Bridging Artificial Intelligence and Data Assimilation: The Data‐Driven Ensemble Weather Forecasting System ClimaX ‐ LETKF

Abstract

While machine learning‐based weather prediction (MLWP) has achieved significant advancements, research on assimilating real observations or ensemble forecasts within MLWP models remains limited. We introduce ClimaX‐LETKF, the data‐driven ML‐based ensemble weather forecasting system that combines the MLWP model ClimaX with a conventional data assimilation program. It operates stably over multiple years, independently of numerical weather prediction (NWP) models, by assimilating the NCEP ADP Global Upper Air and Surface Weather Observations. The system demonstrates similar analysis errors with both relaxation to prior perturbation (RTPP) and relaxation to prior spread (RTPS) when optimal relaxation parameters were applied. RTPP replaces an analysis perturbation with a weighted blend of analysis and background perturbations, whereas RTPS simply rescales the analysis perturbation. A substantial reduction in the ensemble spread was observed at the first forecast timestep when initialized from analyses produced with RTPS, indicating weaker initial ensemble spread growth than with RTPP. This work demonstrates stable long‐term data assimilation cycling with a MLWP model using real observations and represents an important step toward practical, self‐sustained MLWP forecasting systems.