Google WeatherNext 2 Is Available Through the Open-Meteo API
Global DeepMind AI-based ensemble forecast, 0.25° resolution, 6-hourly data
Open-Meteo now provides Google DeepMind WeatherNext 2 through the Ensemble API endpoint. You can access it here.
WeatherNext 2 is produced as a 64-member global ensemble on a regular 0.25 degree latitude-longitude grid. The forecast data is issued at 6-hour intervals with up to 360 hours, or 15 days.
Specifications:
Coverage: global with 0.25 degrees
Ensemble size: 64 members
Native temporal resolution: 6-hourly
Forecast range: 60 forecast steps, from +6 hours to +360 hours
Update cadence in Open-Meteo: every 12 hours, currently processing the 00 UTC and 12 UTC runs
Native forecast fields:
2m Temperature
Sea-Surface Temperature
Mean Sea-Level pPressure
Total Precipitation
10 and 100 m Wind Forecasts
Temperature, Specific Humidity, Geopotential, Vertical Velocity on 13 pressure levels at 1000, 925, 850, 700, 600, 500, 400, 300, 250, 200, 150, 100, and 50 hPa
Computed fields by Open-Meteo:
Relative Humidity, Dew Point and cloud cover on pressure level
Low, Mid, High and Total Cloud Cover
Snowfall and Weather Codes
Open-Meteo, like with other ensemble models, retains the only most recent forecasts for all ensemble members. Ensemble mean and spread (standard deviation) are kept over a longer period.
More information about Google NextWeather 2 can be found in the Google Blog Announcement and its technical documentation.
Great!