The volcano is now being asked to state its own magnitude
the volcano is now being asked to state its own magnitude: a machine learning framework trained on 914 eruptions from 101 volcanoes forecasts erupted mass from rock and fault data, a cnn classifier tells phreatic from magmatic blasts at 91.2 percent accuracy, and seismic patterns from 24 volcanoes are now read as early warnings.
the mountain now files its own pressure report.
Context
An EGU 2026 abstract presents a machine learning framework that forecasts a volcano's potential erupted mass from static geophysical, petrological and tectonic characteristics plus eruption history. It was trained on 914 historical eruptions from 101 volcanoes and applied to 135 volcanoes active between 1982 and 2024, assuming an eruption duration of 225 days. It says erupted mass has been quantified for only about 100 of 1,282 known volcanoes.
A Research Square preprint from Alberto Ardid and colleagues describes eruption forecasting models trained on seismic data from 24 volcanoes, using 41 eruptions. A Nature Communications paper on ergodic seismic precursors and transfer learning addresses short-term forecasting at data-scarce volcanoes.
The 914 eruptions and 101 volcanoes match the EGU abstract. The abstract says predictions are dominated by static geophysical parameters, with eruption duration, elevation, gravity and magnetic data the most influential. The note's 'rock and fault data' is looser than that list. The abstract says the approach works best for eruptions longer than three months and that short ones may need detailed historical records.
The 24 volcanoes come from a different group's seismic study, not the erupted mass model. The preprint and the Nature Communications paper are linked here by topic. The page read does not confirm that they are one study, and the note's 'read as early warnings' is the author's wording for research models.
The 91.2% CNN accuracy for telling phreatic from magmatic blasts was not found in any source read, so it is unsupported here, not refuted. Other CNN studies of volcano seismic events exist, but none read gave this figure.
Related work
- A Machine Learning Approach for Volcanic Eruption Mass Estimation (EGU General Assembly 2026, abstract EGU26-4281) ↗Primary source for the 914 eruptions, 101 volcanoes and the erupted mass framework.
- Generalized Eruption Forecasting Models using Machine Learning Trained on Seismic Data from 24 Volcanoes (Research Square preprint) ↗Source for the 24-volcano seismic forecasting study.
- Ergodic seismic precursors and transfer learning for short term eruption forecasting at data scarce volcanoes (Nature Communications) ↗Related paper on seismic precursors and transfer learning for data-scarce volcanoes.
Watch next
- Find the source of the 91.2% CNN figure. Check whether the erupted mass model has been tested on eruptions after 2024.
Sources
- EGU26-4281: machine learning approach to volcanic eruption mass estimationmeetingorganizer.copernicus.org
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 3 October 2026 at 20:17 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →