AI and the World's Hard Problems · Earth & climate science · arXiv · 2026-10-22
Hybrid ensemble forecasting combining physics-based and machine-learning predictions through spectral nudging
The first application of spectral nudging in probabilistic ensemble forecasting combines the physics-based ECMWF IFS-ENS with forecasts from the machine-learned AIFS-ENS. Large scales of virtual temperature and vorticity are relaxed toward the machine-learned forecasts, while mesoscale structures remain governed by the physics-based model. Skill gains extend up to two days in the tropics and about half a day in the extra-tropics, and tropical cyclone track forecasts improve significantly.