A new study published in Science demonstrates that artificial intelligence systems can generate accurate global weather forecasts faster and with less computational power than conventional physics-based models. Researchers from Google DeepMind and the European Centre for Medium-Range Weather Forecasts (ECMWF) developed GraphCast, an AI model that predicts hundreds of weather variables, including temperature, wind, and precipitation, up to 10 days in advance. In head-to-head comparisons, GraphCast outperformed the world’s leading numerical weather prediction system, the ECMWF’s High RESolution forecast (HRES), on over 90% of 1,380 test targets. The AI model was particularly skilled at predicting severe weather events, such as tropical cyclones and atmospheric rivers, with greater accuracy. While the technology shows immense potential to complement existing forecasting infrastructure, experts note that AI models rely on historical data from traditional systems for training and may struggle with unprecedented extreme weather events driven by climate change. The research highlights a significant shift toward data-driven weather prediction. Read the full article at: https://technologyreview.com/2024/05/15/ai-weather-forecasting-graphcast-outperforms-traditional-models/