number of new elements with all types and numbers of spatial properties, a transformation method has been developed to automatically derive the relational database schemas for arbitrary ADEs from the ADE XML schema files. Since we intended to follow similar rules in the mapping of the object-oriented ADE models onto relational models as we used for the (manual) mapping of the CityGML datamodel onto the 3DCityDB core schema, the Chair of Geoinformatics at TUM developed a new transformation method based on graph transformation systems. This method is described in detail in [YaKo2017] and is implemented within the “ADE Manager” plugin for the Importer/Exporter software tool.

The ADE Manager performs a sophisticated analysis of the XML schema files of an ADE, the automatic derivation of additional relational table structures, and the registration of the ADE within the 3DCityDB. Furthermore, SQL scripts are generated for each ADE for e.g. the deletion of ADE objects and attributes from the database. Please note that in order to support also the import and export of CityGML datasets with ADE contents, a Java library for the specific ADE has to be implemented. This library has to perform the handling of the CityGML ADE XML elements and the reading from and writing into the respective ADE database tables using JDBC and SQL. An example how to develop such a Java library is given for a Test ADE in the 3DCityDB github repository.

2.2.3. Importing and exporting CityGML data

The included Importer/Exporter software tool allows for high performance importing and exporting of CityGML datasets according to CityGML versions 2.0 and 1.0. The tool allows processing of very large datasets (>> 4 GB), even if they include XLinks between CityGML features or XLinks to 3D GML geometry objects. The multi-threaded programming exploits multiprocessor systems or multikernel CPUs to speed up the processing of complex XML-structures, resulting in high performance database access. Objects can be filtered during import or export according to spatial regions (bounding box), their object IDs, feature types, names, and levels of detail. Bounding boxes can be interactively selected using a map window based on OpenStreetMap (OSM).

A tiling strategy is implemented in order to support the export of very large datasets. In case of a very high number of texture images they can be automatically distributed in a configurable number of subdirectories in order to avoid large directories with millions of files which can render a Microsoft Windows operating systems unresponsive. The Importer can also validate CityGML files and can be configured to only import valid features. It considers CityGML ADE contents, if the ADEs have been registered in the database and specific Java libraries for reading/writing the ADE contents from/into the ADE database tables is provided (see above). The Importer/Exporter tool can be run in interactive or batch mode.

2.2.4. Export to KML, COLLADA and glTF

The Importer/Exporter tool can also export city models to KML, COLLADA and glTF formats which can directly be viewed and interactively explored in geoinformation systems (GIS) or digital virtual globes like Google Earth or Cesium WebGL Virtual Globe. A tiling strategy is supported where only tiles in the vicinity of the viewer’s location are being loaded facilitating the visualization of even very large 3D city and landscape models. Information balloons for all objects can be configured by the user. The exported models are especially suited to be visualized using the 3DCityDB-Web-Map-Client (see below), an Open Source 3D web viewer that is based on the CesiumJS Webglobe framework with many functional extensions.

2.2.5. Spreadsheet export

The Spreadsheet Generator (SPSHG) allows exporting thematic data of 3D objects into tables in CSV and Microsoft Excel format which can be uploaded to a Google Spreadsheet within the Google Document Cloud. For every selected geoobject one row is being exported where the first column always contains the GMLID value of the respective object. The further columns can be selected by the user. This tool can be used to export attribute data from e.g. buildings like the class, function, usage, roof type, address, and further generic attributes that may contain information like the building energy demand, potential solar energy gain, noise level on the facades etc. The spreadsheet rows can be linked to the visualization model generated by the KML/COLLADA/glTF Exporter. This is illustrated in Section 6.2.

2.2.6. Interactive 3D web visualization

The 3DCityDB-Web-Map-Client is a WebGL-based 3D web viewer which extends the Cesium Virtual Globe to support efficient displaying, caching, prefetching, dynamic loading and unloading of arbitrarily large pre-styled 3D visualization models in the form of tiled KML/glTF datasets generated by the KML/COLLADA/glTF Exporter. It provides an intuitive user interface to facilitate rich interaction with 3D visualization models by means of the enhanced functionalities like highlighting the objects of interests on mouseover and mouseclick as well as hiding, showing, and shadowing them. Moreover, the 3DCityDB-Web-Map-Client is able to link the 3D visualization model with an online spreadsheet (Google Fusion Table) in the Google Cloud and allows viewing and querying the thematic data of every city object according to its GMLID. For details see also [YaCK2016] and [ChYK2015].

2.2.7. Web Feature Service (WFS) 2.0

The 3DCityDB comes with an OGC compliant implementation of a basic WFS 2.0 allowing web-based access to the 3D city objects stored in the database. WFS clients can directly connect to this interface and retrieve 3D content for a wide variety of purposes. The implementation currently satisfies the Simple WFS conformance class. The WFS considers CityGML ADE contents, if the ADEs have been registered in the database and specific Java libraries for reading/writing the ADE contents from/into the ADE database tables is provided (see above). An implementation of a full, transactional WFS is commercially available from one of the development partners, see Section 6.3.

2.2.8. Docker support

We now provide Docker images for

  1. a complete 3DCityDB installation pre-installed in a PostGIS
  2. a webserver with an installed 3DCityDB-Web-Map-Client
  3. a 3DCityDB WFS

We also provide a Docker-compose script to launch all three Docker containers in a linked way with just a single command. Details are given in Section 1.5 and in the respective github repositories. Docker is a runtime environment for virtualization. Docker encapsulates individual software applications in so-called containers, which are – in contrast to virtual machines – light-weight and can be deployed, started and stopped very quickly and easily. Using our Docker images a 3DCityDB can be installed by a single command.

2.2.9. Open Source and Platform Independence

The entire software is freely accessible to the interested public. The 3DCityDB is licensed under the Apache License, Version 2.0, which allows including 3DCityDB in commercial systems. You may obtain a copy of the Apache License at http://www.apache.org/licenses/LICENSE-2.0. Both the Importer/Exporter tool and the Web Feature Service are imple­mented in Java and can be run on different platforms and operating systems.

2.2.10. Features inherited from CityGML

  • Complex city object modelling: The representation of city objects in the 3D city database ranges from coarse models to geometrically and semantically fine grained structures. The underlying data model is a complete realization of the CityGML data model for the levels of detail (LOD) 0 to 4. For example, buildings can be represented by simple, monolithic objects or can consist of an aggregation of building parts. Extensions of buildings, like balconies and stairs, can be classified thematically and provided with attributes just as single surfaces can be. LOD4 completes a LOD3 model by adding interior structures for 3D objects. For example, LOD4 buildings are composed of rooms, interior doors, stairs, and furniture. This allows among other things to select the floor space of a building, so that it can later be used e.g. to derive SmartBuildings or to form 3D solids by extrusion [DBBF2005]. Buildings can be assigned addresses that are also stored in the 3D city database. Their implemen­tation refers to the OASIS xAL Standard, which maps the address formats of the different countries into a unified XML schema. In order to model whole complexes of buildings, single buildings can be aggregated to form special building groups. The same complex modelling applies to the other CityGML feature types like bridges, tunnels, transportation and vegetation objects, and water bodies.

  • Complex digital terrain models: DTMs may be represented in four different ways in CityGML and therefore also in the 3D city database: regular grids, triangular irregular networks (TINs), 3D mass points and 3D break lines. For every level of detail, a complex DTM consisting of any number of DTM components and DTM types can be defined. Besides, it is possible to combine certain kinds of DTM representations for the same geographic area with each other (e.g. mass points and break lines or grids and break lines). In Oracle Spatial (but not Locator) Grid-based DTMs may be of arbitrary size and are composed from separate tiles to a single overall grid using the Oracle GeoRaster functionality. Please note that the Import/Export tool provides functions to read and write TIN, mass point, and break line DTM components, but not for raster based DTMs. GeoRaster data would have to be imported and exported using other tools from e.g. Oracle, ESRI, or Safe Software.

  • Support of different kinds of multi-representations: Levels of detail, different appearances, (and with Oracle RDBMS only) planning versions and history: Every geoobject as well as the DTM can be represented in five different resolution or fidelity steps (Levels of Detail, LOD). With increasing LOD, objects do not only obtain a more precise and finer geometry, but do also gain a thematic refinement. Different appearance data may be stored for each city object**: Appearance relates to any surface-based theme, e.g. infrared radiation or noise pollution, not just visual properties. Consequently, data provided by appearances can be used as input for both presentation and analysis of virtual 3D city models. The database supports feature appearances for an arbitrary number of themes per city model. Each LOD of a feature can have individual appearances. Appearances can represent – among others – textures and georeferenced textures. All texture images can be stored in the database. (cf. [GKSS2005])

  • Representation of generic and prototypical 3D objects: Generic objects enable the storage of 3D geoobjects that are not explicitly modelled in CityGML yet, for example dams or city walls, or that are available in a proprietary file format only. This way, files from other software systems like architecture or computer graphics programs can be imported directly into the database (without interpretation). However, application systems that would like to use these data must be able to interpret the corresponding file formats after retrieving them back from the 3D geodatabase.

    Prototypical objects are used for memory-efficient management of objects that occur frequently in the city model and that do not differ with respect to geometry and appearance. Examples are elements of street furniture like lanterns, road signs or benches as well as vegetation objects like shrubs, certain tree types etc. Every instance of a prototypical object is represented by a reference to the prototype, a base point and a transformation matrix for scaling, rotating and translating the prototype.

    The geometries (and appearances like textures, colors etc.) of generic objects as well as prototypes can be stored either using the geometry datatype of the spatial database management system (Oracle Spatial/Locator or PostGIS) or in proprietary file formats. In the latter case a single file may be saved for every object, but the file type (MIME type), the coordinate transformation matrix that is needed to integrate the object into the world coordinate reference system (CRS) and the target CRS have to be specified.

  • Extendable object attribution: All objects in the 3D geodatabase can be augmented with an arbitrary number of additional generic attributes. This way, it is possible to add further thematic information as well as further spatial properties to the objects at any time. In combination with the concept of generic 3D objects this provides a highly flexible storage option for object types which are not explicitly defined in the CityGML standard. Every generic attribute consists of a triple of attribute name, data type, and value. Supported data types are: string; integer and floating-point numbers; date; time; binary object (BLOB, e.g. for storing a file); geometry object according to the specific geometry data type of Oracle or PostGIS respectively; simple, composite, or aggregate 3D solids or surfaces. Please note that generic attributes of type BLOB or geometry are not allowed as generic attributes in CityGML (and will, thus, not be exported by the CityGML exporter). However, it may be useful to store binary data associated with the individual city objects, for example, to store derived 3D computer graphics representations.

  • Free, also recursive grouping of geoobjects: Geoobjects can be grouped arbitrarily. The aggregates can be named and may also be provided with an arbitrary number of generic attributes (see above). Object groups may also contain object groups, which leads to nested aggregations of arbitrary depth. In addition, for every object of an aggregation, its role in the group can be specified explicitly (qualified association).

  • External references for all geoobjects: All geoobjects can be provided with an arbitrary number of references to corresponding objects in external data sources (i.e. hyperlinks / linked data). For example, in case of building objects this allows to store e.g. the IDs of the corresponding objects in official cadasters, digital landscape models (DLM), or Building Information Models (BIM). Each reference consists of an URI to the external data store or database and the corresponding object ID or URI within that external data store or database.

  • Flexible 3D geometries: The geometry of most 3D objects can be represented through the combination of solids and surfaces as well as any - also recursive - aggregation of these elements. Each surface may has attached different textures and colors on both its front and back face. It may also comprise information on transparency. Additional geometry types (any geometry type supported by the spatial database management system Oracle Spatial/Locator or PostGIS) can be added to the geoobjects by using generic attributes.