Google updates WeatherNext 3 AI model with real-time satellite data

In short: Google says its new WeatherNext 3 AI weather model uses real-time satellite data to make sharper and more frequent forecasts.

What happened

Google announced an updated AI weather forecasting system called WeatherNext 3. The company says it is more accurate than its earlier version, especially for rain and snowfall.

Weather forecasting usually relies on large computer simulations of the atmosphere, which can take time to run. Google’s approach uses AI, meaning software that learns patterns from data (like learning what cloud shapes often lead to rain). With WeatherNext 3, Google says it now also learns from real-time satellite observations, not just older weather records.

Google says WeatherNext 3 can update forecasts every hour based on the latest satellite images. The previous version, WeatherNext 2, produced forecasts every six hours and used a 25-kilometer grid. WeatherNext 3 can show some details, like temperature and moisture, down to a 5-kilometer grid, which is like switching from a blurry map to a sharper one.

Google says the biggest improvements may be in places that have fewer ground sensors like rain gauges, which is often outside the US and Europe. The company claims precipitation forecasts can be up to 50 percent more accurate at least a day ahead.

Google says WeatherNext 3 is now used in Google Search, Google Maps, Gemini, and other Google products. The company also says it has worked with the US National Hurricane Center and agencies in Asia on AI-based forecasting.

Why it matters

Better short-term rain and snow forecasts can help people plan travel, school, and outdoor work, and can also help communities prepare for fast-changing storms. Google also says the model can help predict wind conditions for renewable energy, such as wind power, which depends heavily on weather.

Source: The Verge AI