Li et al. (2026) Improving CWRF Prediction of Summer Monthly Precipitation Over the Yangtze River Basin With Spatio‐Temporal Graph Neural Network

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: International Journal of Climatology
• Year: 2026
• Date: 2026-04-04
• Authors: Tao Li, Han Wu, Fang Wang, Shenjun Zhong, Jianjun Xue
• DOI: 10.1002/joc.70375

Research Groups

Not available in the abstract.

Short Summary

This study proposes a novel deep learning framework utilizing a spatio-temporal graph neural network (GNN) to enhance the accuracy of summer precipitation forecasts from the Climate-Weather Research and Forecasting (CWRF) model over the Yangtze River Basin (YRB). The GNN model significantly improves prediction skill across multiple metrics, outperforming existing neural network-based models.

Objective

• To improve the accuracy of summer precipitation forecasts over the Yangtze River Basin (YRB) by developing a novel deep learning framework based on a spatio-temporal graph neural network (GNN) to enhance the prediction skill of the Climate-Weather Research and Forecasting (CWRF) model.

Study Configuration

• Spatial Scale: Yangtze River Basin (YRB)
• Temporal Scale: Summer (flooding season), monthly precipitation prediction

Methodology and Data

• Models used: Climate-Weather Research and Forecasting (CWRF) model, Spatio-temporal Graph Neural Network (GNN), other neural network-based models (for comparison).
• Data sources: Raw output from the CWRF model.

Main Results

• The proposed GNN model outperforms existing neural network-based models across multiple evaluation metrics for precipitation prediction over the YRB.
• GNN achieves the most substantial reduction in root-mean-square error (RMSE) across the YRB.
• It outperforms other models in terms of temporal correlation coefficient, showing overall monthly increases exceeding 0.1 in June compared to the raw CWRF output.
• GNN demonstrates obvious enhancements in anomaly correlation coefficient, indicating better year-to-year consistency.

Contributions

• Proposes a novel deep learning framework using a spatio-temporal graph neural network to improve precipitation prediction skill.
• Significantly enhances the accuracy of the CWRF model's summer precipitation forecasts over the Yangtze River Basin.
• Provides valuable guidance for improving precipitation forecasts, particularly in regions with complex topography, irregular spatial domains, and diverse climate conditions.

Funding

Not available in the abstract.

Citation

@article{Li2026Improving,
  author = {Li, Tao and Wu, Han and Wang, Fang and Zhong, Shenjun and Xue, Jianjun},
  title = {Improving <scp>CWRF</scp> Prediction of Summer Monthly Precipitation Over the Yangtze River Basin With Spatio‐Temporal Graph Neural Network},
  journal = {International Journal of Climatology},
  year = {2026},
  doi = {10.1002/joc.70375},
  url = {https://doi.org/10.1002/joc.70375}
}