The 3D GeoInformation research group from the Faculty of Architecture and the Built Environment is working on building a 3D model of all the towns and cities in the Netherlands. It is the only model of its kind on a national scale and can be used to accurately project the size of the population. It is also a useful tool for creating flood models and calculating noise and daylight contours. As part of the research group 3D GeoInformation, PhD researcher Filip Biljecki investigated the level of detail in 3D city models.
Remote estimates of population size for demographic purposes were previously made on the basis of maps and satellite images, by dividing areas into smaller sections and subsequently tallying them together. These days, 3D modelling techniques and geographical information science (GIScience) have significantly expanded the possibilities. The height and volume of buildings can be estimated remotely and it is possible to distinguish between densely populated cities and sparsely populated rural areas. “We can then check the reliability of our calculations against the accurate data on actual numbers of residents we receive from Statistics Netherlands,” says Filip Biljecki. “This enables us to generate the size of the population automatically.”
One important ingredient for the 3D population model is the open dataset of Statistics Netherlands, which provides insight into the population composition of 12,237 districts and neighbourhoods in the Netherlands. The researchers have linked this dataset to the Up-To-Date Height Model of the Netherlands (Actuele Hoogtebestand Nederland) and to the 2D models of buildings available from Kadaster (the Netherlands’ Cadastre, Land Registry and Mapping Agency) to create a stratified image.
The general point of departure is: the larger the building, the more people live in it. Of course, that assumption is not always correct. While the model can distinguish between different building uses, buildings are sometimes vacant. “However, the error margins at the detailed level actually come down to 1 per cent when considered across the country as a whole,” says Biljecki. This makes the 3D model a cost-effective tool for instantly establishing how many people live where.
A breach in the dike
In addition to demographers, the new 3D model is also of interest to architects and urban planners, who can use it to look up building geometry information. With a few mouse clicks, they can view the height differences across a city and shadow effects. This is particularly useful when placing solar panels, but also when calculating the daylight factor for a building. Streets where there is significant traffic noise are marked in red. It is also possible to see how far this noise reaches, both horizontally and vertically.
Data on the height of roof shapes (for placing green roofs), kerbs and green areas facilitate mapping out the potential consequences of flooding and how to prevent this. The model shows which buildings will be flooded in the event of a dike breach and provides an indication of how much damage this would likely cause. This information is useful not only for urban planners, but also for insurance companies.
In total, all 9.9 million buildings in the Netherlands are included in the model, along with their cadastral details. Each building is linked to data from various other databases, which creates a vastly more detailed image than was ever possible in the 2D era. For example, you can see the new Delft station project in three dimensions: the ProRail tunnel, the NS ticket hall, and the municipality’s offices and outdoor area. The 3D model allows you to see the various ownership titles of every part of the project, right down to the lifts and staircases. “You could never have seen that sort of thing in 2D. All in all, the advantages offered by the 3D model are simply enormous,” says Biljecki.
Published: April 2017
- PhD thesis 'Level of detail in 3D city models'