Wang et al. (2026) Planetary Boundary Layer Height Prediction over Karst Plateau: A Study Integrating Multi-Source Remote Sensing and Machine Learning

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

Identification

• Journal: Atmosphere
• Year: 2026
• Date: 2026-09-23
• Authors: Jue Wang, Pengcheng Jia, Yang Li, Qiang Wang, Tao Yang
• DOI: 10.3390/atmos17100917

Research Groups

• Department of Atmospheric Science, University of California, Los Angeles (UCLA)
• Key Laboratory of Mountainous Environment and Ecological Restoration, Institute of Mountain Hazards and Environment, Chinese Academy of Sciences

Short Summary

This study explores the potential of multi-source ground-based remote sensing combined with machine learning to accurately estimate planetary boundary layer height (PBLH) in complex terrain environments. The research demonstrates that locally observation-driven machine learning can outperform global reanalysis for PBLH estimation.

Objective

• Investigate the feasibility and accuracy of using multi-source ground-based remote sensing data and machine learning algorithms to predict PBLH over karst plateaus, where ERA5 reanalysis shows significant systematic errors.

Study Configuration

• Spatial Scale: Local scale (Guiyang National Reference Climatological Station on the eastern Yunnan–Guizhou Plateau)
• Temporal Scale: 17 months of continuous observations from January 2024 to May 2025

Methodology and Data

• Models used: CatBoost gradient boosting model
• Data sources: Ground-based remote sensing data (temperature, humidity profiles, wind profiles, surface data) from the Guiyang National Reference Climatological Station; ERA5 reanalysis product for comparison

Main Results

• The CatBoost model achieved a coefficient of determination (R2) of 0.80 and a root mean square error (RMSE) of 290 m on an independent test set, outperforming ERA5.
• Incorporating vertical profile features improved the R2 by 0.093 compared to using surface observations alone.

Contributions

• This study provides valuable reference data for enhancing weather forecasting and air quality modeling in complex terrain areas.
• The results demonstrate that locally observation-driven machine learning can serve as a viable supplement or alternative to global reanalysis for PBLH estimation.

Funding

• National Natural Science Foundation of China (Grant No. 41875064)
• Key Research Program of the Chinese Academy of Sciences (Grant No. XDA19030401)

Citation

@article{Wang2026Planetary,
  author = {Wang, Jue and Jia, Pengcheng and Li, Yang and Wang, Qiang and Yang, Tao},
  title = {Planetary Boundary Layer Height Prediction over Karst Plateau: A Study Integrating Multi-Source Remote Sensing and Machine Learning},
  journal = {Atmosphere},
  year = {2026},
  doi = {10.3390/atmos17100917},
  url = {https://doi.org/10.3390/atmos17100917}
}