Increasing level of detail of buildings for improved simulation of 4D urban digital twin

Bulatov, Dimitri and May, Marie and Strauß, Eva and Mancini, Francesco and Kottler, Benedikt and Helmholz, Petra; (2022) Increasing level of detail of buildings for improved simulation of 4D urban digital twin. In: Annual Conference Proceedings of the XXVIII International Seminar on Urban Form. University of Strathclyde Publishing, Glasgow, pp. 503-512. ISBN 9781914241161

[thumbnail of Bulatov-etal-ISFU-2021-Increasing-level-of-detail-of-buildings-for-improved-simulation-of-4D-urban-digital-twin]
Text. Filename: Bulatov_etal_ISFU_2021_Increasing_level_of_detail_of_buildings_for_improved_simulation_of_4D_urban_digital_twin.pdf
Final Published Version
License: Creative Commons Attribution 4.0 logo

Download (3MB)| Preview


Buildings represent a crucial component of urban morphology, and their accurate modeling is essential for a number of applications involving Urban Digital Twins. With respect to thermal simulation aiming to identify Urban Heat Islands, a trade-off between accurate modeling of a single building type and large-scale reconstruction of virtual city models needs to be found. In the proposed paper, we analyzed an Australian suburb containing approximately 1700 residential buildings with challenging roof structures. Building outlines are provided by geo-information data and converted into prismatic models of LOD1. Using airborne sensor data (digital orthophotos, high-resolution images, and digital surface models), we identified two ways to increase the LOD and thus, the accuracy of the simulation. Firstly, we used common Computer Aided Graphics software to model interactively a few selected buildings, a process denoted as geo-specific modeling. Here, the outlines were used as foundations for constructing the ground-level walls. We relied on airborne data to retrieve building heights and roof structures. Number of floors and positions of façade elements were modeled on standard typological assumptions and building practices. We developed an interface to import automatically LOD1- based data and to export LOD3 buildings into the simulation. Secondly, we reproduce these models to model other buildings of the dataset. For this so-called geo-typical modeling, a similarity measure based on the outlines was implemented. The final scene consists of triangles modeling LOD3 buildings, terrain, and trees, retrieved using machinelearning- based methods on land cover classification. Together with the semantic class, we store the geometrical and physical properties of every triangle. The environmental data (e.g., cloud coverage, air temperature) is available by means of the weather services. Surface temperature is modeled by considering conductive, convective, and radiative heat transfer. The simulation of updated LOD3 buildings shows a significantly increased realism of the temperature distribution in an urban area. It can used to verify sustainable design of appropriate morpho-typologies for a particular precinct in a given context.

Persistent Identifier