WeatherNext 2 is here, Google’s improved AI weather forecaster

Google is improving the weather forecast you see in the Pixel Weather app, Google Search, and other places. The Alphabet subsidiary has introduced WeatherNext 2, an AI-powered weather forecasting model. Google says this predictive model provides faster and more accurate predictions. How fast is it? Google says this “most advanced and efficient predictive model” generates weather forecasts eight times faster, generates four six-hour weather forecasts per day, and makes future predictions for up to 15 days.

WeatherNext 2 outperforms WeatherNext 1 at predicting 99% of variables such as wind speed and direction, temperature, precipitation, pressure, and humidity. It uses a Tensor Processing Unit (TPU), a special purpose integrated circuit (ASIC) chip designed by . Google accelerates machine learning and AI workloads, so weather forecasts can now be created in less than a minute. Compare this to the time it would take to make the same predictions on a supercomputer using a physics-based model.

In the coming weeks, WeatherNext 2 will also take over the weather forecasting duties for Google Maps. So if you notice an increase in the quality of weather forecasts you see in the Pixel Weather app, Google Maps, the Pixel Weather app, or Google Search, it might be because of Google’s improvements to its WeatherNext 2 predictive model.

How Google uses AI to predict the weather. |Image credit – Google

WeatherNext 2 better understands the different variables that make up a weather forecast. These variables, such as temperature, wind, and pressure, are related, and their relationships can be captured in equations. AI weather forecasting does not rely on data processing using supercomputers. Using AI instead, users look for patterns that are expected to repeat.

WeatherNext 2 allows you to generate a 15-day weather forecast in just 1 minute. Overall, hundreds or thousands of predictions can be created quickly.