Su et al. (2026) An online spectral nudging-based correction system: improving physical model forecasts by incorporating large-scale circulations derived from machine learning models

Identification

• Journal: Geoscientific model development
• Year: 2026
• Date: 2026-09-17
• Authors: Yong Su, Jincheng Wang, Xueshun Shen, Couhua Liu, Xingliang Li, Jin Zhang, Hao Jing, Yingying Hu
• DOI: 10.5194/gmd-19-8673-2026

Research Groups

• CMA Earth System Modeling and Prediction Centre (CEMC), Beijing, China
• State Key Laboratory of Severe Weather Meteorological Science and Technology (LaSW), Beijing, China
• Key Laboratory of Earth System Modeling and Prediction, CMA, Beijing, China
• School of Atmospheric Sciences, Lanzhou University, Lanzhou, China

Short Summary

This study presents an online spectral nudging-based correction system that improves physical model forecasts by incorporating large-scale circulations derived from machine learning models. The hybrid system combines the strengths of the FuXi model in forecasting circulation patterns with the advantages of the CMA-GFS in representing precipitation intensity and fine-scale details.

Objective

• To develop a novel approach for improving physical model forecasts by leveraging the strengths of machine learning models.
• To investigate the effectiveness of spectral nudging-based correction system in enhancing large-scale circulation prediction.

Study Configuration

• Spatial Scale: Global scale, with a horizontal resolution of 0.125° and 87 vertical levels.
• Temporal Scale: Forecasting up to 10 days ahead at 6-hourly intervals.

Methodology and Data

• Models used: CMA-GFS (China Meteorological Administration Global Forecast System) and FuXi model (a novel cascaded machine learning system).
• Data sources: ERA5 reanalysis data, which has a spatial resolution of 0.25° and a temporal resolution of 6 hours.

Main Results

• The hybrid system demonstrates comparable performance to the FuXi model in large-scale circulation prediction.
• The forecast leading time is extended by several days compared to the CMA-GFS alone.
• Verification against high-impact weather events shows that the hybrid system integrates the strengths of both models, improving forecasting capabilities for precipitation distribution and tropical cyclone tracks.

Contributions

• This study presents an independent implementation of the spectral nudging method using a different combination of physical and machine learning models.
• The results provide additional evidence for the potential operational application of the spectral nudging-based correction system in enhancing large-scale circulation prediction.

Funding

• This research was supported by the China Meteorological Administration (CMA) and the National Natural Science Foundation of China (NSFC).

Citation

@article{Su2026online,
  author = {Su, Yong and Wang, Jincheng and Shen, Xueshun and Liu, Couhua and Li, Xingliang and Zhang, Jin and Jing, Hao and Hu, Yingying},
  title = {An online spectral nudging-based correction system: improving physical model forecasts by incorporating large-scale circulations derived from machine learning models},
  journal = {Geoscientific model development},
  year = {2026},
  doi = {10.5194/gmd-19-8673-2026},
  url = {https://doi.org/10.5194/gmd-19-8673-2026}
}