Climate change-amplified extreme weather events are causing increasingly devastating and unpredictable societal, economic, and natural ecosystem impacts. Numerical weather prediction (NWP) forecast models that meteorological services use are expensive, time consuming, and limited.
NVIDIA’s FourCastNet, a physics-ML model that emulates the dynamics of global weather patterns, predicts extremes with unprecedented speed and accuracy.
FourCastNet is GPU-accelerated, powered by the Fourier Neural Operator, and trained on 10 TB of Earth system data. Each 7-day forecast requires only a fraction of a second on a single NVIDIA GPU to generate potential impacts based on thousands of scenarios, five orders-of-magnitude faster than NWP.
NVIDIA’s FourCastNet, a physics-ML model that emulates the dynamics of global weather patterns, predicts extremes with unprecedented speed and accuracy.
FourCastNet is GPU-accelerated, powered by the Fourier Neural Operator, and trained on 10 TB of Earth system data. Each 7-day forecast requires only a fraction of a second on a single NVIDIA GPU to generate potential impacts based on thousands of scenarios, five orders-of-magnitude faster than NWP.
Sign in or sign up to post comments.
Be the first to comment