AI will now understand the weather every hour; Google launches WeatherNext 3

Weather predictions can now be made using AI. Google has launched WeatherNext 3, which can generate new weather forecasts every hour. Learn about Google's new AI model, WeatherNext.

 

Google WeatherNext 3

It's not always easy to predict when and how quickly the weather will change. Sometimes, temperature differences can be seen just a few kilometers away, while sudden rain or strong winds can completely change the weather. Therefore, accurate and timely weather information is crucial for daily planning, as well as for agriculture, power generation, and preparedness to deal with bad weather. In this direction, Google has introduced its new AI model, WeatherNext 3, which claims to provide weather forecasts faster and with better resolution than before. So, let's find out what Google's new AI model, WeatherNext, is.

What is WeatherNext, Google's new AI model?

Google DeepMind and Google Research launched WeatherNext 3 on September 3rd. According to the company, it is its most advanced global weather AI model to date. It draws directly from live geostationary satellite data and can generate hourly weather forecasts.

The previous model, WeatherNext 2, produced forecasts every six hours on a 25-kilometer grid. WeatherNext 3 displays key surface weather data at a resolution of up to 5 kilometers. Other surface data uses a resolution of 10 kilometers, and atmospheric data uses a resolution of 25 kilometers. According to Google, this results in a global weather picture approximately five times clearer than the previous model.

AI will tell when and where it will rain.

WeatherNext 3 has been specifically optimized for rapidly changing weather conditions like rainfall. The model is trained on NASA's IMERG and satellite radar-based global precipitation data. According to Brightband, WeatherNext 3 has improved medium-range weather forecasts by up to 60 percent compared to IMERG, and by up to 30 percent compared to MRMS. Early-range forecasts also outperform rain gauge data by up to 10 percent.

Solar and wind energy will also work.

WeatherNext 3's uses go beyond just weather monitoring. The model can estimate wind energy production by forecasting wind speeds at a height of 100 meters. It also provides high-resolution cloud cover and sunshine levels, allowing solar farms to better estimate their potential power generation capacity.

WeatherNext 3 will be available on these Google platforms.

Google is rolling out its new WeatherNext 3 in Search, the Gemini app, Google Maps, the Google Maps Platform Weather API, and Google Earth Engine. Weather data is also being made available to developers, researchers, and business users through BigQuery, Earth Engine, and Google Cloud Storage. According to Google, this model is particularly important for regions like Latin America, Africa, and Asia-Pacific.