WeatherNext 3
A next-generation global meteorological AI model released by Google, positioned as an AI weather forecasting system with improvements in medium-to-long-range forecast accuracy and inference speed over both its predecessors and traditional numerical weather prediction baselines.
Timeline (last 90 days)
Technical details about WeatherNext 3 regarding the model's forecast accuracy, temporal resolution, and coverage have not been fully disclosed.
WeatherNext 3 provides solar radiation forecasts.
WeatherNext 3 provides wind speed forecasts at turbine height.
Google has launched the WeatherNext 3 weather forecasting system to provide meteorological prediction data for grid operators and wind and solar power generation companies.
The WeatherNext series optimizes general-purpose AI weather models for the specific altitude levels and physical quantities required by the energy industry, representing a vertically specialized industry deployment.
WeatherNext 3 has been described as the most accurate and capable global AI weather model to date.
WeatherNext 3 completely bypasses physical simulation, learning weather evolution rules end-to-end directly from real-time satellite observation data, with inference speed typically hundreds of times faster than traditional NWP.
WeatherNext 3 can provide hourly forecasts on a 5-kilometer grid, with spatial resolution five times greater than the previous generation.
WeatherNext 3 specifically added customized data interfaces for renewable energy, capable of outputting photovoltaic radiation data and wind power wind speed and direction profiles.
Google has released WeatherNext 3, which uses satellite observations to update global forecasts every hour at a spatial resolution of 5 kilometers
11 more timeline events
All Facts (18)
WeatherNext 3 introduces 100-meter wind speed forecasts, high-resolution cloud cover forecasts, and solar irradiance level forecasts, specifically serving renewable energy production.
70%UnverifiedWeatherNext 3 learns directly from real-time satellite observation data, rather than relying on output data from traditional numerical weather prediction (NWP) models.
70%UnverifiedWeatherNext 3 supports temperature and humidity forecasts at 5 km resolution, other surface variable forecasts at 10 km resolution, and atmospheric variable forecasts such as wind speed at 25 km resolution.
70%UnverifiedWeatherNext 3 provides wind speed forecasts at turbine height.
50%UnverifiedThe WeatherNext series optimizes general-purpose AI weather models for the specific altitude levels and physical quantities required by the energy industry, representing a vertically specialized industry deployment.
50%UnverifiedWeatherNext 3 provides solar radiation forecasts.
50%UnverifiedWeatherNext 3 completely bypasses physical simulation, learning weather evolution rules end-to-end directly from real-time satellite observation data, with inference speed typically hundreds of times faster than traditional NWP.
50%UnverifiedWeatherNext 3 specifically added customized data interfaces for renewable energy, capable of outputting photovoltaic radiation data and wind power wind speed and direction profiles.
50%UnverifiedWeatherNext 3 can provide hourly forecasts on a 5-kilometer grid, with spatial resolution five times greater than the previous generation.
50%UnverifiedWeatherNext 3 is a flagship AI weather forecasting model released by Google, officially positioned as the most advanced and most accurate global weather AI model to date.
50%UnverifiedWeatherNext 3 has been fully integrated into Google's product matrix, including Google Search, Gemini, Google Maps, Google Maps Platform, and Google Cloud.
50%UnverifiedWeatherNext 3 uses the NASA IMERG dataset and Google's proprietary precipitation reanalysis data for precipitation forecast training.
50%UnverifiedWeatherNext 3 provides approximately five times clearer global weather picture than the previous generation WeatherNext 2.
50%UnverifiedWeatherNext 3's precipitation forecasts show improvements of up to 60% relative to the IMERG dataset, 30% relative to MRMS, and 10% relative to rain gauge measurements at earlier lead times, as measured by the Continuous Ranked Probability Score (CRPS).
50%UnverifiedGoogle DeepMind and Google Research jointly launched WeatherNext 3.
70%UnverifiedGoogle has launched the WeatherNext 3 weather forecasting system to provide meteorological prediction data for grid operators and wind and solar power generation companies.
50%UnverifiedWeatherNext 3 has been described as the most accurate and capable global AI weather model to date.
50%UnverifiedGoogle has released WeatherNext 3, which uses satellite observations to update global forecasts every hour at a spatial resolution of 5 kilometers
50%