Google Releases Open-Source AI Models for Accurate Cyclone Prediction

Google has announced that it is making its cyclone prediction AI models open source. The company's WeatherNext 2 and WeatherNext Cyclones models have achieved state-of-the-art accuracy in predicting a cyclone's track, intensity, and wind structure.

The models are able to accurately predict what will happen with cyclones up to 15 days into the future, which is a degree of accuracy that has never been attained before. Google says its WeatherNext model bridges the gap between global and local modeling techniques by predicting both the track and intensity of a cyclone with high accuracy.

The company claims that its WeatherNext Cyclones model gains more than a full day (24 hours) of lead time advantage for predicting cyclone tracks, intensity, and wind structure. This is based on evaluations using historical cyclones from 2023 to 2024, benchmarking the deterministic and probabilistic performance against other top weather models.

Google says that by making these models available on an open source basis, they can be used in academic research, operational forecasting, or developing more specialized, localized models. The company's goal is to improve forecasting for global weather overall as well as cyclones.