WeatherNext 2: Our most advanced weather forecasting model
5 days ago
- #Google-DeepMind
- #AI
- #weather-forecasting
- Google DeepMind and Google Research introduced WeatherNext 2, their most advanced weather forecasting model.
- WeatherNext 2 can generate forecasts 8x faster with up to 1-hour resolution.
- The model provides hundreds of possible weather scenarios, aiding in decision-making for weather agencies.
- Forecast data is now available in Earth Engine and BigQuery, with an early access program on Google Cloud’s Vertex AI.
- WeatherNext technology has been integrated into Google Search, Gemini, Pixel Weather, and Google Maps Platform’s Weather API.
- The model uses a Functional Generative Network (FGN) to inject noise, ensuring physically realistic forecasts.
- WeatherNext 2 excels in predicting both 'marginals' (individual weather elements) and 'joints' (complex interconnected systems).
- The model outperforms the previous WeatherNext model on 99.9% of variables and lead times.
- Google aims to advance the technology further by integrating new data sources and expanding access.
- Resources for learning more include a research paper, developer documentation, and access to forecast data in BigQuery.