Simulation of Urban Density Scenario according to the Cadastral Map using K-Means unsupervised classification

Due to the increasing trend of urbanization over the last few decades, city development and planning have become a major concern for the authorities. Many efforts have been made in the field of urban growth modelling by researchers and urban planners to investigate the factors and effects of urbanization and urban growth. The work presented here is part of a project that simulates urban densification scenarios. This work describes the automatic detection and classification of neighbourhoods based on vector data of buildings and parcels. During the urbanization process, an empty parcel is divided into sub-parcels to accommodate new buildings. In general, the sub-parcels created have the same characteristics (surface area, building height, etc.) as the already urbanized parcels in their vicinity. Here, the aim is to detect and identify the different neighbourhoods in the study area to determine the characteristics of the new parcels. This is done in three steps: first, classifying the urbanized parcels, then identifying the different neighbourhoods based on this classification, and finally dividing the parcels into sub-parcels with features depending on the surrounding neighbourhoods. The model developed here is designed to automatically identify different urban zones and simulate the division of parcels over time according to a variety of urban density scenarios. This model is also compatible with the creation of new buildings based on these levels of density.