Geoscience data and AI prospectivity mapping for mineral exploration.

From country-scale geology and open ground to ranked targets, using published geochemistry, geophysics, satellite imagery and machine learning.

  • Open ground: tenure registers of 55 jurisdictions, read to show the ground no one holds.
  • Prospectivity models: 51 minerals ranked, with the reasoning behind every score.
  • Ranked targets in Australia, the United States and Canada; a global model everywhere else.
  • Spectral mapping: alteration minerals from satellite, regional to camp scale, including WorldView-3.
  • Due diligence: tenure, land access and the real cost to hold and drill.
  • The record: over 200 million published sample records from 162 sources.

Eastern Goldfields, Western Australia. Open ground, with open pegging blocks ranked by recorded endowment. Rendered from our own data.

  • U.S. Geological Survey
  • British Geological Survey
  • CSIRO
  • European Space Agency
  • AuScope National Virtual Core Library
  • Geological Survey of Finland (GTK)
  • German Aerospace Center (DLR)
  • Geological Survey of Japan (AIST)
  • GNS Science
  • Council for Geoscience, South Africa
  • Alberta Energy and Minerals
  • Alberta Geological Survey
  • Botswana Geoscience Institute
  • British Columbia Geological Survey
  • Bureau of Land Management
  • EuroGeoSurveys
  • Geological Survey of Brazil (SGB)
  • Geological Survey of Denmark and Greenland (GEUS)
  • Geological Survey of Norway (NGU)
  • Geological Survey of Sweden (SGU)
  • GeoSphere Austria
  • Guyana Geology and Mines Commission
  • Manitoba Geological Survey
  • Mineral Resources Authority, Papua New Guinea
  • Minerals Commission of Ghana
  • Nova Scotia Department of Natural Resources and Renewables
  • ONHYM, Morocco
  • Polish Geological Institute (PGI-NRI)
  • Saskatchewan Geological Survey
  • Saudi Geological Survey
  • SERNAGEOMIN, Chile

Data published by 80 organisations ·Names and logos belong to their owners; no endorsement or partnership is implied. All 80 sources

The record

The groundwork, measured.

Counted from our holdings, not estimated. Deepest in Australia, the United States and Canada.

How we hold it
18.2M
Geoscience features
Mapped geology, structure and deposits across 40 countries
8.7M
Tenure records
Live and historic titles from the registers of 48 countries
110M
Distinct samples
Published geochemistry from 162 sources, 94.8% of records Australian
1.01B
Laboratory determinations
Element, oxide and volatile values reported on those samples
GeoSpectra

Partner

Spectral remote sensing, with GeoSpectra.

Australian specialists in spectral remote sensing and geospatial intelligence for mineral exploration. Our spectral and structural mapping is delivered with them, from free imagery to camp-scale WorldView-3.

geospectra.com.au ↗

Machine learning, in service now

Models that rank the ground.

Machine-learning prospectivity for 51 minerals, trained on 13.8M samples. Every target comes with the evidence behind it.

How prospectivity works
130+
National models
51 minerals across Australia, the United States and Canada
13.8M
Training samples
Soil, stream-sediment and rock-chip samples with laboratory assays
44
Minerals in the global model
Beyond Australia, the United States and Canada, on exposed ground
10 km
Test separation
National models are scored only on ground at least this far from their training data

National models

For each mineral in Australia, the United States and Canada. A run scores the ground and returns ranked targets, each with its evidence and a flag where that evidence is thin.

A global model

44 minerals everywhere else, read from satellite alteration, radar texture and terrain on exposed ground. It returns a prospectivity map and a confidence read for every cell.

Next, a foundation model

One model over the full record, in map view and in 3D, tested on ground held back from training and then against history before we use it.

Next, the models are retrained on the full record and new sources, such as hyperspectral imagery, before we use them to rank and select ground.

In development

Next, reading the drill core.

We hold the raw spectra from Australia's national drill-core library, matched to laboratory assays where they exist. Next, we will train a model to read mineralogy and grade from the spectrum.

Already in the record: which scanned holes sit near any piece of ground, what the scanner recorded down each, and, in four states and territories, whether assays exist for the same intervals.

Where this is heading
489.3M
Raw spectra
Visible to thermal infrared, measured down the core
6,546
Scanned holes
In six states and territories
2,746
Holes matched to assays
Spectra lined up with laboratory chemistry, metre by metre

Scans: AuScope National Virtual Core Library, CSIRO and the state and territory geological surveys.

Reach

Where we hold data.

Data held for 56 countries, and a full country atlas built for 28 of them.

Browse the countries
Dot map of the world: 28 countries with a built atlas highlighted, and 56 countries where published geoscience is held
Atlas built · 28 countries Data held · 56 countries in all Other land

Open Ground

Find ground you can actually peg.

In many jurisdictions a petroleum title does not stop a mineral one. We read the two separately, so open ground is not written off by mistake.

Alberta

42.3%

open for a mineral title

23.6% if every title counts as held

Northern Territory

80.2%

open for a mineral title

62.8% if every title counts as held

How Open Ground works
Open ground in the Eastern Goldfields of Western Australia, with open pegging blocks ranked by recorded mineral endowment
Eastern Goldfields, Western Australia: open ground, with open blocks ranked by recorded endowment. Data: WA tenure register · deposit record

Questions

Straight answers.

What does MineDSS do?

MineDSS ranks open ground from the published record and shows why. The engine combines country geoscience, open-ground analysis, machine-learning prospectivity for 51 minerals, spectral mapping and tenure due diligence.

What is mineral prospectivity mapping?

A map of where the ground is most likely to carry anomalous values of a target metal, built from geology, geophysics, terrain and satellite evidence. Our machine-learning models score the ground and show the evidence behind each target. In Australia, the United States and Canada a run also returns ranked targets.

Which minerals and countries do you cover?

51 minerals: national models in Australia (47), the United States (51) and Canada (37), and a global model for 44 minerals everywhere else. We hold published geoscience for 56 countries, with a full country atlas for 28.

How do you find open ground?

We read each licensing authority's own register, keep mineral and petroleum titles separate, remove excluded land wherever we hold an exclusion layer, and rank the open blocks on the recorded deposit record. 55 jurisdictions are covered.

How are your models tested?

Every model is scored on ground held back from its training; for the national models that test ground sits at least 10 km from anything they trained on. A model that fails is not used, for any mineral, in any country. Next, the models are retrained on the full record and new sources, such as hyperspectral imagery, before we use them to rank and select ground.

Talk to us

Talk to us.

  • Partners

    Exploration and mining companies interested in working with us.

  • Investors

    The record, the models and the ground they point to.

  • Publishers and researchers

    Geological surveys, universities and programmes whose work the record is built on.