Northern Territory Zinc-Lead Prospectivity Concordance and Critical Co-Product Evidence Mapping
A completed geology study comparing published Northern Territory clastic-dominated and Mississippi Valley-type zinc-lead prospectivity surfaces and auditing critical co-product evidence.

Software compatibility
The completed workflow uses checksummed Northern Territory extracts from the Geoscience Australia Critical Minerals Mapping Initiative and the pinned Python 3.14 packages. Replacing the source data, deposit classes, region, or thresholds requires a new provenance and interpretation review.
Project definition
Problem statement
Published regional prospectivity models can overlap, disagree, and inherit some of the same training information. Treating a high-scoring cell as a discovery ignores model lineage, retained-area rules, spatial dependence, and geological uncertainty.
The geology problem is to compare two related zinc-lead model surfaces, measure their agreement and sensitivity, and separate critical co-product data coverage from any claim of an economic deposit.
Project objectives
- Verify and retain the Northern Territory CD, MVT, and deposit-geochemistry extracts with exact hashes.
- Measure shared cells, union cells, Jaccard agreement, percentile concordance, and published geological attributes.
- Test descriptive capture of represented target deposits at 10, 25, and 50 km distances.
- Build connected high-priority screening clusters and retain their area, support, centroid, and score evidence.
- Audit positive, below-detection, zero, and missing Ag, Cd, Ga, Ge, and In analyses.
Project structure
Project components
Source validator
Checks exact hashes, coordinate systems, geometry, required fields, and published source counts before analysis.
Model concordance
Aligns H3 cells and measures presence, percentile agreement, exact overlap, and geological category summaries.
Deposit capture
Deduplicates represented target deposits and calculates distance-based capture with a seeded bootstrap summary.
Cluster builder
Groups neighbouring high-support cells into deterministic connected screening clusters with CSV and GIS outputs.
Co-product audit
Separates available positive assays from below-detection encodings and missing values for five critical co-products.
Methodology
Project workflow
- 01Verify the sources
The workflow checks three retained GeoJSON extracts against declared SHA-256 hashes and required schemas.
- 02Align the models
CD and MVT H3 cells are joined without discarding model-only areas, and percentile evidence is kept separately.
- 03Measure agreement
Exact overlap, Jaccard concordance, model presence, threshold sensitivity, and geological summaries are calculated.
- 04Build screening clusters
High-support neighbouring cells form deterministic clusters with reproducible IDs, areas, centroids, and scores.
- 05Appraise limitations
Deposit capture, bootstrap results, data completeness, model lineage, and claim boundaries are reported together.
Demonstration scenario
The retained Northern Territory data contain 44,792 CD cells and 22,106 MVT cells. Their 10,550 shared cells give a Jaccard index of 0.187229 across a 56,348-cell union. The workflow identifies 1,011 connected screening clusters. All three represented target deposits lie within 10 km of both surfaces, but the report treats this as weak descriptive evidence because the sample is very small and may share model lineage.
Engineering
Tools and method
- Tools
- The project uses Python 3.14, GeoPandas, H3, Pandas, NumPy, SciPy, Shapely, Matplotlib for subject analysis, simulation, and results.
- Geospatial layer
- GeoPandas, Pyogrio, PyProj, and Shapely validate, project, combine, dissolve, measure, and export the source geometry.
- Index layer
- H3 identifiers provide exact cell matching and deterministic neighbourhood connectivity.
- Analysis layer
- Pandas, NumPy, and SciPy calculate percentile ranks, overlap, capture, bootstrap intervals, sensitivity, and assay summaries.
- Evidence layer
- CSV, GeoJSON, JSON, and PNG retain every numerical result, screening cluster, map, and chart used in the report.
- Verification
- Automated tests, branch coverage, dependency audit, deterministic reruns, Docker, document checks, and repository validation verify the delivery.
Testing
Evaluation
Evaluation measures
- Exact CD, MVT, shared, and union cell counts
- Jaccard agreement and percentile concordance between the published surfaces
- Capture of represented target deposits at 10, 25, and 50 km
- Cluster area, cell count, mean percentile, maximum percentile, and dual-model support
- Threshold sensitivity across the 75th, 85th, 90th, and 95th retained-score percentiles
- Positive, below-detection, zero, and missing Ag, Cd, Ga, Ge, and In records
Project boundaries
- The published services retain only their high-scoring cells, and model scores are not calibrated deposit probabilities.
- The three represented target deposits are too few for strong independent predictive validation and may have influenced the original model lineage.
- Co-product assay availability does not establish mineralogical association, recoverability, continuity, grade, tonnage, or economic value.
- A screening cluster is not evidence of a deposit, resource, reserve, ownership, access permission, or exploration feasibility.
Included
- 01Complete Python and geospatial source code
- 02Three checksummed public Geoscience Australia source extracts
- 03Complete concordance evidence for 56,348 union cells
- 04One thousand and eleven connected screening-cluster records in CSV and GeoJSON
- 05Known-deposit capture, bootstrap, and threshold-sensitivity tables
- 06Silver, cadmium, gallium, germanium, and indium evidence audit
- 07Eighteen generated figures and three attributed literature images
- 08Forty-five annotated references, including current 2026 research
- 09Seventeen automated tests with 98.72 percent branch-aware coverage
- 10Complete project files, calculations, results, and analysis material in a private GitHub repository
- 1183-page project documentation in PDF and editable Word formats
- 123-page setup and usage guide in PDF and editable Word formats
Project record
No information is collected on this page.
- Permanent project ID
- GP-GE-1SRIZSI
- Catalogued
- 21 Aug 2026
- Completed
- 28 Aug 2026
- Verified
- 28 Aug 2026
- Demonstration
- Included in repository
Handover
After purchase
- 01Payment is confirmed
The project is marked unavailable and cannot be purchased again.
- 02Repository access is granted
The buyer's submitted GitHub account receives access to the private repository.
- 03The purchase record is delivered
The certification sheet is prepared from the reviewed buyer details and sent privately by email.