Prospectivity
ni-v2.0
Magmatic nickel sulphide, Selebi-Phikwe and Tati
All runs Ranks ground
Mineral system
Ni Nickel Definition 1.1
Magmatic nickel sulphide, Selebi-Phikwe and Tati
Signed by K. Dlamini, FGSSA on 18 Aug 2026
Gradient-boosted trees, magmatic sulphide model · 200 m
Ran 04 Sept 2026, 110 min
Run by: Amini modelling team · Accepted by: K. Dlamini, FGSSA
Capture efficiency
The top 10% of ranked ground holds 71% of the known deposits.
Search-space reduction 95.6%
The share of the modelled area a competent person can set aside for this commodity.
Validation
Blind-site hit rate 75%
Threshold agreed at MS1 65%
Block cross-validation 5 folds, 20 km blocks
AUC 0.85 ± 0.05
Physical constraints
- Prediction requires a mafic-ultramafic host: sedimentary cover sequences are excluded.
- A conductor without a coincident gravity high cannot score high on its own.
Features and importance
| Feature | Critical process | System role | Importance |
|---|---|---|---|
| EM conductor amplitude | Trap: massive sulphide accumulation is conductive | Trap | |
| Residual gravity high | Trap: dense sulphide body against silicate host | Trap | |
| Mafic-ultramafic host proximity | Source: the magma that carried the nickel | Source | |
| Basal contact geometry | Transport: sulphide settles at an embayment on the basal contact | Transport |
Inputs
| Dataset | Version | Confidence filter | Coverage |
|---|---|---|---|
| Selebi-Phikwe airborne EM | 2016 | A and B only | 67% |
| National gravity | 2021 | A and B only | 90% |
| BCL historical drilling | 2019 digitisation | assayed intervals, C or better | 23% |
| Geology 1:250k | 2023 | all, lineage recorded | 100% |
BCL historical drilling covers 23% of the modelled area and is the layer that governs the bounds.
Validation
| Withheld site | Withheld by | Class | In the high class |
|---|---|---|---|
| BS-NI-01 Selebi North deeps | K. Dlamini, FGSSA | High | In the high class |
| BS-NI-02 Phikwe west limb | K. Dlamini, FGSSA | High | In the high class |
| BS-NI-03 Dikoloti | K. Dlamini, FGSSA | Medium | Missed |
| BS-NI-04 Lentswe | K. Dlamini, FGSSA | High | In the high class |
Outputs
- Prospectivity surface 2.0 · COG, 200 m
- Confidence surface 2.0 · COG, 200 m
- Feature-importance maps 2.0 · COG per feature
Parameters
- learningRate
- 0.05
- maxDepth
- 5
- trees
- 700
- classCut
- high ≥ 0.77, medium ≥ 0.52
- negatives
- 3:1, barren mafic ground
Limitations and completeness
- EM coverage is 67% of the belt and stops at the 2016 survey boundary.
- BCL drilling is digitised from paper logs; depth registration carries about 2 m of error.
Against the previous edition
This is the first edition of this model.
Targets from this run
2 targets are ranked from this run.
| Target | Score | Class | Competent person |
|---|---|---|---|
| SPK-NI-007 Selebi North Offset | 0.84 | High | K. Dlamini, FGSSA |
| OKV-NI-002 Xade Complex Margin | 0.52 | Low | K. Dlamini, FGSSA |