Silobela block 3

A four-peg block of 29.0 ha (the supplied pegs compute to exactly the stated 29 ha, which confirms the UTM 35S reading), immediately north-east of the Silobela site 2 point — peg B sits ~320 m from the site 2 coordinate. Context from the holder: sampling around site 2 averaged about 1 g/t Au, but that ground is already pegged by others, and this block is the only free ground available nearby. The question asked was whether this block is worth pegging. 36 known gold mines fall within 40 km; the nearest confirmed producer is ~2.9 km away. This is a where-to-walk assessment on free satellite data, not a claim that gold is present.

Worth it — this block contains the best spot we found in the entire site 2 analysis. When site 2 was assessed, the strongest ground in its 1 km² window was not at the supplied point but ~310 m NNW of it, at P(gold) 0.73 at 733161 E / 7911440 N (UTM 35S). That spot falls inside this block, ~200 m east of peg A. In other words, the free ground is not a consolation prize: the model's pick of the site-2 neighbourhood is on it.

Whole-block, the fused model reads a median P(gold) of 0.346, the 69th percentile of the local background — statistically the same ground as the site 2 window (median 0.359, 71st percentile), the ground that assayed ~1 g/t. Walk first from peg A ~200 m east to 733161 E / 7911440 N (P 0.73), then the ~240 m line east to 733401 E / 7911440 N (P 0.50). The remaining spots (0.39–0.46) string eastward along the block's long axis, so one ~1 km traverse peg A → peg D covers all six.

Ground truth arrived (2026-07-30), and it backs the pick. The holder walked the walk-first ground and took five rock grabs from 733150 E / 7911411 N, 31 m from spot #1 (P 0.73; the model reads 0.571 at the sampled point itself, the 99.2nd percentile of the local background). Fire assay, report #0034, clean QC (blank ND, 1 ppm control read 1.01): 2.33, 1.52, 1.34, 0.16, 0.02 g/t Au. Three of five grabs ran at or above 1.34 g/t, against the roughly 1 g/t neighbourhood average previously reported from the pegged ground next door. This is the first grade-bearing test of a model pick in this study, and the highest grades reported anywhere in the Silobela work came from it. Two honest caveats: the spot was sampled because the model ranked it, so this validates walk-first targeting rather than block-wide grade, and the 0.16 and 0.02 g/t grabs from the same point are the usual nuggety-oxide scatter. The missing control is 2 or 3 grabs from a low-P patch on the same block (the SW corner near peg B reads P below 0.2): if those also run over 1 g/t, the model gets no credit, the whole belt does. The holder registered a 25 ha block on this ground the following week. Log: research/mineral-prospecting/in-situ/ASSAYS.md.

Every 10 m pixel over the block (and the ground around it, including the site 2 point) is inspectable: the heatmap is the fused P(gold) surface (dark = low, bright = high), and tapping anywhere shows the score, its rank against the local background and within the block, and every layer behind it, with the coordinate in peg format. The low ground is as visible as the high ground on purpose: control samples from dark pixels (for example the SW corner near peg B) are what calibrate this model. Scores read surface geology, not grade.

The best-ground spots (UTM 35S). Spots 1 and 2 are the walk-first pair at the west end near peg A (purple); spots 3 to 6 string eastward along the block (grey), so a single peg A → peg D traverse covers all six:

The five checks behind that

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

Check This block What it means
Regional P(gold) 0.346 whole-block (69th percentile); best spot 0.73 the combined estimate; the same statistical ground as the site 2 window (0.359, 71st), and the top spot from the site 2 analysis sits inside this block
Terrain match 88th percentile the block matches the terrain signature of 16 known gold mines within 25 km (LOO AUC 0.895; spatial CV at 8 km holdout 0.791)
Red-oxide (ferric) 45th percentile background; fully bare ground (0% vegetated), so the reading is clean, there is simply no oxide anomaly at surface
Structure / lineaments interior at the 0th percentile no DEM-scale lineament density crosses the 29 ha interior; at this block size the lineament arm has little resolving power, and the nearest density highs sit just outside the boundary
ASTER alteration (SWIR) Al-OH 56th, carbonate/Mg-OH 29th, chlorite 69th percentile archive 2006 scene, 0% cloud, fully bare. Nothing anomalous, and the mine calibration on this scene is weak (Al-OH separability +0.16 sigma), so ASTER neither supports nor damages the case here

Compared to your other ground

Same estimate, run over each site:

Site P(gold) median Percentile vs local background
Belingwe-West 0.394 n/a
Silobela site 2 window (~1 g/t assays) 0.359 71st
Silobela block 3 (this) 0.346 69th
Silobela site 1 window 0.331 63rd
Kwekwe block 3 0.285 36th
Kwekwe block 2 0.254 45th
Kwekwe block 4 0.238 34th
Kwekwe block 1 0.129 8th

The read in one line: this block is statistically indistinguishable from the ground next to it that assayed ~1 g/t, it holds the single best model spot of the whole Silobela neighbourhood, and only Belingwe-West ranks above it in this study. Of free ground available to peg, this is the strongest case the model has made.

One honesty note: the ~1 g/t figure belongs to samples taken around site 2, not on this block. The model says the surface here looks the same; only samples on this ground can say the grade carries across. But 29 ha is a small, cheap block to test: the peg A → peg D traverse with pan-and-grab or pXRF sampling at the six spots settles it in a day.

Data used

The P(gold) grid is 10 m (~0.01 ha cells), so it pins the spot to walk, but it reads surface geology. It tells you where to look, not where a vein is.