Carbonatites are among the strangest rocks on Earth. They are igneous rocks composed of more than fifty percent carbonate minerals, yet they crystallise from molten carbonate rather than forming in oceans or lakes. They are volumetrically tiny, arising from extremely low degrees of partial melting of carbonated peridotite in the mantle, with melt fractions below roughly 0.05 weight percent. Despite their scarcity, carbonatites punch far above their weight economically: they host the highest rare earth element concentrations of any igneous rock type and have supplied the world with niobium, phosphate and light rare earths for nearly half a century. China’s Bayan Obo, the largest carbonatite-hosted rare earth deposit on the planet, illustrates the scale of what these rocks can deliver. With rare earth elements now classified as critical metals by most industrialised nations because of their indispensable role in green energy technologies, electric vehicle motors and wind turbines, every new carbonatite occurrence attracts immediate attention.

The Khaderpet carbonatite was discovered in 2002 by Rio Tinto Exploration during diamond-focused work on their 3-016 prospect in the Anumpalli Kimberlite Cluster. The plug intrudes an ultramafic lamprophyre known as aillikite, which itself was emplaced into Archaean granites, producing a distinctive granitoid breccia that geologists have likened to the carapace of a tortoise. Field relationships tell a clear story: the carbonatite contains angular clasts of the host granite but lacks any fragments of the lamprophyre, indicating that the carbonate magma arrived late, punching through the older breccia at the periphery of the complex. The rock weathers in the classic ‘panther skin’ pattern typical of carbonatites, and the surrounding granitoids display intense fenitisation, the brick-red alkali metasomatism produced by fluids exsolved from the carbonatitic magma. Caustic fusion of drill core from the prospect has even yielded good quality octahedral diamonds of transparent and grey varieties, adding a second commodity to an already unusual package.

Under the microscope, the Khaderpet carbonatite reveals an equigranular fabric dominated by calcite, which makes up ninety to ninety-five percent of the rock. The calcite grains display rhombohedral cleavage, twinkling relief and the high-order interference colours characteristic of carbonate under cross-polarised light. Scattered through this groundmass are accessory phases of considerable economic interest: euhedral fluorite crystals, apatite, phlogopite, garnet, magnetite and chromite, along with the rare earth bearing minerals monazite, titanite and allanite. These three minerals are the storehouses of the light rare earths, and their presence in thin section directly explains the exceptional whole-rock compositions. Flow banding in the finer grained portions of the rock, with aligned calcite crystals recording the movement of the magma, further attests to the intrusive, magmatic character of the body. Alteration of calcite to dolomite is common, particularly toward the rock margins.

The geochemistry is where Khaderpet truly stands out. Ten fresh outcrop samples analysed by X-ray fluorescence and inductively coupled plasma mass spectrometry show calcium oxide contents of 46.8 to 53.8 weight percent with negligible magnesium and iron, classifying the rock unambiguously as a calcio-carbonatite. Total rare earth element concentrations range from about 4100 to 4500 parts per million, of which the light rare earths contribute nearly everything, between 4092 and 4509 parts per million, while the heavy rare earths amount to a mere 19 to 24 parts per million. The ratio of light to heavy rare earths reaches values of 180 to 230, and the chondrite-normalised patterns show the steep, step-like profile typical of carbonatites worldwide, with a lanthanum-to-lutetium ratio of about 29 and an extraordinarily high lanthanum-to-ytterbium ratio near 282. Strontium concentrations of nearly 11,000 parts per million, barium around 900 parts per million, and elevated niobium, thorium and uranium complete the classic carbonatite trace element signature.

This extreme fractionation of light over heavy rare earths carries a profound petrogenetic message. It indicates that the Khaderpet magma formed by very low degrees of partial melting of carbonated mantle peridotite, followed by liquid immiscibility that separated the carbonate melt from its silicate sibling. Ratios of zirconium to hafnium, ranging from 85 to 115, and niobium to tantalum, from 110 to 270, show positive correlations consistent with a mantle origin. The primitive mantle-normalised trace element patterns display peaks for uranium, lanthanum, cerium and neodymium with troughs for rubidium and niobium, matching the fingerprints of metasomatised upper mantle sources. In short, the Khaderpet carbonatite is a genuine window into processes operating deep within the lithospheric mantle beneath the Eastern Dharwar Craton, a region whose thick lithospheric root has already proven fertile ground for more than 150 kimberlite and related intrusions.

Stable isotope analysis provides the final confirmation of primary character. Carbon and oxygen isotope measurements on five samples, performed at the Wadia Institute of Himalayan Geology, yield oxygen isotope values between 6.6 and 8.76 per mil and carbon isotope values between minus 7.45 and minus 6.33 per mil. These compositions fall squarely within the ‘primary igneous carbonatite’ box defined by decades of global data, overlapping the fields established for unaltered carbonatites from Precambrian terranes. Primary isotopic compositions essentially mean the rocks escaped crustal contamination and represent unmodified partial melts of the mantle. Combined with the petrography and trace element systematics, the isotope data leave little doubt that the Khaderpet carbonatite is a magmatic rock of deep origin, not a hydrothermal or secondary carbonate.

Perhaps the most consequential finding is economic. On geochemical discrimination diagrams plotting strontium-to-barium ratios and barium against total rare earth content, the Khaderpet samples plot in the ‘mineralised’ field, the same territory occupied by the Weishan rare earth deposit in China. The total rare earth content of Khaderpet exceeds that of several better-known south Indian carbonatites and approaches the values recorded at Hogenakkal, although it remains below the concentrations at Samalpatti, Pakkanadu and the Kamthai deposit, which to date is the only carbonatite-hosted rare earth occurrence in India proven feasible for mining. A drone magnetic survey has delineated an areal extent of roughly 100 by 200 metres with an average depth of 40 metres for the intrusion, but previous exploration drilled only a single shallow hole targeting diamonds. The authors argue that detailed drilling, ore characterisation and beneficiation studies are now essential to establish the order of magnitude of the rare earth oxide resource and to determine whether profitable extraction is possible.

The broader context amplifies the stakes. India currently hosts more than fifty confirmed and suspected carbonatite occurrences, and the Geological Survey of India has expanded its critical mineral exploration projects from 65 in 2020-2021 to 195 in 2024-2025, with several alkaline and carbonatite complexes among the targets. Recent discoveries across the Dharwar Craton, at Gollapalli in Telangana, Gundlupet in Karnataka and Krishtipadu in Andhra Pradesh, suggest the craton is far more carbonatite-fertile than previously recognised. The Khaderpet occurrence carries a double dividend: if the rare earth potential is confirmed, the site benefits from well-developed roads, water infrastructure and abundant local labour, and development would create employment while adding to national rare earth production. The study also issues a practical warning for explorers: carbonatites can hide alongside kimberlites, so any investigation of mantle-derived exotic rocks in the Indian Shield should watch carefully for associated carbonate intrusions. A small plug in a farmer’s field may yet prove to be a strategic national asset.

Subject of Research: Geochemistry and economic potential of a light rare earth element enriched carbonatite intrusion associated with kimberlitic rocks in the Eastern Dharwar Craton, India

Article Title: Geochemical studies and economic perspective of light rare earth element (LREE) enriched carbonatite from Khaderpet, Eastern Dharwar Craton

Article References: Phani, P. R. C., & Sengupta, P. (2026). Geochemical studies and economic perspective of light rare earth element (LREE) enriched carbonatite from Khaderpet, Eastern Dharwar Craton. Discover Geoscience, 4(1), Article 308. https://doi.org/10.1007/s44288-026-00663-7

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00663-7

Keywords: carbonatite, rare earth elements, LREE, kimberlite, Dharwar Craton, India, geochemistry, stable isotopes, monazite, critical minerals, diamonds, mineral exploration

The research, led by Sharon Christa and Tushar Mane of MIT Art Design and Technology University in Pune together with Ketut Tomy Suhari of Universiti Geomatika Malaysia, focuses on the Dharwar Craton, an ancient block of continental crust spanning roughly 42,291 square kilometers in Karnataka and Andhra Pradesh. The Dharwar Craton is one of India’s most storied mineral provinces, hosting gold deposits and metallogenic belts that have been worked and debated for well over a century. Rather than relying on scarce deposit inventories to teach the model what mineralized ground looks like, the team built a framework that first teaches itself the fundamental patterns of the landscape, then applies that knowledge to the prospectivity task with only limited labels.

The technical heart of the approach is masked image modelling, a self-supervised pre-training strategy popularized in computer vision. The idea is elegantly simple: take an image, hide most of it, and ask the network to reconstruct the missing pieces. To succeed, the model must internalize the spatial grammar of the data, the way rock units adjoin one another, how magnetic anomalies trace buried structures, and where geological boundaries align with geophysical gradients. The authors pre-trained a ViT-Base encoder separately for each fold of a spatial cross-validation scheme, meaning the network developed its own understanding of the terrain before ever seeing a single prospectivity label. Crucially, the pre-training data for each fold excluded the spatial region reserved for testing, a deliberate design choice that eliminates what the authors call transductive data leakage, a subtle but serious flaw in many geospatial machine learning studies.

That leakage concern deserves emphasis, because it is where many published geospatial AI results quietly inflate their performance. When training and test data come from geographically adjacent or overlapping areas, the model can effectively memorize the answer key rather than learn transferable geological relationships. The team addressed this with fivefold spatial cross-validation and a 20-kilometer exclusion buffer around each test region, a rigorous protocol borrowed from ecological modelling literature that ensures the model is genuinely predicting unseen territory. The result is a performance estimate that reflects real-world exploration conditions, where a geologist asks the algorithm about ground it has never been shown.

Once pre-training was complete, the framework faced the second great challenge of geoscience machine learning: fusing data that come in different forms and at different scales. Geological maps exist at 1:25,000 and 1:50,000 scales, capturing lithology, structure, and alteration at complementary resolutions, while aeromagnetic survey data record the magnetic fingerprint of subsurface rocks from the air. The architecture the researchers devised is dual-branch, with one pathway processing the multi-scale geological information and another handling the aeromagnetic features. These branches communicate through cross-modal attention, a mechanism that lets each data type learn which features of the other are relevant, and through deep canonical correlation analysis, a technique that finds shared structure between the two representations. In effect, the model learns how surface geology and magnetic signatures conspire to reveal hidden mineral systems.

The performance gains are striking. Across the five spatial cross-validation folds, the full framework achieved a mean area under the precision-recall curve, or AUC-PR, of 0.7172, with a standard deviation of 0.2020. That figure towers over the baselines: a Random Forest classifier managed only 0.3672, the classical Weights of Evidence method reached 0.3619, and a Vision Transformer without self-supervised pre-training scored 0.3994. The comparison isolates the value of pre-training itself, confirming a mean improvement of roughly 0.32 AUC-PR attributable to masked image modelling. Because mineral exploration is a classic imbalanced-data problem, where prospective ground is a tiny fraction of the total landscape, the precision-recall metric is the appropriate yardstick, and the margin over the baselines is not incremental but transformative.

The authors were equally careful about uncertainty and honesty in their predictions. The framework employs a Positive-Unlabelled learning setup, which acknowledges that the map contains confirmed mineralized zones, the positives, and vast areas whose mineral status is simply unknown, rather than known to be barren. This is a more truthful description of exploration reality than binary labelled classification. On top of that, Monte Carlo dropout provides uncertainty-aware outputs, running the network repeatedly with different dropout configurations to quantify how confident each prediction should be. The model also produces three prospectivity prediction heads at different depths, but the researchers are explicit that only the 0 to 500 meter surface head is the primary validated output; the deeper heads are exploratory and should not be treated as validated predictions. That kind of methodological candor is rare and welcome.

Perhaps the most reassuring result is geological rather than statistical: the high-prospectivity zones identified by the model are spatially consistent with known metallogenic belts in the Dharwar Craton. The algorithm, trained largely on unlabeled data, independently rediscovered the regions that generations of field geologists have flagged as fertile ground. That convergence suggests the network is learning genuine geological structure rather than exploiting artifacts of the data pipeline. It also hints at the framework’s real promise, which is not confirming what is already known but highlighting analogous, previously overlooked terrain where the same structural and geophysical signatures appear without any recorded deposit.

The practical implications extend well beyond one craton in southern India. Global demand for critical minerals is accelerating, and greenfield exploration increasingly depends on extracting maximum insight from existing geophysical and geological datasets before committing to expensive drilling campaigns. Airborne magnetic surveys cover enormous territories, and geological maps exist at multiple scales for most of the planet’s exposed land, yet the labelled deposit inventories needed for supervised learning remain sparse. A framework that learns rich representations from the abundant unlabeled data and fuses heterogeneous sources through attention mechanisms offers a template for prospectivity mapping in exactly those data-poor settings. The authors have also made their work reproducible, releasing the code publicly on GitHub, and the underlying aerogeophysical magnetic dataset from the Geological Survey of India is available through the AI Kosh platform, lowering the barrier for other teams to build on the approach.

There are, of course, caveats. The high variance across cross-validation folds, reflected in that 0.2020 standard deviation, reminds us that prospectivity prediction in some regions remains much harder than in others, and the exploratory status of the deeper prediction heads means three-dimensional targeting is not yet validated. The framework is also surface-focused, so deposits concealed deep beneath cover sequences may evade detection. But the core demonstration stands: self-supervised representation learning, applied with rigorous spatial validation and honest uncertainty quantification, can nearly double the detection performance of established methods under realistic data scarcity. As exploration geologists confront ever subtler targets and shrinking discovery rates, the marriage of Vision Transformers, masked image modelling, and multi-modal geophysical fusion may prove to be one of the most consequential tools to enter the mineral exploration toolkit in decades. The Earth’s next great ore deposits, the study suggests, may be found first by machines that taught themselves to read the rocks.

Subject of Research: Self-supervised multi-modal deep learning for mineral prospectivity mapping in the Dharwar Craton, India

Article Title: Self-Supervised Multi-modal Fusion for Mineral Prospectivity Mapping Using Vision Transformers Integrating Multi-scale Geological Maps with Aeromagnetic Data

Article References: Christa, S., Suhari, K. T., & Mane, T. (2026). Self-Supervised Multi-modal Fusion for Mineral Prospectivity Mapping Using Vision Transformers Integrating Multi-scale Geological Maps with Aeromagnetic Data. Natural Resources Research. https://doi.org/10.1007/s11053-026-10771-3

Image Credits: AI Generated

DOI: 10.1007/s11053-026-10771-3

Keywords: mineral prospectivity mapping, Vision Transformer, self-supervised learning, masked image modelling, multi-modal fusion, aeromagnetic data, geological maps, Dharwar Craton, Positive-Unlabelled learning, spatial cross-validation, uncertainty quantification, deep learning