Google’s AI model bears down on traditional weather forecasting in storm track success

Google has launched a new AI model for forecasting tropical storms and created Weather Lab, a website to share its AI weather research, while partnering with the US National Hurricane Center to test the technology. The initiative comes as the Trump administration has cut National Weather Service staff and capacity, raising questions about the future of public weather forecasting versus privatized services.

What you should know: Google’s experimental AI model can generate 50 different scenarios for a storm’s possible track, size, and intensity up to 15 days in advance.

• The model’s five-day predictions for cyclone tracks in the North Atlantic and East Pacific were 87 miles (140 km) closer, on average, to actual storm tracks than predictions from the European Center for Medium-Range Weather Forecasts in 2023 and 2024.
• Google released an unreviewed research paper claiming its model’s predictions are at least as accurate as traditional physics-based models.
• The National Hurricane Center is evaluating the model’s effectiveness as the Atlantic hurricane season continues through November.

How it works: The cyclone model is trained on Europe‘s ERA5 archive, which includes hundreds of millions of weather observations from agencies worldwide combined with traditional model predictions.

• Google also used ERA5 data to train its previous GenCast model, which outperformed one of the European Center for Medium-Range Weather Forecasts’ leading physics-based models 97.2 percent of the time, according to research published in Nature in December 2024.
• Weather Lab’s interactive website allows users to compare AI models against traditional physics-based forecasts, though Google emphasizes it’s only a research tool.

The big picture: AI weather models are positioned to assist rather than replace traditional forecasting, as they still rely heavily on real-world observations and historical weather data.

• Google is collaborating with the Cooperative Institute for Research in the Atmosphere at Colorado State University and researchers in the UK and Japan to improve its AI weather models.
• The importance of collecting and analyzing new data on increasingly extreme weather events remains crucial for adapting to climate change.

Political context: The Trump administration and the Department of Government Efficiency have significantly impacted federal weather research capabilities.

• The National Weather Service reduced weather balloon launches after staffing cuts, with the National Oceanic and Atmospheric Administration increasingly relying on private company data.
• Project 2025 called for dismantling NOAA and privatizing much of its weather services, raising concerns about turning forecasts into paid products.

What they’re saying: Google DeepMind research scientist Peter Battaglia addressed privatization concerns during a press call.

• “For a long time, weather has been viewed as a public good, and I think, you know, most of us agree with that … Hopefully we can contribute to that, and that’s why we’re trying to kind of partner with the public sector,” Battaglia said.

Notable omission: Unlike previous announcements, Google’s launch today doesn’t mention the climate crisis, contrasting with its December 4 GenCast announcement that stated: “As climate change drives more extreme weather events, accurate and trustworthy forecasts are more essential than ever.”