A CityGML extension for handling very large tins

A CityGML extension for handling very large tins

Kumar, Kavisha (TU Delft Urban Data Science)
Ledoux, H. (TU Delft Urban Data Science)
Stoter, J.E. (TU Delft Urban Data Science)

Dimopoulou, E. (editor)
van Oosterom, P. (editor)

2016

In addition to buildings, the terrain forms an important part of a 3D city model. Although in GIS terrains are usually represented with 2D grids, TINs are also increasingly being used in practice. One example is 3DTOP10NL, the 3D city model covering the whole of the Netherlands, which stores the relief with a constrained TIN containing more than 1 billion triangles. Due to the massive size of such datasets, the main problem that arises is: how to efficiently store and maintain them? While CityGML supports the storage of TINs, we argue in this paper that the current solution is not adequate. For instance, the 1 billion+ triangles of 3DTOP10NL require 686 GB of storage space with CityGML. Furthermore, the current solution does not store the topological relationships of the triangles, and also there are no clear mechanisms to handle several LODs. We propose in this paper a CityGML extension for the compact representation of terrains. We describe our abstract and implementation specifications (modelled in UML), and our prototype implementation to convert TINs to our CityGML structure. It increases the topological relationships that are explicitly represented, and allows us to compress up to a factor of ∼ 25 in our experiments with massive real-world terrains (more than 1 billion triangles).

CityGML
GML
Terrains
TINs

http://resolver.tudelft.nl/uuid:65fc90a6-7c00-42ca-88ab-e949d853fe33

https://doi.org/10.5194/isprs-annals-IV-2-W1-137-2016

ISPRS

ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, IV-2/W1

11th 3D Geoinfo Conference, 2016-10-20 → 2016-10-21, Athens, Greece

Institutional Repository

conference paper

© 2016 Kavisha Kumar, H. Ledoux, J.E. Stoter