Kwekwe block 3

An approximately 93 ha eight-peg block near Kwekwe, about 4 km south-east of the first Kwekwe block. The corner order supplied was self-intersecting, so the boundary shown here is the polar-ordered approximate footprint of the eight pegs. 77 known gold mines fall within 40 km. This is a where-to-walk assessment on free satellite data, not a claim that gold is present.

One caution on the boundary itself: three of the eight pegs (D, F, G) sit inside the outer hull of the others, and D and G are only ~65 m apart, so the true claim shape is genuinely ambiguous (one irregular block, or two adjacent blocks, read differently). Plausible boundaries range from about 71 to 107 ha. The NE corner (pegs A, B, C) is unambiguous and holds the two walk-first spots, but spots 3 and 4 sit within 10 m of the approximate western boundary and could fall outside the actual claim. If the supplier can confirm the intended peg order, or whether this is one block or two, the footprint and the whole-block percentile should be re-run.

Walk the NE corner first. Start at 788416 E / 7906777 N (UTM 35S), the rank-1 spot at P(gold) 0.67, roughly 76 m from peg C. Walk the 180 m arc to the rank-2 spot at 788586 E / 7906717 N (P(gold) 0.60, ~72 m from peg B). Both spots sit in the same NE pocket where terrain and the fused model converge at their highest values on this block; a single traverse covers them before committing time to the southern edge or interior control points.

As a whole block the fused model places it at the 36th percentile of the local background (P(gold) 0.285, 80 m grid; 0.277 at 10 m native), which is background-level prioritization. The terrain arm is the outlier: the block sits at the 97th percentile of the terrain signature of 52 known gold mines within 25 km (LOO AUC 0.944), a stronger single-arm terrain reading than any earlier block in the set. The iron-oxide arm reads at the 80th percentile, but the mine calibration check shows that the 52 known mines in this scene are themselves only at the 43rd percentile in ferric ratio (separability -0.119 sigma), so elevated ferric here most likely reflects laterite and farmed red soil rather than gossan. The oxide and fused signals read as background-level; terrain is the signal worth following.

The best-ground spots (UTM 35S). Spots 1 and 2 are the NE corner target (purple); spots 3 and 4 are the southern edge cluster (amber); spots 5 and 6 are interior scatter points included for coverage and calibration (grey):

The four checks behind that

Each is scored over the block against the ground immediately around it.

Check This block What it means
Terrain match 97th percentile the strongest single-arm terrain reading in the set; the block geometry closely matches the ensemble of 52 known gold mines within 25 km (LOO AUC 0.944, spatial CV at 8 km holdout 0.884)
Red-oxide (ferric) 80th percentile reads above background, but the 52 known mines in this scene are themselves only at the 43rd percentile in ferric ratio (separability -0.119 sigma); elevated ferric here most likely reflects laterite and farmed red soil, not gossan
Structure / lineaments 55th percentile slightly above background; dominant trend 150-165 degrees; farmed flat terrain caveat applies, as roads and field edges are the dominant false positive in cultivated ground
Regional P(gold) 0.285 whole-block (36th percentile); NE corner P(gold) 0.60-0.67 the combined estimate; background-level overall, with the NE corner as the sole above-average patch

Compared to your other blocks

Same estimate, run over each block:

Block P(gold) median Percentile vs local background
Belingwe-West 0.394 n/a
Zvishavane 0.282 67th
Kwekwe block 2 (whole block) 0.254 45th
Kwekwe block 2, NW corner spots 0.65-0.70 n/a
Kwekwe block 3 (this), whole block 0.285 36th
Kwekwe block 3 (this), NE corner spots 0.60-0.67 n/a
Kwekwe block 1 0.129 8th

Whole-block, kwekwe-3 sits just above kwekwe-2 in raw probability (0.285 vs 0.254) but below it in percentile rank (36th vs 45th) because the two blocks sit in different local distributions. The NE corner of kwekwe-3 reaches the same range as the NW corner of kwekwe-2 (0.60-0.67 vs 0.65-0.70). The first Kwekwe block remains the weakest in the set. The terrain arm here (97th percentile) is the strongest single-arm terrain reading across all blocks scored; its divergence from the 36th-percentile fused result is a notable signal rather than a confirmation.

Data used

The P(gold) grid is 10 m (~0.01 ha cells), so it pins the corner to walk, but it reads surface geology. It tells you where to look, not where a vein is. The single most useful next step is ground truth from the NE corner pair, around 788416 E / 7906777 N (UTM 35S): a handheld pXRF reading on outcrop or soil, or simple pan-and-grab sampling of surface material, on the one 180 m traverse. That single visit tests whether the terrain arm or the embedding model is right here, and recalibrates confidence for every Kwekwe-area block at once.