Google boosts AI to provide more accurate weather forecasts
That artificial intelligence is used in weather forecasting is to be expected, but Google has just announced WeatherNext 2. Self-described as its “most advanced and efficient forecasting model”, this last incarnation improves not only accuracy, but also speed.
Forecast generation can be up to eight times faster, and Google is already using the data produced by WeatherNext 2 – this is not something that is coming, it is here now. After heavy research work, Google is ready to give the tool to users.
Google has already upgraded the forecasts that appear in Search, Gemini, Pixel Weather and Google Maps Platform’s Weather API. On top of this, there are plans to bring the same technology to the forecasts that are used in Google Maps.
The company says: “This breakthrough is enabled by a new model that can provide hundreds of possible scenarios. Using this technology, we’ve supported weather agencies in making decisions based on a range of scenarios through our experimental cyclone predictions”.
It continues:
We're now taking our research out of the lab and putting it into the hands of users. WeatherNext 2's forecast data is now available in Earth Engine and BigQuery. We’re also launching an early access program on Google Cloud’s Vertex AI platform for custom model inference.
Google has produced a video that goes into some more details about the prediction model:
One of the key advancements of WeatherNext 2 compared to its predecessor is it ability to use limited data to imagine numerous scenarios. As Google explains:
WeatherNext 2 can predict hundreds of possible weather outcomes from a single starting point. Each prediction takes less than a minute on a single TPU; it would take hours on a supercomputer using physics-based models.
Our model is also highly skillful and capable of higher-resolution predictions, down to the hour. Overall, WeatherNext 2 surpasses our previous state-of-the-art WeatherNext model on 99.9% of variables (e.g. temperature, wind, humidity) and lead times (0-15 days), enabling more useful and accurate forecasts.
This improved performance is enabled by a new AI modelling approach called a Functional Generative Network (FGN), which injects ‘noise’ directly into the model architecture so the forecasts it generates remain physically realistic and interconnected.
Google says that it plans to make its tools “available to the global community”. More information can be found here.