Case Study: ECMWF boosts weather forecasting accuracy and speed with Eviden AI and HPC

Winds of Change With ECMWF's Digital Transformation

The European Centre for Medium-Range Weather Forecasts (ECMWF) sought to enhance its ability to predict extreme weather by improving the accuracy of its numerical weather simulations. To tackle this challenge, ECMWF partnered with Eviden, selecting them through an international tender to provide a BullSequana supercomputer and establish a center of excellence for HPC and AI.

Eviden's AI experts developed a neural network solution to enhance ECMWF's ecRad model. This solution achieved a 97% reduction in the accuracy difference between the production model and the high-fidelity benchmark while running significantly faster. The integration of this AI dramatically accelerated ECMWF's weather forecasting capabilities and boosted climate discovery.

ECMWF

Matthew Chantry

Machine Learning Coordinator