New AI weather model could predict hurricanes faster, more accurately
Google has developed a new AI weather model that can help forecast tropical cyclones and is partnering with the National Hurricane Center to make their models available to forecasters for testing.
Google's Weather Lab website allows users to interact with its AI weather models and compare live and historical tropical cyclone tracks to traditional weather forecasting models.
"When read together, these predictions can help weather agencies and emergency service experts better anticipate a cyclone's path and intensity," Google wrote in a recent press release. "This could help experts and decision-makers better prepare for different scenarios, share news of risks involved and support decisions to manage a cyclone's impact," they added.
The new AI model is trained using key historical data from nearly 5,000 observed cyclones over the past 45 years, along with a comprehensive global weather re-analysis dataset comprising millions of weather observations.

Unlike traditional weather models, which rely on atmospheric and physics-based simulations that require significant time and computing power to complete, AI weather models can produce forecasts within seconds. The new AI model is also capable of generating 50 possible scenarios for a tropical cyclone's formation, track, intensity, size and shape up to 15 days in advance.
Google found that their new AI tropical cyclone model performed at a comparable level in most cases to traditional models and outperformed them in a few tests.
In addition to their partnership with the National Hurricane Center, Google's Weather Lab team is working with the Cooperative Institute for Research in the Atmosphere at Colorado State to evaluate the new model's performance. According to Google, CIRA scientists found the new AI model had "comparable or greater skill than the best operational models for tracking and intensity."
-ABC News meteorologist Kyle David