AI Model Enhances Tropical Cyclone Forecasting Capabilities
A new artificial intelligence model, WeatherNext Cyclones (WN-C), has been developed to improve the forecasting of tropical cyclones. This model, created by a team including Google DeepMind and NOAA, offers advanced ensemble forecasts for cyclone track, intensity, and size. Trained on global analysis data and historical cyclone records, WN-C provides forecasts extending 15 days into the future. The model has demonstrated a lead time advantage of a day or more over existing models, marking a significant improvement in forecasting accuracy. This advancement is achieved using coarser data inputs, challenging the notion that high-resolution data is necessary for accurate intensity forecasting.