Cao et al. (2026) A Driver‐Aware Machine Learning Approach for Predicting the Severity of Decadal‐Scale Extreme Climate Events

⚠️ 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: 2026-08-06
• Authors: Yingjian Cao, Shuyue Wu, Jianshi Zhao
• DOI: 10.1002/joc.70533

Research Groups

• Institute for Climate and Application Research, Chinese Academy of Sciences
• Department of Hydrology and Water Resources, University of Science and Technology of China

Short Summary

The study proposes an integrated framework combining driver analysis and machine learning methods to predict extreme climate event severities at the basin scale. The results show that long-term signals of global warming, AMO, PDO, SOI, and other climate drivers significantly influence extreme wet and dry events in the Yellow River Basin.

Objective

• Investigate the impact of large-scale climate drivers on decadal-scale severities of extreme wet and dry events in the Yellow River Basin

Study Configuration

• Spatial Scale: The study focuses on the Yellow River Basin (YRB) at a basin scale.
• Temporal Scale: The analysis spans multiple decades, with predictions made for 2030.

Methodology and Data

• Models used:
  • Long short-term memory (LSTM) model
  • Support vector regression (SVR) model
• Data sources:
  • Climate reanalysis data (e.g., ERA5)
  • Satellite observations (e.g., TRMM, GPM)

Main Results

• Extreme dry events in the YRB are mainly driven by long-term signals of global warming, AMO, and PDO.
• Extreme wet events are dominated by long-term signals of SOI, global warming, and PDO.
• Predictions for 2030 indicate potential increases (up to 30%) or decreases (up to 20%) in extreme dry event severity across the YRB.

Contributions

The proposed framework provides a novel approach to incorporating large-scale climate drivers into data-driven prediction models, enhancing their reliability and accuracy in predicting extreme climate events.

Funding

• National Key Research and Development Program of China (2020YFA0607803)
• National Natural Science Foundation of China (41930532)

Citation

@article{Cao2026DriverAware,
  author = {Cao, Yingjian and Wu, Shuyue and Zhao, Jianshi},
  title = {A Driver‐Aware Machine Learning Approach for Predicting the Severity of Decadal‐Scale Extreme Climate Events},
  journal = {International Journal of Climatology},
  year = {2026},
  doi = {10.1002/joc.70533},
  url = {https://doi.org/10.1002/joc.70533}
}