A new artificial intelligence weather forecasting model developed by researchers at the University of California, Berkeley, has demonstrated superior accuracy compared to established global systems like the European Centre for Medium-Range Weather Forecasts (ECMWF). The model, named ClimaNet, was trained on four decades of historical weather data and can generate a 10-day global forecast in under one minute on a single high-performance computer. In head-to-head tests, ClimaNet provided more accurate predictions for key atmospheric variables, including temperature and geopotential height, at 3-day and 10-day ranges. Researchers note that while the AI model excels at mid-latitude forecasts, traditional physics-based models still hold an edge in predicting extreme weather events in the tropics. The team plans to integrate physical laws more directly into the AI’s training to improve its performance in these areas. The findings suggest AI could significantly accelerate and enhance operational weather prediction. Read the full article at: https://sciencedaily.com/releases/2024/10/240410145210.htm