Assessment of different building representations in numerical urban flood modeling

According to the latest IPCC report, climate change directly impacts the intensity and frequency of floods in urban areas. As a result, Flash floods, characterized by a rapid increase in flood peak during a short duration, are becoming more common. Managing these flash floods is a crucial yet challenging task for water authorities. One important tool supporting the management of flash floods are 2-D hydrodynamic models.

In this study, we use the 2-D module of the openTELEMAC-MASCARET software to investigate the effect of the building representation on an urban flash flood. For this, we isolate a sinuous-shaped building group within an artificial study area. The building group itself and the topography in the model are derived from the Moabit district in Berlin. The buildings are cut out from the model as a hole, and the impact on the model results from the vertex distribution around these holes is assessed. We compare different algorithms to resample the vertices of the building. First, we design an algorithm that resamples the vertices of the building edges at an even distance while conserving the overall shape of the building. The resampling distance is set globally but slightly varies for every building edge. This algorithm is then compared with built-in resampling tools from other software: one tool also conserves the shape of the building yet cannot resample the vertices at an even distance (from QGIS), and another tool resamples at an even distance, yet does not conserve the building shape (from SMS). A change in the distribution of vertices causes a change in the mesh distribution around the buildings. Hence, the flow pattern around the buildings also changes. Additionally, we study the numerical stability of the different distributions. With this study, we aim to deepen the understanding of building representation in urban flood modeling and set the path for further investigations within this topic.