Ground-penetrating radar (GPR) is often presented as the answer to unmarked graves: roll a cart over the site and the burials appear on screen. Used well, it is one of the best non-intrusive tools available for finding graves. But it does not see skeletons, and in the wrong conditions it can be confidently wrong.
At Koffiefontein Mine in the Free State, where an estimated 600 people were buried in historic mass graves, we put GPR to a direct test. The results changed how we use it on every project since.

How GPR finds graves
A GPR antenna sends short pulses of radio energy into the ground and records the echoes that bounce back from changes below the surface. A grave shaft, a coffin or disturbed backfill can produce a distinctive reflection, known as an anomaly. By surveying a grid in closely spaced lines, the data can be combined into “time slices” that show what the ground looks like at a given depth.
The choice of antenna matters. Higher frequencies give sharper detail but do not reach as deep, and they pick up more clutter from stones, roots and rubbish. Lower frequencies penetrate further and produce cleaner results, at the cost of fine detail. Soil matters just as much: clay-rich or saline ground absorbs the signal, and wet ground changes how it travels.
The Koffiefontein test
In June 2022 Coen Nienaber, the GPR specialist we contracted for the project, surveyed three areas of the mine where there was good reason to expect graves. Grids of between 5 × 4 m and 8 × 10 m were surveyed in lines 50 cm apart with a dual-frequency (300 and 800 MHz) GPR system. Every promising anomaly was then checked by archaeological excavation, a step known as ground truthing.
| Survey grid | Why we expected graves | What the GPR showed | What excavation found |
|---|---|---|---|
| Grid 5 | Historic maps showed the edge of the old cemetery extending here | Two grave-shaped anomalies | No graves. Natural calcrete deposits and geology |
| Grid 7 | Human remains had previously been found in these mine dumps | Anomalies in neat rows, like a cemetery layout | No graves. Differences in compaction and drainage caused by heavy machinery on the tailings |
| Grid 11 | Burials had just been uncovered during construction | Nothing grave-like | Graves present, with human remains in situ |
In two areas the radar pointed to graves that were not there. In the third, where graves were confirmed, it showed nothing at all. On this site, GPR on its own would have sent a project in exactly the wrong direction.
Why the radar struggled
- Kimberlite gravel. The mine ground is full of highly reflective kimberlite, which produces strong echoes that look like features.
- Disturbed ground. Tailings dumps, construction fill, drainage pipes and buried cables create clutter. At Grid 11 the graves lay beneath 60–80 cm of compacted gravel and clay used as a vehicle yard.
- Waterlogged soil. Heavy rain before the survey left the ground wet and, in places, waterlogged.
- Old, degraded burials. After many decades, graves may offer very little contrast with the surrounding soil.
- Equipment and settings. The antennas and processing used were not well suited to these conditions.
What we changed
Koffiefontein led us to change both our equipment and our method. We now use our own Sensors & Software Noggin 250 MHz system on a SmartCart. Its lower frequency reaches deeper and produces less clutter in difficult ground. Just as importantly, we treat GPR as a screening tool, not a verdict:
- We assess ground conditions before promising results, and survey when the ground is drier where the programme allows.
- We never sign off an area as clear, or confirm a grave, on radar alone.
- We ground-truth anomalies with controlled archaeological excavation before permits are applied for or budgets are fixed.
When it works, it pays for itself
In the right conditions GPR is invaluable. On our recent N8 expansion project, the original assessment identified 8 graves. After GPR work it was determined that there were in fact 36. Finding them before construction, rather than during it, meant the relocation could be planned and budgeted properly.
What this means for your project
- A GPR survey tells you where to look, not what is there. Budget for ground truthing.
- Mine land, tailings, fill and waterlogged or clay-rich soils reduce reliability. Ask how the specialist will account for this.
- A “clear” radar result is not a clearance. Unmarked graves can still be present.
- The combination of GPR, archival research, community knowledge and targeted excavation is what gives you a reliable grave count.
The Koffiefontein GPR survey was carried out for G&A Heritage by Coen Nienaber in June 2022.
Need to know what is under your site before you build? Request a site assessment.