BRENNEN

Industries / Mining and materials / Mining and quarrying / Uranium mining

Finding deposits

Finding deposits is the search for uranium ore: measuring radiation from the air and on the ground, tracing radioactive boulders and sampling rock, then drilling to prove what lies underground.

6

Milestones achieved, from 1947 to 2014.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1947

    Counting radiation from the air

    Uranium and its decay products give off gamma rays, which instruments can count. Tube counters were developed in the first decade of the 1900s, and scintillation detectors, crystals that flash when a gamma ray strikes them, made instruments far more sensitive in the 1940s. The hunt for uranium led to the first airborne radiometric surveys in the United States, Canada and the Soviet Union in 1947 and in Australia in 1951, and from the 1960s spectrometers could tell potassium, uranium and thorium apart by the energy of their gamma rays, by the International Atomic Energy Agency's (IAEA) account.

    LessonOver 98% of the gamma rays measured come from the top 35 cm of ground, the IAEA notes, so a survey from the air maps only a thin surface layer and cannot see ore buried under cover.

    www-pub.iaea.org

    Credit

    government uranium survey programs in those countries (the IAEA's account names no one)

  2. 1965

    A world uranium count

    The OECD's Nuclear Energy Agency (NEA) and the IAEA first published Uranium Resources, Production and Demand, known as the Red Book, in 1965. It compiles what governments report about their uranium. The edition published on 14 September 2026, Uranium 2026, drew on 46 country reports and counts more than 8.1 million tonnes of uranium (tU) in identified resources that could be recovered at under US$260 per kilogram, and 61,924 tU mined in 2024, the most since 2016.

    LessonThe NEA itself warns that resources in the ground do not guarantee supply; mines still have to be found, financed and built in time.

    world-nuclear-news.orgoecd-nea.org

    Credit

    OECD Nuclear Energy Agency; International Atomic Energy Agency

  3. 1973

    Flying the whole United States

    The US Atomic Energy Commission began the National Uranium Resource Evaluation (NURE) in 1973 to find the country's uranium. Contract crews flew airplanes and helicopters about 400 ft above the ground on lines 3 to 6 miles apart over the lower 48 states and much of Alaska, recording gamma rays and the magnetic field, and reflew about forty promising areas on lines a quarter mile apart. The program passed to the Energy Research and Development Administration in 1974 and the Department of Energy in 1977, ended in the early 1980s, and its data went to the US Geological Survey (USGS), which publishes them.

    LessonSurvey data outlive their program only if someone keeps and converts them: about a quarter of the NURE tapes were still in their original formats when the USGS received them, and data from several early surveys are not available.

    pubs.usgs.gov

    Credit

    US Atomic Energy Commission and its successors; Bendix Corp. (main contractor); Patricia Hill, Robert Kucks and Dhananjay Ravat (2009 USGS compilation)

  4. 1975

    A copper search finds uranium

    In 1975 Western Mining Corporation found the Olympic Dam deposit in South Australia while exploring for copper in sedimentary rocks, by combining geology, geophysics and an analysis of large-scale fault lines. The deposit lies under about 300 m of younger, flat-lying rock. Lachlan Reynolds, a geologist at the mine, put its resource in 2000 at about 30 million t of copper and 930,000 t of U3O8 (uranium oxide, the form in which uranium is sold), with gold and silver.

    LessonOre under hundreds of meters of barren rock gives no radiation signal at the surface; this one was found by reasoning about deep structure and drilling geophysical targets.

    portergeo.com.au

    Credit

    Western Mining Corporation; E. S. T. O'Driscoll (fault-line analysis, cited by Reynolds)

  5. 1988

    Ore half a kilometer down

    In 1981 prospectors for the Saskatchewan Mining Development Corporation found a train of radioactive boulders, rocks dragged by glaciers away from their source, in the Athabasca Basin, a sandstone basin in northern Saskatchewan. Years of electromagnetic, gravity and radon surveys and drilling followed, much of it along the P2 fault, which runs through graphite-rich rock that conducts electricity and so shows up in electromagnetic surveys. In 1988 drill hole MAC-195 cut 4.27% U3O8 over 10 m, the discovery of the McArthur River deposit, which Cameco now mines between 530 and 640 m below the surface.

    LessonDeep discoveries take years: seven passed between the boulders and the discovery hole, and twelve more before production.

    mineraldeposits.saskatchewan.cacameco.com

    Credit

    Saskatchewan Mining Development Corporation (merged into Cameco in 1988)

  6. 2014

    Ore in the basement rock

    In 2014 NexGen Energy drilled the Arrow discovery on its Rook I property in the southwest Athabasca Basin, next to the Patterson Lake South discovery; by July all 14 holes drilled there had hit uranium along a 470 m strike length. Arrow lies in basement rock, the older crystalline rock beneath the basin's sandstone, starting about 100 m below the surface and extending downward for about 800 m. Mining Weekly's 2026 summary of the project gives measured and indicated resources, the categories drilled closely enough to estimate with confidence, of 256.7 million lb of U3O8 at 3.1%.

    newswire.caworld-nuclear.orgminingweekly.com

    Credit

    NexGen Energy (Leigh Curyer, chief executive; Garrett Ainsworth, exploration)

Watch for

  1. Uranium found 1,820 m down

    Chinese teams report uranium mineralization rich enough to count as "industrial" grade in sandstone 1,820 m deep under the desert of the Tarim Basin in Xinjiang, which they call the deepest of its kind in the world and a reliable sign of deposits worth mining nearby. They credit a new regional prediction model and deep-drilling methods for hard ground.

    Achieved whena resource estimate for the find is published and an outside body, such as the Red Book's compilers, counts it as an identified resource, or a field test shows the uranium can be recovered from that depth.

    subsites.chinadaily.com.cn

    Claimed by

    China Atomic Energy Authority, with CNNC's Beijing Research Institute of Uranium Geology and Geologic Team No. 216, July 2025

Open

  1. 01

    Score the Red Book's forecasts

    A public table of every near-term production-capability projection in every edition of the Red Book since 1965, by country and target year, matched with the production that later editions reported, plus error figures per edition and per country.

    The brief
    Credit · unclaimed
  2. 02

    Answer key for gamma logs

    A public test set of drill-hole gamma logs, the radiation counts recorded by a probe lowered down a hole, each paired with laboratory assays of the same core, from several kinds of deposit, with a scoring script that measures how well any method turns counts into uranium grade.

    The brief
    Credit · unclaimed

Open 01 · the brief

Score the Red Book's forecasts

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public table of every near-term production-capability projection in every edition of the Red Book since 1965, by country and target year, matched with the production that later editions reported, plus error figures per edition and per country.

Why it is still open

The NEA's "Red Book Retrospective" compiled the 20 editions from 1965 to 2004 into a historical profile of resources, production, demand and the time from discovery to production, not a test of its projections. Michael Dittmar of ETH Zurich compared a few editions in a 2009 preprint and quoted the Red Book's own note that world production has never exceeded 89% of reported capability. Searches for this brief found no edition-by-edition scorecard.

inis.iaea.orgarxiv.org

Built with what exists

Red Book PDFs (the IAEA's INIS repository holds digitized files for several older records under the Red Book's title, dated 1975, 1986, 1988, 1990 and 1993, which should be checked against the editions); Python with pdfplumber or Camelot for text tables; vision-language models such as Claude or Gemini to read scanned tables, with every figure checked against its page.

  1. First step

    Take the projections of 2009 to 2015 capability in the 2007 edition and set them beside the production reported for those years in later editions.

  2. Done looks like

    Capability projections two to ten years ahead from every edition, for the 15 largest producing countries, each matched with reported output, published with a median error per edition and per country and a check that the table reproduces the Red Book's own "never exceeded 89%" figure.

Open 02 · the brief

Answer key for gamma logs

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public test set of drill-hole gamma logs, the radiation counts recorded by a probe lowered down a hole, each paired with laboratory assays of the same core, from several kinds of deposit, with a scoring script that measures how well any method turns counts into uranium grade.

Why it is still open

Explorers still warn that total-count gamma readings "may not be directly or uniformly related to uranium grades". The USGS compared logs and core from 56 holes on the Colorado Plateau in 1959 and found log grades between 10% below and 25% above the chemical assays. An open-source Python revival of a 1962 Atomic Energy Commission method for sharpening gamma logs appeared on GitHub in August 2026 with no test data to score it on. Searches for this brief found no shared, multi-deposit answer key.

newswire.capubs.usgs.govgithub.com

Built with what exists

Paired log and assay tables in public USGS reports such as Bunker (1959); logs and assays filed with provincial and state mining departments; Python with lasio for standard LAS log files; the GAMLOG code; scikit-learn or PyTorch for learned methods.

  1. First step

    Put the paired log and core values from Bunker's 1959 report into a spreadsheet and count how many of the 56 holes have enough detail to score a method.

  2. Done looks like

    At least 300 holes from at least four deposit types (sandstone, unconformity, calcrete, intrusive), each log paired with assays, released under an open license with published scores, such as error in grade times thickness, for at least three methods including the 1962 method and one learned model.