Much of the Limpopo Belt is covered by thin soil, savanna and younger cover, and field access is patchy. Yet its economic geology depends on structure: where major shear zones sit, how they link, and which smaller faults branch from them. A geological map drawn at 1:250,000 will show the big shear zones but few of the second-order structures that you can drill. This article shows how to extract linear features from a free digital elevation model (DEM), turn them into a directional dataset, and test whether they reflect real basement structure. It also covers the limits of what a surface-only dataset can tell you about deep crust.

The principle: why deep structures leave a surface trace

Plainly put, big shear zones are zones of weaker, differently textured rock. Erosion treats them differently from their neighbours, so they often appear as straight valleys, aligned stream segments, scarps or abrupt changes in ridge direction. A lineament is any such straight or gently curved feature on a landscape. It is a description, not an interpretation: a lineament may be a fault, a dyke, a joint set or a fence line.

The Limpopo Belt is a good place to practise this because its structure is large and well documented. It lies between the Zimbabwe and Kaapvaal cratons and is divided by major shear zones into the Central Zone, the Northern Marginal Zone and the Southern Marginal Zone. The Central Zone is described as an allochthonous thrust sheet (one moved in from elsewhere), bounded by the Tuli-Sabi shear zone in the north and the Palala shear zone in the south. The scale is considerable: the Palala Shear Zone is part of a lineament extending over 1,000 km, from central Botswana to the Soutpansberg. At its best exposure, it is a mylonite zone about 15 km wide (mylonite is rock that has been intensely ductilely sheared).

Now the technical caveat. A DEM sees only the surface. Depth information comes from other methods. Magnetotelluric data image the composite Sunnyside-Palala-Tshipise shear system, 20–30 km wide, as a sub-vertical conductive structure, and gravity data place the southern boundary between the Kaapvaal Craton and the Limpopo Belt at the Hout River shear zone. A DEM lineament map therefore does not "see" the deep crust. It records where deep structures, or later reactivations of them, have shaped the landscape. Your task is to link that surface pattern to the deeper evidence.

The workflow

Choose the DEM first. Copernicus GLO-30 gives global coverage at 1 arc-second, about 30 m. Its absolute vertical accuracy is stated as better than 4 m at 90% linear error. It is a digital surface model, so it includes vegetation and buildings, not bare ground. It derives from TanDEM-X radar data acquired between December 2010 and January 2015. The older SRTM, flown in 2000, is the usual alternative. Comparing the two helps you spot artefacts. Digital Earth Africa hosts GLO-30, and SRTM is available through USGS EarthExplorer.

1. Mosaic and reproject. Merge tiles and convert to a projected coordinate system (UTM zone 35S or 36S for most of the belt) so distances and angles are in metres and degrees. Do this in QGIS, GDAL or Python.
2. Fill and smooth lightly. Fill voids and apply a mild low-pass filter, such as a 3 × 3 mean, to suppress radar speckle without erasing narrow valleys.
3. Make multi-azimuth hillshades. Shaded relief hides features that run parallel to the light direction. Generate at least four illuminations (0°, 45°, 90°, 135°) at a low sun altitude, around 30–45°. GDAL's gdaldem does this in one command.
4. Add derivative layers. Slope, curvature and flow-accumulation channels reveal straight drainage segments independently of illumination.
5. Extract lineaments. Trace them by hand for control, or use an edge detector (Canny, or the LINE algorithm in commercial software) followed by vectorisation. Record length and azimuth for each.
6. Filter and statistics. Remove short segments, set a minimum length that suits your DEM resolution, and plot a length-weighted rose diagram (a circular histogram of azimuths).
7. Compare with other data. Overlay geological maps, aeromagnetic and gravity grids, and published shear-zone traces.

Worked example (illustrative only)

This example is hypothetical and not drawn from any real project. Imagine a 60 × 60 km tile on the boundary of a high-grade gneiss terrane, with a thin cover of Quaternary sand. At 30 m per pixel the tile is about 2,000 × 2,000 cells.

You produce four hillshades and trace 1,240 candidate lineaments. After discarding anything under 2 km (about 65 pixels, long enough to be unlikely noise), 310 remain. The rose diagram, weighted by length, shows two clusters: a dominant set trending about 070° (ENE) holding roughly 45% of total length, and a weaker set near 340° holding about 20%.

The ENE set is consistent with the regional trend described for the belt: the Limpopo Complex has an ENE–WSW trend. You then overlay a magnetic grid. If the ENE lineaments coincide with linear magnetic breaks or offsets, that supports a basement origin. If the 340° set aligns with narrow, strongly magnetic linear highs, it may be a dyke swarm. The belt hosts later intrusive rocks, including the Great Dyke and its associates, and younger dyke swarms. In that case you would reclassify those lineaments as dykes, not shear zones. Finally, you would pick two or three intersections of the ENE and 340° sets as sites for ground checking, not as drill targets.

Common mistakes and limitations

• Illumination bias. One hillshade direction exaggerates features perpendicular to the light and hides parallel ones. Always use several azimuths and check that your rose diagram is not simply mirroring your light angles.
• Treating every lineament as a fault. Roads, fences, field boundaries, dykes, bedding and joint sets all create lineaments. Remove cultural features by comparing with satellite imagery.
• Surface-model artefacts. A DSM includes tree canopy. Dense vegetation can create texture that edge detectors pick up. Radar DEMs also contain striping and void-filling seams.
• Resolution mismatch. A 30 m DEM cannot resolve structures narrower than a few pixels. Absence of a lineament does not mean absence of a fault.
• Confusing age and depth. Landscape lineaments may record much later reactivation. In the Palala system, two phases of movement are recorded, at 2.7–2.6 Ga and 2.0–1.8 Ga, so surface expression may reflect only the younger event.
• Over-interpreting statistics. Do not read a dominant direction as a kinematic result. Orientation says nothing about the sense of movement.

To check your results, compare against independent datasets: aeromagnetics (the best single check for basement structure), gravity, published geological maps and field measurements. Repeat the extraction with a different DEM and a different operator or algorithm. If the main azimuth clusters survive, they are probably real. Finally, test the lineaments against what you know: do they cluster along known shear-zone traces, and does their density change across mapped lithological boundaries?

Key points to remember

• A lineament is a descriptive observation, not a fault. Interpretation needs independent evidence.
• A DEM records surface expression. Gravity, magnetics and magnetotellurics tell you about depth.
• Use several illumination azimuths and derivative layers, and filter by length.
• Use rose diagrams to compare azimuth populations with known regional structural trends, and with dyke and cultural patterns.
• Treat lineament intersections as candidates for field checking, not as finished targets.

Sources

• The Limpopo Belt is subdivided into three zones: the Central Zone (CZ),... (researchgate.net)
• Shear zones bounding the central zone of the Limpopo Mobile Belt, southern Africa - ScienceDirect (sciencedirect.com)
• The deep structure of the Limpopo Belt from geophysical studies - ScienceDirect (sciencedirect.com)
• Tectonic model of the Limpopo belt: Constraints from magnetotelluric data - ScienceDirect (sciencedirect.com)
• Gravity evidence for a larger Limpopo Belt in southern Africa and geodynamic implications (academic.oup.com)
• Copernicus DEM (COP-DEM) — Digital Earth Africa 2026 documentation (docs.digitalearthafrica.org)

About Orex — Orex is a mineral exploration intelligence platform based in Mwanza, Tanzania. We combine satellite remote sensing, elevation-derived structural analysis and open geoscience data to help explorers, licence holders and investors focus their fieldwork on the ground that matters.

Want to see fault structures and intersection targets on your area of interest — for free? Install GoldRadar on your phone or desktop: it maps lineaments and automatically flags fault intersections derived from satellite elevation data, giving you a structural framework for preliminary exploration before you spend a dollar on the ground.