Zhou et al. (2026) Simulation of Extreme Weather Events in the Wanzhou Region of the Three Gorges Reservoir Area Using the WRF Model Coupled With Machine Learning Techniques

⚠️ 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: 2026-03-25
• Authors: Yulin Zhou, Lang Xu, Xing Wei, Ruibo Shi, Yuanjun Chen, Libo Ran
• DOI: 10.1002/joc.70354

Research Groups

Not explicitly mentioned in the abstract.

Short Summary

This study systematically evaluates the performance of the Weather Research and Forecasting (WRF) model with various physical parameterisation schemes for extreme precipitation and high-temperature events in the Wanzhou District of the Three Gorges Reservoir region. It identifies optimal WRF configurations and demonstrates that coupling these with machine learning models, particularly Random Forest, significantly enhances prediction accuracy and reliability.

Objective

• To systematically evaluate the performance of the Weather Research and Forecasting (WRF) model under multiple physical parameterisation schemes for extreme precipitation and high-temperature events in the Wanzhou District of the Three Gorges Reservoir region.
• To identify optimal WRF physical parameterisation scheme combinations for these extreme events.
• To optimise WRF outputs through post-processing using extreme gradient boosting (XGBoost), random forest (RF), and long short-term memory (LSTM) models.

Study Configuration

• Spatial Scale: Wanzhou District of the Three Gorges Reservoir region, characterized by complex terrain and reservoir regulation.
• Temporal Scale: Focus on extreme precipitation and high-temperature events; specific overall temporal range not detailed in the abstract.

Methodology and Data

• Models used:
  • Weather Research and Forecasting (WRF) model with various physical parameterisation schemes:
    • Scheme A5: WSM6-GD-Noah-RRTM/Dudhia
    • Scheme B5: WSM6-GD-Noah-RRTMG/CAM
  • Machine learning models for post-processing:
    • Extreme Gradient Boosting (XGBoost)
    • Random Forest (RF)
    • Long Short-Term Memory (LSTM)
• Data sources: Observational data for validation and assessment of model performance (specific sources not detailed in the abstract).

Main Results

• Scheme A5 (WSM6‐GD‐Noah‐RRTM/Dudhia) provided the best overall performance for extreme precipitation simulations.
• Scheme B5 (WSM6‐GD‐Noah‐RRTMG/CAM) performed best for high-temperature events.
• Coupling the optimal WRF configurations with the Random Forest (RF) model yielded the greatest improvement in predictions.
  • For extreme precipitation, temporal consistency indices increased by 38.03% and spatial consistency indices by 57.30%.
  • For high-temperature events, temporal consistency indices increased by 25.25% and spatial consistency indices by 55.79%.
• The integration of machine learning models into WRF substantially improved the accuracy and reliability of extreme event predictions.

Contributions

• Provides a systematic evaluation of WRF model performance under multiple physical parameterisation schemes in a complex terrain and reservoir-regulated region.
• Identifies optimal WRF scheme combinations specifically tailored for extreme precipitation and high-temperature events in the study area.
• Demonstrates the significant value of integrating machine learning models (particularly Random Forest) for post-processing WRF outputs, leading to substantial improvements in prediction accuracy and reliability for extreme events.
• Offers vital scientific evidence and technical support for disaster risk reduction and meteorological forecasting in the Wanzhou District of the Three Gorges Reservoir region.

Funding

Not explicitly mentioned in the abstract.

Citation

@article{Zhou2026Simulation,
  author = {Zhou, Yulin and Xu, Lang and Wei, Xing and Shi, Ruibo and Chen, Yuanjun and Ran, Libo},
  title = {Simulation of Extreme Weather Events in the Wanzhou Region of the Three Gorges Reservoir Area Using the <scp>WRF</scp> Model Coupled With Machine Learning Techniques},
  journal = {International Journal of Climatology},
  year = {2026},
  doi = {10.1002/joc.70354},
  url = {https://doi.org/10.1002/joc.70354}
}