Assessing the impacts of gridded population model choice on degree of urbanisation metrics

Defining urban and rural areas is crucial for assessing the accessibility of services and opportunities that impact people worldwide. The Degree of Urbanisation framework, endorsed by the UN Statistical Commission, primarily uses population grids to classify areas through a harmonised, population-centric approach, enabling international comparisons. However, variations in the distribution of population counts at the grid-cell level across different population datasets can significantly influence the resulting patterns. We applied the Degree of Urbanisation to 16 countries in Africa and the Caribbean, using four population grids to evaluate these effects. It shows that differences primarily occur in the classification of urban cluster. On average, 27.5 % of the population falls into mixed classes, with 17.5 % in mixed rural and urban cluster areas and 7.8 % in mixed urban cluster and urban centre areas. Population grids that only model populations within satellite-detected settlements show limited disagreement, with mixed rural and urban cluster population classifications decreasing by 5.6 percentage points and mixed urban cluster and urban centre populations by 1.4. Population modelling approaches that distribute populations more broadly, including outside of detected built-up areas, substantially reduce settlement identifications, resulting in 42.3 % fewer urban centres and 66.2 % fewer dense urban clusters than the average across the study countries. Our analyses highlight the potential sensitivity of Degree of Urbanisation to gridded population modelling assumptions and provide guidance on its implementation.