AI & Climate
Google introduces WeatherNext 3 with hourly satellite-grounded forecasts
Google says its new global AI weather model refreshes hourly, resolves selected surface variables at up to five kilometres and is rolling into consumer and developer products.
What happened
Google DeepMind and Google Research introduced WeatherNext 3, a global AI weather model that ingests hourly geostationary satellite mosaics alongside historical analysis and weather-station observations. Google says the model issues a new forecast every hour, resolves selected surface variables at five kilometres and is beginning to power Search, Gemini, Maps, the Maps Platform Weather API, Earth Engine, BigQuery and downloadable Cloud data.
Why it matters for AI
More frequent, higher-resolution forecasts could improve planning for local weather, renewable energy and operations in regions where coarse global models miss fast or geographically narrow changes. Broad product integration also means one model update can affect decisions made by consumers, developers and organisations at the same time.
AIRunsThis View
The useful signal is the combination of fresh observations, an inspectable research paper and distribution through production interfaces. The accuracy figures still require workload- and region-specific validation, and WeatherNext output must not replace official warnings from national meteorological services.
What this could mean for us
Relevant to future AIRunsThis geospatial and scenario products: weather-derived features should record model version, observation time, spatial resolution and provenance, and should route public-safety decisions to the applicable official authority.
Source and method
This item is based on a primary source. AIRunsThis separates the source’s factual announcement from its own interpretation. Vendor performance claims have not been independently reproduced unless stated.