Google DeepMindの「WeatherNext 2」AIモデルがサイクロン予測で大幅な精度向上を達成Our WeatherNext 2 AI model demonstrated a massive leap forward in predicting cyclones.

Predicting cyclones

Google DeepMind says its WeatherNext 2 AI model has made a major advance in cyclone prediction. The improvement matters because tropical cyclones are among the most destructive weather events on Earth, while accurate and timely warnings can support earlier disaster response and efforts to reduce harm.

Predicting how hurricanes and cyclones develop remains a longstanding challenge. Meteorologists face a constant race against time when issuing warnings, and every additional hour of warning counts.

State-of-the-art accuracy

In a paper published in Nature, Google researchers report that the WeatherNext AI model achieved state-of-the-art accuracy in predicting three important aspects of a cyclone: its track, its intensity, and its wind structure.

Track prediction concerns where a cyclone is expected to move, while the model’s other outputs address how intense it may become and how its winds are structured. Taken together, these predictions provide information relevant to issuing accurate and timely warnings as a storm develops.

Google describes the result as a massive leap forward in AI-driven weather forecasting. According to the announcement, the improvement represents roughly a decade of meteorological progress in a single model and establishes a new standard for cyclone forecasting accuracy.

Open-sourcing WeatherNext 2

Google is open-sourcing the WeatherNext 2 model to the global research community. The stated purpose is to help build climate resilience worldwide by making the model available for research.

The reported gains in forecasting cyclone tracks, intensity, and wind structure have direct safety relevance. More accurate forecasts can contribute to earlier responses to destructive weather events and may help reduce damage by giving meteorologists better information when preparing timely warnings.