Google’s best weather model now skips physics simulations entirely and learns from live satellite…
google’s best weather model now skips physics simulations entirely and learns from live satellite data, and weathernext 3 forecasts are 50% more accurate on precipitation. it runs hourly on a 5km grid and already powers search, maps and gemini.
the physics that took meteorology a century to build is now optional.
Context
Verified date: Google's blog post Introducing WeatherNext 3 is dated Sep 3, 2026. It says the model now uses real-time satellite data, refreshes hourly, runs at higher resolution, gives precise precipitation forecasts and is integrated across Search, Gemini, Maps, Google Maps Platform and Cloud. The post says it shows temperature and moisture at 5-kilometer resolution, other surface variables at 10 kilometers and atmospheric variables such as wind at 25 kilometers.
Google's developer research page says precipitation shows up to a 50% reduction in Brier score and CRPS compared to numerical weather prediction baselines, evaluated against IMERG observations.
The Sep 3 launch, the hourly refresh, live satellite input and the Search, Maps and Gemini integration match Google's pages. The 50% is a reduction in Brier score and CRPS, stated as up to, against numerical weather prediction baselines, which is not the same as 50% more accurate. The 5 km grid applies to some surface variables, with others at 10 and 25 km. 'Physics is now optional' is the author's opinion, and the paper describes the model taking two analysis frames from physics-based analysis plus satellite frames as inputs.
Related work
- Google DeepMind: WeatherNext ↗First global model to forecast every hour.
- Google for Developers: WeatherNext 3 models ↗Hourly initialization from satellites.
- Google blog post ↗Sep 3, 2026.
Watch next
- Read the WeatherNext 3 paper for what the model takes as input besides satellites.
Sources
- Google: introducing WeatherNext 3blog.google
- Google for Developers: WeatherNext research and benchmarksdevelopers.google.com
- DeepMind: WeatherNext 3 paperstorage.googleapis.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 1 October 2026 at 12:14 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 →