Mineral prospectivity multilayer mapper
A geospatial modelling system for combining geological, geochemical, geophysical, structural, and remote-sensing evidence into mineral-prospectivity maps.
Project definition
Problem statement
Mineral exploration evidence comes from datasets with different scales, coverage, accuracy, and geological meaning. Combining layers without a deposit model or spatially independent validation can produce an attractive but unreliable map.
The engineering problem is to transform each layer into documented evidence, compare expert and data-driven integration methods, and report uncertainty and validation without claiming a mineral resource.
Project objectives
- Define a selected mineral-system or deposit model and its expected geological controls.
- Prepare public geology, structure, geochemistry, geophysics, terrain, remote-sensing, and occurrence layers.
- Transform raw layers into reproducible proximity, anomaly, categorical, and continuous evidence rasters.
- Compare weighted-overlay and selected machine-learning prospectivity models.
- Validate using spatially separated known occurrences and analyse uncertainty and weight sensitivity.
System design
System modules
Layer catalogue
Records source, scale, resolution, projection, date, coverage, licence, quality, and geological interpretation.
Evidence processor
Creates aligned fault-distance, lithology, geochemical, geophysical, alteration, and terrain evidence layers.
Prospectivity modeller
Runs transparent weighted overlays and selected data-driven models under versioned configurations.
Spatial validator
Uses blocked or leave-area-out validation and compares predictions with held-out occurrence records.
Target dashboard
Shows evidence, prospectivity, uncertainty, sensitivity, ranked areas, and complete source provenance.
Methodology
System workflow
- 01Define model
The student selects a prepared study area, target mineral system, conceptual controls, and validation plan.
- 02Prepare evidence
Layers are cleaned, projected, resampled, transformed, and masked to comparable analysis cells.
- 03Build models
Expert-weight and data-driven approaches combine the same evidence under saved configurations.
- 04Validate spatially
Held-out areas or occurrence groups measure predictive ranking and reduce spatial leakage.
- 05Review targets
High-scoring areas are compared by evidence support, uncertainty, sensitivity, access, and data gaps.
Demonstration scenario
A prepared study area combines lithology, fault distance, geochemistry, magnetics, and alteration evidence. Weighted and machine-learning models are trained outside one spatial block. The dashboard compares their held-out occurrence ranking, uncertainty, sensitivity, and the evidence supporting each target area.
Engineering
Technical architecture
- Web application
- Nuxt and Vue for layer control, map comparison, target profiles, sensitivity, and export.
- Geospatial API
- FastAPI for layer jobs, evidence transformations, modelling, validation, ranking, and tiles.
- Spatial data layer
- PostgreSQL and PostGIS for geology, occurrences, structures, targets, metadata, and model results.
- Raster pipeline
- Python, Rasterio, and GeoPandas for alignment, distance, interpolation, anomaly, zonal, and masking operations.
- Modelling layer
- Transparent weighted methods and scikit-learn models with spatial splits, calibration, and feature importance.
Testing
Evaluation
Evaluation measures
- Spatial alignment and evidence-transformation correctness
- Area under ROC and precision-recall curves under spatial validation
- Success-rate and prediction-area curves for held-out occurrences
- Target-rank stability under weight, threshold, resolution, and data-removal sensitivity
- Calibration and uncertainty across evidence-coverage conditions
- Processing time across layer count, raster resolution, and study-area size
System boundaries
- The map shows modelled prospectivity and does not establish the presence, grade, quantity, ownership, or economic value of a mineral deposit.
- Only public, licensed, synthetic, or explicitly permitted geoscience data is included.
- Model results depend on the selected deposit concept, evidence coverage, resolution, and known-occurrence bias.
- Exploration decisions require field mapping, geophysics, geochemistry, drilling, permissions, environmental review, and qualified geologists.
Included
- 01Geoscience layer and study-area interface
- 02Preprocessing, evidence, prospectivity, validation, and ranking modules
- 03Layer, target, uncertainty, sensitivity, and validation maps
- 04Prepared public geoscience data, occurrences, and results
- 05Complete source code in a private GitHub repository
- 06Complete project documentation with synopsis, abstract, methodology, geological workflow diagrams, prospectivity maps, screenshots, and conclusion
- 07Setup and usage guide
Project record
No buyer information is collected on this page.
- Permanent project ID
- GP-GE-1SRIZSI
- Catalogued
- 21 Aug 2026
- Completed
- Pending
- Verified
- Pending
- Demonstration
- Added when ready
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.