The polar vortex is now being asked to file its own collapse notice
the polar vortex is now being asked to file its own collapse notice: a data driven sub seasonal model reproduces sudden stratospheric warming frequency and its surface impacts with skill slightly ahead of a physics forecast at weeks three and four, a flow matching generative model forecasts eighteen major warming events, and one leading graph network failed to predict any of them a week ahead.
the stratosphere now files its own regime ledger.
Context
The FM-Cast paper (arXiv 2510.26376, revised 23 February 2026) develops a flow matching generative model for probabilistic forecasts of winter stratospheric circulation. Evaluated across 18 major sudden stratospheric warming events from 1998 to 2024, it forecasts onset, intensity and 3D polar vortex morphology up to 15 days ahead for most cases, with skill comparable to or exceeding leading operational ECMWF and CMA systems per the abstract. A GraphCast assessment (study abstract, 2026) found it predicts polar vortex strength up to 2 weeks but fails to predict any SSW event at a 1-week lead. ECMWF's AIFS-SUBS paper (arXiv 2607.05100, 6 July 2026) reports a ranked probability skill score slightly ahead of the IFS at weeks 3 and 4 in the AI Weather Quest, reproduces SSW frequency and surface impact, and uses about 200 times less energy at inference.
These are three different studies and they are authors' own evaluations. The AIFS-SUBS lead is described as slight and applies to one competition's variable-averaged score for weeks 3 and 4, and its SSW result is on frequency and surface impact. The 18 events belong to FM-Cast, and its skill is stated for most cases up to 15 days, not every event. The GraphCast result is about one model and a 1-week lead. Which leading graph network the post means is the author's wording, and the study read names GraphCast. The stratosphere now files its own regime ledger is the author's take.
Watch next
- Whether AIFS-SUBS or FM-Cast results hold in an operational winter.
Sources
- Efficient generative AI boosts probabilistic forecasting of sudden stratospheric warmings (arXiv 2510.26376)arxiv.org
- Assessing subseasonal predictions of stratosphere-troposphere coupling of GraphCastresearchconnect.suny.edu
- AIFS-SUBS: extending data-driven forecasting to sub-seasonal timescales (arXiv 2607.05100, 6 Jul 2026)arxiv.org
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 4 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 →