Where in South Africa is your utopia?
For Day 8 of the 30 Day Map Challenge, the theme is Utopia. Thomas More coined the word in 1516 from two Greek roots that sound the same: ou-topos, "no place", and eu-topos, "good place". That pun became the idea for the map. There is no single best place to live in South Africa. It depends on what matters to you, and for some combinations of wishes there is no place at all.
After Day 7's sunrise bands, I wanted a map that every reader would use differently. So instead of showing one answer, this report lets you build your own.
Explore the interactive report
The page has three controls on the right:
- What matters to you? Tick any of six items: Weather, Town size, Wildlife and game reserves, Mountains, Beach, and Easy to fly out. Nothing ticked means everything counts.
- Your kind of weather: Mild all year, Warm winters, Mediterranean, Sunny and dry, Green and lush, or Four distinct seasons.
- Your kind of place: Rural and quiet, Small town, or City.
Every town gets a score from 0 to 100 for each item, and your Utopia score is simply the average of the items you tick.
- The map shows all 413 towns as 3D columns on a dark basemap with hillshading. The ten best towns rise as tall gold columns. If even the best town scores below 70, they turn bronze: your choices pull in opposite directions. The rest of the country is shaded from dark (weaker) to bright blue (stronger).
- Your top 10 lists the winners with their two best items. When no town fits well, the subtitle says so.
- The scorecard shows all six items for the town you click (or the number 1 town). Blue bars are the items you ticked, and the score is their average. Grey bars show what the town is like on everything else.
- Province re-ranks the towns within one province.
Interactive report: Day 8, Paradise Is Personal
This is a guide to setting, not services: climate, landscape, wildlife, coast, pace of life and flights. Safety, housing, jobs, healthcare and municipal services are not included, and a high score doesn't mean a town is a practical place to move to.
Starting with 413 towns
The towns come from OpenStreetMap's place=city and place=town points, extracted from the Geofabrik South Africa file of 30 October 2026 with a small Python reader (no GDAL or osmium needed). I kept those in mainland South Africa with at least 4,000 people within 10 km, using the European Commission's GHS-POP 2025 population grid.
Two rules stopped the cities dominating the list:
- No double counting. Starting with the largest places, a town within 15 km of one already kept is dropped, so suburbs fold into their city.
- Townships are part of their city. Large townships and suburbs inside the metros, such as Soweto, Mitchells Plain, Bellville and Mamelodi, fold into the metro's main city. Separate towns inside a metro boundary, such as Simon's Town and Bronkhorstspruit, stay in.
That left 413 towns, from Johannesburg to Rietfontein.
Six things you can tick
Weather: six kinds, not one "best"
The first version scored weather as "more comfortable days is better", and it handed the Cape a big advantage for its cool, grey winter days. Climate is a preference, not a ranking, so the dropdown now offers six kinds of weather, and a town can score well on more than one. They are built from two datasets:
- Feels-like temperature from the Copernicus ERA5-HEAT Universal Thermal Climate Index (UTCI), 2016–2025, at 0.25°. UTCI combines air temperature, humidity, wind and sun into one "how it feels" figure.
- Sunshine and rainfall from NASA POWER's climatology API. I wrote a small standard-library Python script that fetched one point per town, which you can run from PowerShell.
| Option | Scored on | Top towns |
|---|---|---|
| Mild all year | Neither too cold nor too hot: the lower of two scores for freezing and very hot (over 32 °C feels-like) days | Port Shepstone, Port St Johns, East London, Port Alfred |
| Warm winters | Freezing winter days, 100 at 3 or fewer, 0 at 40 or more | Komatipoort, Hoedspruit, Malalane, Richards Bay |
| Mediterranean | The share and amount of rain that falls in winter | Melkbosstrand, Paarl, Malmesbury, the Swartland |
| Sunny and dry | Sunshine and low rainfall | Augrabies, Upington, Springbok |
| Green and lush | Rainfall and cloud | Ixopo, Howick, Port Shepstone |
| Four distinct seasons | Summer thunderstorms, dry cold winters and altitude | Ermelo, Carolina, Phuthaditjhaba, Johannesburg |
Town size: closer to your kind of place
Density is the number of people within 10 km divided by the area of that circle. The score is how close a town is to the kind of place you pick: 100 at about 25 people per km² or fewer for Rural and quiet, around 120 for Small town, and 1,000 or more for City, losing 60 points for every tenfold difference.
Wildlife and game reserves
This counts the area of national parks, game reserves (public and private) and the large provincial parks within 50 km, such as Hluhluwe-iMfolozi, iSimangaliso and uKhahlamba-Drakensberg. A town scores 100 at 2,000 km² or more.
The protected areas are mostly OpenStreetMap relations, so I wrote a second reader that joins their member ways into rings. I then rasterised the rings on a 0.02° grid, about 2 km, and counted the cells within 50 km of each town.
Mountains, Beach and Easy to fly out
- Mountains: how far the highest of 4,749 OpenStreetMap peaks within 50 km rises above the town. Bergville, below Champagne Castle, tops the list with almost 2,000 m.
- Beach: 100 within 5 km of the Natural Earth coastline, falling to 0 at 60 km, roughly a day trip.
- Easy to fly out: 100 within 30 km of an airport with scheduled flights in OurAirports, falling to 0 at 250 km.
Data is not perfect
As with any analysis drawing on multiple data sources, the results are only as reliable as the underlying data and the assumptions used to interpret it. Every effort has been made to identify and correct anomalies, but some inaccuracies may remain; the findings are presented in good faith and are accurate to the best of my knowledge.
Bringing it together in Icon Map Slicer
3D columns
Icon Map Slicer has a Columns layer that extrudes each point as a cylinder or a square prism. I set the height in metres: the top 10 rise 4,000 m per point of score, so a town on 90 is a 360 km column. Every other town is a low column of 1.5 km plus 1.2 km per point above 40. With the map tilted to 55° on the Dark Matter basemap with hillshading, the gold towers stand out while the rest of the country reads as a low skyline.
Scores that react to every tick
The model is small: a town table with each 0–100 indicator, a six-row priorities table for the tick list, and two disconnected tables for the dropdowns. One measure scores a town on the priority in context. The Utopia score is AVERAGEX over the ticked priorities. Ranks are taken among the towns in the selected province, while ignoring clicks on the map and in the table, so clicking a town doesn't make it number 1.
When the answer is "no place"
Colour does the storytelling. If a top-10 town scores 70 or more, its column is gold. Below 70, it is bronze, and the table's subtitle explains that no town fits all your choices well. Seventy is not arbitrary: it is just above the best score any town reaches when all six items count.
What the map reveals
No place has everything. With all six items ticked, the best score is 69 and no town scores 80 or more on more than four of the six. The Kwazulu-Natal south coast has the climate but no mountains or big game. The Garden Route has mountains, forest and sea, but cold, windy winters. The Lowveld has wildlife and warmth, but heat and no beach.
Some utopias don't exist. Ask for mild weather all year and wildlife, and the best town scores 54. Kruger-edge towns follow on 50, all wildlife and no mild weather, and the KZN coast on 48, all mild weather and no wildlife. The columns turn bronze. That's More's "no place".
These are scores for a setting, not a verdict on where to live. They depend on the thresholds I chose, on how well OpenStreetMap is mapped, and on national data at 25–50 km resolution.
What I learned from Day 8
1. Every anomaly is a definition problem. Each measured something slightly different from what its label promised.
2. Rank versus fixed scale matters. A percentile rank always gives the middle town 50, even when its climate is unpleasant. A fixed scale makes a zero mean something.
3. "Neither too hot nor too cold" means the lower score, not the average. Averaging two conditions gave the Lowveld half-marks for a climate nobody would call mild.
4. Fewer, better items beat more, patchier ones. Six items on national datasets were easier to explain and harder to argue with than nine, and they showed the trade-offs more clearly.
5. Let the map say "no". A colour change and one line of text turn a weak top 10 into a finding.
Final thoughts
Most scoring maps hide their weightings and present one winner. Letting readers tick what matters to them, with a score they can check bar by bar, turns the same data into many answers, and occasionally into none. The approach works for any "where should we..." question, from site selection to where to open a branch: keep the items few, score them on scales people understand, and let the map show the trade-off rather than hide it.
Sources and attribution
- Towns, peaks and parks: © OpenStreetMap contributors, available under the Open Database Licence (ODbL). Geofabrik South Africa extract, 30 October 2026.
- Feels-like temperature: Copernicus Climate Change Service, Thermal comfort indices derived from ERA5 reanalysis (ERA5-HEAT), 2016–2025. DOI: 10.24381/cds.553b7518. Contains modified Copernicus Climate Change Service information.
- Sunshine and rainfall: NASA Langley Research Center, POWER Project climatology API, accessed November 2026.
- Population: European Commission, Joint Research Centre, GHS-POP R2023A, 2025 epoch, 30 arc-second grid. CC BY 4.0.
- Coastline and outline: Natural Earth, 1:10m. Public domain.
- Airports: OurAirports, airports with scheduled service. Public domain.
Methodology note: Towns are OpenStreetMap city and town points in mainland South Africa with at least 4,000 people within 10 km (GHS-POP 2025), with places within 15 km of a larger one removed and large metro townships folded into their city. Each item is scored 0–100: weather options from ERA5-HEAT feels-like temperature (UTCI) and NASA POWER sunshine and rainfall at each town's grid cell; town size from people per km² within 10 km; wildlife from km² of national parks, game reserves and major provincial parks within 50 km (OpenStreetMap, rasterised at 0.02°); mountains as the percentile rank of the relief of the highest peak within 50 km above the town (town height from NASA POWER, approximate); beach and flights from straight-line distance to the coast and to the nearest airport with scheduled flights. The Utopia score is the unweighted average of the items ticked. Climate and population data are 25–50 km and 1 km resolution, so values describe the area around a town, not the town centre. OpenStreetMap is not a complete inventory.