Kwekwe block: a free-EO peg assessment

A second case study after the Zvishavane claim. A holder supplied a 157 ha block near Kwekwe in the Midlands greenstone belt, available to peg, chosen on the "red oxides, like the neighbouring mines" thesis, and asked how it performs on free satellite data. Kwekwe is a genuinely high-production district: there are 85 known gold mines within 40 km (the biggest neighbour produced 124 tonnes), and they are the local reference this analysis is built on.

Three independent folia-native methods, no Earth Engine: red-oxide (ferric) mapping, an explicit DEM terrain-signature match to the surrounding mines, and the validated AlphaEarth embedding analog. The map below carries all of them. The detail is in the Zvishavane method pages; this page is the verdict for the block.

Bottom line: free EO gives no peg-worthy signal here. The block sits in a favourable district and on mine-like terrain (right neighbourhood), but by the strongest validated method its surface signature reads as ordinary country rock, background (49th to 57th percentile depending on resolution), best corner ~0.2 against the 0.7 gold band. The red-oxide rationale is not corroborated: the block is not iron-anomalous, the nearby mines are themselves only weakly ferric in the imagery, and hyperspectral EMIT finds no clean iron-oxide mineral (hematite/goethite) over the block. One limit frames all of it: every method reads at pixel scale, 10 m (Sentinel-2), 80 m (AlphaEarth), 60 m (EMIT, spectrally the sharpest but spatially the coarsest), so this is evidence against a block-scale gold system, not proof that a narrow vein or a small gossan outcrop, the scale you see in the field, is absent. The spots below are best-case corners to walk first, never gold.

How this block compares to the first study (Zvishavane). By the AlphaEarth embedding analog (the strongest cross-validated method here, and the one common to both studies, the DEM terrain arm, also cross-validated at AUC 0.87, was added only for this block), the original Zvishavane claim scored higher: 71st percentile, best corner 0.50 (about two-thirds of the way to its gold band) versus this block's 49th percentile, best corner ~0.2 (about a quarter). So the original claim resembled gold ground more. The flip side: Kwekwe is the better district (known mines ~4 km away vs ~11 km, denser, with a 124-tonne producer), and this block's negative is the more robust one, the ground here is fully exposed, so Sentinel-2, EMIT, and the embedding all got a clean look and all agree, whereas Zvishavane's higher score rested on the embedding alone because canopy blinded the optical arms there. Better parcel: Zvishavane. Better district: Kwekwe. Neither is a deposit on free EO alone.

The block spots, as a list (UTM 35S, the holder's grid):

What the map shows

Embedding P(gold), 10 m (on by default) is the validated method: how much the ground resembles known gold sites in AlphaEarth's 64-D satellite embedding, scored by a RandomForest trained on the 85 local mines (scored at AlphaEarth's native 10 m for the block). Over the block it is dim, the block reads like background. This is the layer to trust.
Red-oxide ferric ratio (S2 B04/B02) is the holder's thesis layer. It is not anomalous over the block, and (the test that matters) the known mines are only weakly ferric (nearest-pixel +0.26σ, 60th pct), so a bright patch here is weak evidence at best.
Gold mines sized by production. The block sits among real producers: the district is favourable. That is the strongest thing in the block's favour, and it is about the neighbourhood, not the block.
Spots are each method's best-case ground inside the block. They cluster in the north / north-west corner, which is the only constructive read here. None is a gold indication; all are walk-first-with-low-expectations points.
Lineament density (DEM, off by default) is the structural arm: candidate shears from a 30 m DEM. The block reads mildly above-average (72nd percentile) with a NW-SE grain (the real Midlands structural trend), but the NW corner the other arms liked is only 58th percentile, so structure does not reinforce it. And the terrain is very flat (95 m of relief over 8 km), so most of these linears are field edges and tracks, not bedrock. Its top lineament highs are on the map as blue structural candidates (coordinates in the table) for completeness, but this is the lowest-confidence layer here, unlike the rugged Zvishavane belt where structure was a primary target.

How this was made, and how it was validated

Each method trains on the 85 local mines and reports leave-one-out skill and a spatial cross-validation, the test for whether the skill is real or just spatial autocorrelation (clustered mines memorised by the model):

Method	Leave-one-out	Spatial CV (8 km buffer)	Block verdict
Embedding analog	AUC 0.94	0.90 (holds, strong)	49th percentile, background-like
DEM terrain	AUC 0.91	0.87 (holds, coarse)	85th percentile (mine-like terrain type)
Red oxide	mines weakly ferric (+0.26σ pixel)	n/a	not anomalous (40th percentile)

The embedding skill holds up strongly under spatial cross-validation, so the block's background-like score is trustworthy, not an artefact. The terrain signal is real but coarse (the mines sit on slightly more dissected greenstone terrain than the granite plateau, a "right terrain type" filter, necessary not sufficient). Recipes: research/mineral-prospecting/kwekwe-block/ (embedding.py, dem_terrain.py, iron_oxide.py).

Soundness check. Because known mines are physically disturbed ground (pits, tailings) and AlphaEarth embeds surface appearance, a ring test checked whether the model scores the disturbed footprint or real geology: P(gold) around the mines decays from 0.30 at the pit to 0.10 at 1.5 km but stays 5 to 7x background throughout, so the signal is largely areal greenstone-belt geology, not a mining artefact (a modest disturbance component remains). The block scores below even that 1.5 km mine-halo, so its background-like reading is not just an artefact of the block being un-pegged. The coordinate decode was independently validated against Globe & Phoenix and Cam & Motor (the latter's grade matches the published value exactly). Recipe: research/mineral-prospecting/kwekwe-block/validate.py.

Hyperspectral cross-check (EMIT). Because the holder's thesis is red oxides, I pulled a dry-season, cloud-free EMIT scene (285 contiguous bands, 60 m) over the block, the one instrument that can actually name the iron mineral. On bare ground the ferric (~900 nm) absorption is weak and mis-positioned (~958 nm, outside the hematite ~880 / goethite ~920 range), so no clean iron-oxide mineral is resolved. Al-OH (clay) and a 2330 nm feature are present but ambiguous (ordinary soil clay and dry crop litter on this farmed plateau), not a confirmed alteration assemblage. So the best available instrument does not find the red oxides either. Recipe: research/mineral-prospecting/kwekwe-block/emit_mineralogy.py.

The single biggest lever from here is one real ground observation from the block or a neighbour: it wires straight into the iron-oxide calibration and the embedding reference, and re-weights every layer against real data.