Discretization of urban areas using POI-based tesselation

Urban area tessellation is a crucial aspect in many spatial analyses. While regular tessellation methods, like square-grid or hexagon-grid, are suitable for addressing pure geometry problems, they cannot take the unique characteristics of different subareas into account. Irregular tessellation methods allow the border between the subareas to be defined more realistically based on the urban features like road network or POI data. This paper studies and compares five different tessellation methods: Squares, hexagons, adaptive squares, Voronoi diagrams, and city blocks. We explain how (open-source) POI data can be integrated into the tessellation process to build what we call 'Local Geographic Units' (POI-based tiles). These units are flexible and adaptable to the structure of the studied area and underlying data and could improve the performance of further analyses. The results of the various tessellation methods are demonstrated for the city of Frankfurt am Main in Germany. A simple clustering of Local Geographic Units for the studied city indicates that city blocks perform better than the other methods in the city segmentation in terms of reflecting the structure of this city.