30 Day Map Challenge 2026 - Day 2 (Icon Map Pro) - Lines

How does South Africa's electricity network connect the country, and what can its geography tell us?

For Day 2 of the 30 Day Map Challenge, the theme is Lines. After using population and supermarket locations for Day 1's Points map, I wanted to explore a very different type of spatial data: the networks that link places together.

Electricity infrastructure is an especially interesting example. Long-distance, high-voltage transmission corridors connect distant parts of South Africa, while lower-voltage lines form more local networks around towns and cities. When displayed together and coloured by voltage, these lines reveal a structure that is difficult to appreciate from a conventional table.

The Microsoft Power BI report uses the Icon Map Pro visual, with electricity lines at the centre and substations and power generation facilities available as supporting context. The key technical feature is PMTiles imported directly through the newer Shapes from Files functionality in Icon Map Pro. That lets the report display detailed network geometries without storing the line coordinates in the Power BI semantic model.

Explore the interactive report

Use the Province slicer to focus on a region, compare mapped network lengths in the voltage-band chart, and zoom in to discover substations and power stations. The coloured lines represent voltage categories, not electricity flows or available network capacity.

*Interactive report: Day 2, Powering South Africa.

Starting with OpenStreetMap's electricity network

The electricity lines and substations come from OpenStreetMap, using the South Africa extract supplied by Geofabrik.

Rather than download a handful of lines through a web query, I used Osmium Tool to filter the national .osm.pbf file for power=line and power=minor_line. Those tags cover major overhead lines and smaller lines mapped by OSM contributors.

The cleaned source contained 12,936 electricity line features, with approximately 87,731 km of mapped line geometry. These are mapped route kilometres. They are not necessarily circuit kilometres, and the extract is not an authoritative inventory of every electricity line in South Africa.

Making sense of voltage tags

OSM records voltage in volts, sometimes with multiple values on a single feature. Other lines have no usable voltage information at all.

After normalising the available voltage values to kV and assigning each line to one of eight display bands, ranging from below 10 kV to 550 kV and above, with a separate Unknown category. The map and chart use the same category colours, making it easier to compare the geographical pattern with the reported line lengths.

From network geometry to province-level analysis

I also wanted visitors to be able to explore the network one province at a time.

For the final report, I used the nine province boundaries from OpenStreetMap, rather than a heavily generalised boundary dataset. I extracted the administrative relations with admin_level=4 and kept their complete multipolygon geometries.

The electricity lines were then intersected with the provincial polygons. A line crossing a provincial boundary was split into separate portions, each with its own province identifier and calculated length.

This step matters: simply assigning the entire line to the province containing its midpoint would distort regional kilometre totals.

The resulting dataset contained 13,262 province-split line segments. Approximately 86,335 km intersected the province polygons. The remainder was not assigned, so I have avoided presenting the provincial totals as an official network inventory.

Each processed segment received a unique segment_id. That identifier became the link between its geometry in PMTiles and its attributes in Power BI.

Bringing PMTiles into Icon Map Pro with Shapes from Files

Instead of loading over 13,000 detailed line geometries as WKT text into the Power BI dataset, I used Tippecanoe to turn the province-split GeoJSON into a PMTiles vector-tile archive.

Crucially, Icon Map Pro's newer Shapes from Files functionality can import PMTiles directly. I used that route in the report, rather than treating the PMTiles archive as a basemap or relying on an externally hosted tile URL.

The vector tiles contain the line geometry and the segment_id needed for matching. The Power BI table contains the analytical attributes, such as province, voltage band, line colour and length. Icon Map Pro matches the tile features to their Power BI records using the shared identifier.

That separation delivers a practical advantage: the map can render detailed lines while Power BI slicers and measures operate on a conventional, much smaller set of attributes.

A simple optimisation saved 8 MB

My first PMTiles export was 26.8 MB, a little too large for the file-upload limit I was working with. I initially considered reducing maximum zoom or simplifying the lines more aggressively, but neither was necessary.

The source GeoJSON contained properties already represented in Power BI, so rebuilt the PMTiles archive with only the essential properties, including the feature identifier, rather than copying every source attribute into each tile.

That reduced the PMTiles file by approximately 8 MB, to about 18.8 MB, while retaining the same zoom range and without introducing additional geometry simplification. I could keep the detailed linework and the existing voltage styling.

The lesson is simple: before simplifying spatial data to make a file smaller, check whether you're duplicating attributes that already live in your semantic model.

One map for lines, substations and power stations

To add context without obscuring the Lines theme, I also prepared two point datasets:

  • Substations, extracted from OpenStreetMap, with representative coordinates and voltage-tier SVG markers.
  • Power generation facilities, taken from Global Energy Monitor's Global Integrated Power Tracker, with generating technologies and project statuses.

For the generation data, I kept unit/phase records for analytical totals but consolidated them into plant-location markers. Records flagged with TransitionZero identifiers were excluded from the report.

The final mapping table contains separate record types for electricity lines, substations and power plants. The line rows carry a feature reference for the PMTiles geometry; the point rows carry coordinates and encoded SVG data URIs. This allows the same Icon Map Pro visual to display all three without repeating the line coordinates in Power BI.

I sized power plant markers at 40 × 40 pixels and substations at 20 × 20 pixels, using zoom-dependent visibility to keep the national view focused on the network. The accompanying chart shows mapped network length by voltage band, while the Province slicer filters the network geographically.

What the map reveals

The strongest visual impression comes from the contrast between long-distance corridors and more closely spaced regional networks.

The higher-voltage routes create connections across large areas of the country. In and around major urban regions, particularly Gauteng, the mapped network becomes denser and more interconnected. Other parts of the country show long stretches of infrastructure separated by comparatively sparse local linework.

The chart adds another perspective by comparing the mapped route length within each voltage category. Selecting a province helps reveal how that balance varies regionally.

These are patterns in OpenStreetMap's recorded data. A sparsely mapped area does not necessarily have a sparse electricity network: differences in contributor coverage and missing voltage tags can also affect the result.

What I learned from Day 2

Three practical lessons stood out.

1. Lines require a different data strategy from points. Keeping geometry in vector tiles and the useful attributes in Power BI proved more manageable than loading large WKT strings into the semantic model.

2. Administrative segmentation needs proper geometry operations. Splitting lines at provincial boundaries makes region-level network lengths much more defensible than assigning entire routes to a single province.

3. Retain detail before reducing it. The PMTiles archive became small enough to import simply by removing redundant properties. The geometry did not need to be simplified further.

Final thoughts

A line on a map is not just a shape. As part of a network, it connects places, crosses administrative boundaries and helps explain how infrastructure is organised geographically.

For this challenge, Icon Map Pro and the Shapes from Files PMTiles import made it possible to explore South Africa's mapped electricity network in Power BI, with provincial filtering and consistent voltage styling, without placing the detailed line geometry inside the data model.

Next time you encounter a large road, rail, pipeline or utility network, it is worth considering the same separation between geometry and analytical attributes.

Sources and attribution

Methodology note: the figures represent mapped OSM route geometry and derived provincial statistics, not official transmission or distribution network totals. The map does not show electricity flow, grid congestion or available connection capacity.