Predicting how hurricanes and cyclones develop is a longstanding problem, and each hour of warning counts. Tropical cyclones are among the many most damaging climate occasions on Earth, and for meteorologists, issuing correct, well timed warnings is a continuing race in opposition to time.
As we speak, in a paper printed in Nature, Google researchers present that our WeatherNext AI mannequin achieved state-of-the-art accuracy in predicting a cyclone’s monitor, depth, and wind construction. This represents a large leap ahead — roughly a decade of meteorological progress in a single mannequin.
To assist construct local weather resilience worldwide, we are actually open-sourcing the WeatherNext 2 mannequin to the worldwide analysis neighborhood. For extra info, examine the Google DeepMind blog.
