Multi-scale gridded urban morphometrics for settlement classification and population mapping

Urban areas are expanding rapidly around the world, and much of this growth is expected in low- and middle-income countries. Policy makers, researchers, and those implementing development projects need up-to-date and consistent information on cities in order to plan and track progress towards Sustainable Development Goals. Yet in many places experiencing rapid growth, information on urban areas and their population is lacking, outdated or incomplete. In recent years, increasing availability of very high spatial resolution imagery (<1 m resolution) and computing power is enabling sets of building footprint polygons to be automatically extracted from the imagery and mapped for whole countries. These building footprint datasets provide a unique resource to study urban morphometrics in places which may lack other local data. This paper demonstrates the use of a spatial grid to classify urban fabric into settlement types. This unit of analysis is in contrast to plots or parcels which are more commonly used in urban morphology studies, and a case study in Southampton, UK is used to explore the sensitivity of the results to varying the parameters used to define the size of the grid. These initial results suggest that multiple scales of observation windows can be combined to identify key patterns across space and that multiple grid resolutions can give relatively consistent classification results. Future work is needed to explore the use of grids to study urban form in other settings.