Urban form simulation in 3D based on cellular automata and building objects generation

Modeling urban form change is fundamental to mitigate the negative impacts of urban development. Despite the extensive applications of cellular automata (CA) models in urban simulation, an important drawback of contemporary urban CA models is that they cannot explicitly represent urban form change in a 3D manner, hence hampering the proper evaluation of the impacts of urban form. This study addresses this issue by developing a hybrid approach of 3D urban form simulation. The proposed approach integrates a patch-based urban CA model and a tool for 3D modeling of building objects. A case study is carried out in Panyu, China, to evaluate the effectiveness of the proposed simulation approach. The experimental results show that the simulated urban land use pattern achieves an overall similarity of 83% as compared against the observed one. The building height predictions yielded by the random forest algorithm have relatively good performance, with the errors close to a standard floor height (3.18–4.57 m). Three scenarios of future urban form are simulated according to the narratives of the shared socioeconomic pathways (SSP). The SSP simulations are used to produce three sets of gridded morphological parameters that are compatible with mainstream regional climate models. The resulting urban form parameters are compared to infer the potential effects of different urban forms in the SSP scenarios.