Skip to main content

Urban modelling in higher dimensions: embedding generalisation of 3D data in a 4D model

Periodic Reporting for period 3 - UMnD (Urban modelling in higher dimensions: embedding generalisation of 3D data in a 4D model)

Reporting period: 2019-09-01 to 2021-02-28

3D geographic information about urban objects (buildings, roads) is needed to monitor and control processes within modern urban areas (noise, flooding, energy demand/supply). However, each specific process requires 3D data with its own specific semantic and geometric Level of Detail (LoD), and current approaches require enormous manual efforts to collect general-purpose 3D data and to transform it to make it suitable for a specific application.

In this project, we will develop a fundamental solution for providing 3D data at application specific LoDs. For this, we will use an innovative method that goes far beyond the state-of-the-art by introducing higher dimensional (nD) mathematical models, which will enable us to enforce consistency by modelling the LoDs as an extra dimension to the 3D spatial dimensions in an integrated 3D+LoD (4D) model.

We have defined three key research lines for the project: (i) a groundbreaking extension of 2D cartographic generalisation solutions to 3D, enabling us to automatically derive application-specific coarse 3D data from fine 3D data; (ii) embedding multiple 3D city models at different levels of detail into a single 4D model; and (iii) “slicing” operations that extract custom 3D cross-sections of the 4D model.

By combining the results of these three lines of research, we aim to generate error-free 3D data at application specific LoDs for urban applications. This will advance urban applications to better support the decision-making process in order to make cities more sustainable, livable, clean, resilient, inclusive, less noisy, better accessible, etc.

If successful, the 4D approach opens a new horizon for modelling parametrisable aspects of urban environments, which may establish new modelling paradigms in the future.
Since the start of the project, we have studied how to reconstruct and generalise different 3D data from one 3D dataset serving different needs at different levels of detail. This provides the Level of Details for the 4D model.
In addition, we have studied and developed the appropriate 4D data structure for storing the generalisation knowledge. Although LoD (scale) is a well-known concept in the GIS domain, modelling it as an extra dimension of geographic data is new.
Finally, a slicing methodology has been developed to derive error-free, application specific 3D data from the 4D model, so that practitioners have access to the specific 3D data that they need in their applications and can advance in their own domains by building 3D-aware urban applications.
To accomplish the urban modelling in nD, we use an innovative method for 3D+LoD modelling by introducing higher dimensional (nD) mathematical models to the well-established discipline of cartographic generalisation in 2D. We develop a 3D+LoD data structure and a slicing method to derive error-free application-specific 3D data from the 4D model. This will be the first solution that is able to store different LoDs of 3D urban models in one 4D data structure, to visualise 4D urban data and to derive application specific urban data from the 4D model.
Since the start of the project, we have succeeded in developing the fundamentals for this proposed innovative methodology (i.e. 3D generalisation to obtain 3D data at different LoDs; 4D data structure; 4D visualiser). We have also studied key principles that support the success of the proposed methodology and 3D urban data applications in general. More specifically, we have developed a method to identify errors and clean 3D data, we have developed a standard to compactly encode and manipulate 3D urban data, and we have studied and identified the 3D+LoD data needs of a wide variety of urban applications.
Our study on use cases for 3D data for urban applications and automatically reconstructing application specific 3D+LoD data accordingly, is new to the prevailing method of maintaining one 3D data set that is supposed to serve any application. The last one has proven to be inadequate to meet the 3D data needs of urban applications. Therefore, the problem remains of independently acquired and stored LoDs of an urban model. This is solved with our approach.
Finally, our extensive studies on the integration of Geo and BIM data, has yielded new findings and insights regarding the GeoBIM domain in how to address the challenges of this integration in order to use the potentials of the integration.