Central African Power Pool - distribution network
S\u00e3o Tom\u00e9 and Pr\u00edncipe distribution network map
The medium-voltage lines that carry power from substations to towns, farms and villages, mapped from public data. Real lines are solid; where no mapped line exists, dashed lines show where one is predicted along the roads.
Click a line, substation or plant for details. Click the map to zoom with the scroll wheel. Lines are indicative, not for engineering or investment decisions.
Lines by voltage and source
There is no public low-voltage data for S\u00e3o Tom\u00e9 and Pr\u00edncipe, so the map shows medium voltage only.
Each line keeps its source. Where two sources map the same line, we keep one, in this order: utility data, the World Bank 2017 map, OpenStreetMap, then predicted lines.
| Voltage | km |
|---|---|
| 30 to 33 kV | 98 |
| Medium voltage, not recorded | 8 |
| Predicted along roads | 111 |
| Total | 217 |
| Source | km | Share |
|---|---|---|
| OpenStreetMap (newer lines not in the 2017 map) | 106 | 49% |
| Predicted along roads (night-time lights, roads and substations) | 111 | 51% |
| Total | 217 | 100% |
How we built this map
- We start from the World Bank Africa electricity grid (2017), the most complete public medium-voltage map for many countries, and add OpenStreetMap lines below 40 kV and power=minor_line, each only where it runs more than 250 m from a line already on the map.
- Predicted lines follow major roads (motorway to tertiary, from OpenStreetMap), because rural feeders mostly run along roads. A road section is predicted to carry a line where gridfinder (Arderne et al. 2020), a model built on night-time lights, predicts one running along it and no mapped line already explains it. Predicted sections are then joined to the nearest substation or mapped line along the roads, so they form connected feeders, and short gaps along the same road are closed. Smaller (unclassified) roads are used only to reach villages and towns in OpenStreetMap that sit within 1.5 km of a night-time-light prediction. They may be lines that exist but are not mapped, lines at another voltage, or no line at all. We tested the model on Kenya with Kenya Power's own map left out: 85% of the predicted line length lay within 1 km of a real Kenya Power line, and mapped plus predicted lines reached 54% of the real network, against 6% for mapped lines alone.
- The 2017 map does not show lines built since then, and OpenStreetMap coverage depends on volunteers. Low-voltage networks are rarely public.
- Transmission lines (40 kV and above) are left out of this map; see the grid-edge atlas for the wider grid. World Bank lines with no recorded voltage are left out, since some are transmission.
- Geometry is simplified to about 30 m for the web. Lengths are measured before simplifying.
Sources
- Lines, substations and small plants: OpenStreetMap contributors (ODbL), downloaded 12 November 2026.
- Predicted lines: our road-based model, using gridfinder (Arderne, Zorn, Nicolas and Koks 2020, Predictive mapping of the global power system using open data, Scientific Data, CC BY 4.0), OpenStreetMap roads and settlements (ODbL) and the substations on this map.
The map is free to view. The data is available on request: email [email protected] and tell us what you need it for. Last built 12 November 2026.
Back to S\u00e3o Tom\u00e9 and Pr\u00edncipe grid and access data