The weather forecast went from physics to data
the weather forecast went from physics to data: google’s weathernext 3 now runs off real-time satellite imagery with hourly predictions five times sharper than its predecessor, and germany’s dwd put the aicon ml model into full operational use in march at 13km resolution. ecmwf’s aifs v2 ships the first data-driven wave and snow forecasts.
the atmosphere stopped needing to be simulated to be predicted.
Context
Google introduced WeatherNext 3 on Sep 3, 2026; its developer docs say it ingests live geostationary satellite observations as a direct model input, and DeepMind calls it the first global weather model that generates forecasts every hour of the day.
DWD made AICON forecasts open data on Mar 3, 2026, and the AICON paper describes a global model at 13 km resolution with a 3-hour time step. ECMWF says AIFS v2 adds its first operational data-driven wave forecasts and a new snow variable.
Satellite input, hourly forecasts, 13 km AICON and the AIFS v2 wave and snow additions match the sources. The 'five times sharper than its predecessor' claim and the AICON 'March full operational use' wording were not seen in the sources read, so unsupported here, not refuted. DWD's March 3 notice is about open data availability of AICON output. 'The atmosphere stopped needing to be simulated' is the author's.
Related work
- Google: introducing WeatherNext 3 ↗Sep 3, 2026.
- DWD: AICON open data announcement ↗Mar 3, 2026.
- ECMWF: implementation of AIFS Single v2 ↗Forecast user wiki.
Watch next
- Find the sharpness comparison against WeatherNext 2.
Sources
- Google DeepMind: WeatherNext 3deepmind.google
- DWD: AICON database descriptiondwd.de
- ECMWF: implementation of AIFS v2ecmwf.int
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 1 October 2026 at 23:01 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 →