Who lives within reach of Harare's lifestyle centres?
How many people can reach Sam Levy's Village, Arundel Village, Honeydew and Kamfinsa, how have those neighbourhoods changed since 2012, and where is Harare growing instead?
4 November 2026 · Real estate, Retail · Read the article on dotzedw.com · City map
Every lease, loan and store opening at a Harare shopping centre rests on a catchment number that is rarely checked. We checked it for four of the city's best-known centres, with drive and walk times on the city's road network, Google's satellite outlines of 1.24 million buildings, and ZIMSTAT's census counts for each of Harare's 78 wards.
We also tested three population maps that analysts commonly use against the 2022 census. They disagree by a factor of four to five on how many people live in Harare's northern suburbs, and only one of them agrees with the census. This page holds the full results.
What members get
- People within 5, 10, 15 and 20 minutes' drive and 15 and 30 minutes' walk of each centre, checked against the census
- How each centre's catchment changed between the 2012 and 2022 censuses
- Roof area per resident, a measured proxy for wealth, and the overlap between centres
- The census check of three population maps, ward by ward, with downloads
- Maps of the catchments and of population change by ward
- A car-park activity index for seven centres from ten years of free satellite images: weekends, month-ends, holidays and the 2020 lockdown
This study is part of the Individual plan
Subscribe on dotzedw.com and you come straight back to this page with Individual access. Already a member? Sign in with your dotzedw.com account.
How we did it
Drive and walk times come from the dotzedw city model: OpenStreetMap roads with free-flow speeds by road class, so they are off-peak times, and 15 km/h from each home to the nearest road junction.
Population is WorldPop's 2024 constrained map at 100 m. We chose it after testing it, GHSL and Meta's High Resolution Settlement Layer against ZIMSTAT's 2022 census counts by ward (compiled by citypopulation.de), using ZIMSTAT's ward boundaries as published by UN OCHA. GHSL was compared through our 0.74 km² hexagons.
Change from 2012 to 2022 shares each ward's census counts across a catchment in proportion to where WorldPop places people today.
Buildings come from Google Open Buildings v3 (outlines, for roof area per resident, leaving out land mapped in OpenStreetMap as business or industrial and buildings over 1,500 m²) and Open Buildings 2.5D Temporal (building growth, 2016 to 2023).
Each centre is located at its mapped point; Sam Levy's Village is measured from Village Walk, the centre beside it, about 270 m away.
Car parks: the open paved ground inside each centre's outline from OpenStreetMap (or 160 m around its point where none is mapped), less roofs, public roads and anything green. Its mean brightness on every Sentinel-2 image with at least 80% of it clear, 2017 to 2026 (from Element 84's Earth Search), modelled with the season, the year and terms for weekdays, month-ends, holidays and lockdowns in a robust regression. The radar check used Sentinel-1 terrain-corrected backscatter. Contains modified Copernicus Sentinel data 2017 to 2026.
What this cannot tell you
- Drive times assume free-flowing roads; at peak hours real catchments are smaller.
- Roof area is a proxy for wealth, not a measure of income or spending.
- The census gives totals by ward; how people are spread within a ward is WorldPop's model.
- OpenStreetMap and Overture miss many shops, especially in high-density and peri-urban areas, so where shops are absent needs checking on the ground.
- Catchments show who could reach a centre, not who does.
- Free 10 m imagery cannot count cars, compare crowds between centres, or see the rainy season (wet asphalt is darker); it shows each car park's rhythm against itself.