Original title: Bayesian deep learning integration of geophysical and drilling data for 3D prediction of copper mineralization and drill targeting: a case study from the Kogodai prospect, Rudny Altai
Margarita Veshchezerova, Egor Barashov, Evgenii Gusev, Michael R. Perelshtein +3 · September 2026
Deciding where to drill for copper in complex rock zones is tough because data is sparse, mixed, and hard to interpret. To help, researchers built a three-dimensional computer workflow that tracks its own uncertainty. They tested it on existing exploration records from the Kogodai prospect in the Rudny Altai province. The team aligned underground drillholes, trenches, and rock measurements onto a shared three-dimensional grid. Next, they trained a smart learning system to map copper grades alongside electrical rock properties, calculating continuous physical patterns rather than disconnected points. By running many simulations, the computer flags its own level of doubt. The model successfully highlighted a main copper trend and several smaller targets matching high electrical responses. It also maps where predictions are solid and where new drilling would gather the most information. Users can set grade cutoffs to check target options and reduce risks before drilling.
Imagine trying to find a hidden treasure in a large field by digging random holes. You have a few old maps that show where metal detectors beeped, but the maps are blurry and the beeps could mean anything. Normally you would dig where the beeps are strongest, but you might miss the treasure because the maps are unreliable. In this paper, instead of looking at each beep separately, the model combines all the maps and hole data to create a continuous picture of where the treasure likely is, and it also shows where the picture is fuzzy so you know where to dig next.
Exploration drilling is expensive and often misses targets because data is sparse and messy. This workflow combines existing drilling and geophysical measurements into a single 3D model that shows where copper is likely and where uncertainty is high. That means drillers can focus on the most promising spots and avoid wasting money on low-confidence areas.
The abstract notes that outputs are intended for qualitative interpretation and risk-aware drill targeting, not resource estimation. Key limitations include incomplete provenance metadata for geophysical products and heterogeneity of historical sampling. The method does not replace detailed resource modeling and its reliability depends on the quality and consistency of input data.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Bayesian deep learning integration of geophysical and drilling data for 3D prediction of copper mineralization and drill targeting: a case study from the Kogodai prospect, Rudny Altai — Margarita Veshchezerova, Egor Barashov, Evgenii Gusev et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.