Liu et al. (2025) Feasibility Study of Microwave Radiometer Neural Network Modeling Method Based on Reanalysis Data

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

Identification

• Journal: Atmosphere
• Year: 2025
• Date: 2025-10-16
• Authors: Xuan Liu, Qinglin Zhu, Dong Xiang, Houcai Chen, Tingting Shu, Wenxin Wang, Bin Xu
• DOI: 10.3390/atmos16101194

Research Groups

Not explicitly stated in the provided text.

Short Summary

This study proposes and validates a neural network retrieval method, based on high-resolution FNL reanalysis data, to derive atmospheric profiles from microwave radiometer brightness temperatures, effectively addressing the challenge of limited radiosonde data availability in certain regions.

Objective

• To develop and validate a Back Propagation (BP) neural network model for retrieving atmospheric temperature, relative humidity, water vapor density, and integrated water vapor (IWV) profiles from microwave radiometer brightness temperatures, utilizing FNL reanalysis data, particularly for regions lacking radiosonde observations.

Study Configuration

• Spatial Scale: Local validation using data from Qingdao, China, with proposed applicability to broader regions lacking radiosonde data (e.g., oceans, plateaus).
• Temporal Scale: Analysis of seasonal variations, implying data spanning multiple seasons or at least one year.

Methodology and Data

• Models used: Back Propagation (BP) neural network.
• Data sources:
  • High-resolution Final Reanalysis (FNL) reanalysis data (from Qingdao, China) for model training.
  • Synchronous radiosonde data (from Qingdao, China) for model validation and accuracy assessment.
  • Microwave radiometer brightness temperature data (implicit input to the retrieval model).

Main Results

• The Root Mean Square Error (RMSE) for temperature profiles is 1.15 °C in the near-surface layer (0–2 km) and 2.05 °C in the mid-to-upper layers (>2 km).
• Comprehensive RMSE values are 17.27% for relative humidity, 0.96 g/m³ for water vapor density, and 1.37 mm for Integrated Water Vapor (IWV).
• Retrieval results show strong spatiotemporal consistency with radiosonde data, with overall small errors.
• The most rapid error increase is observed within the lower atmosphere (<2 km).
• Seasonal variations in accuracy were noted: temperature and relative humidity retrievals are most accurate in summer, while water vapor density and IWV retrievals perform best in winter and worst in summer.

Contributions

• Proposes a feasible and effective neural network retrieval method for microwave radiometer modeling in regions with limited or no radiosonde data coverage.
• Confirms that reanalysis data-based modeling can successfully overcome the issue of sparse radiosonde observations.
• Expands the potential applications of microwave radiometers and reanalysis data for atmospheric remote sensing in challenging environments like oceans and plateaus.

Funding

Not explicitly stated in the provided text.

Citation

@article{Liu2025Feasibility,
  author = {Liu, Xuan and Zhu, Qinglin and Xiang, Dong and Chen, Houcai and Shu, Tingting and Wang, Wenxin and Xu, Bin},
  title = {Feasibility Study of Microwave Radiometer Neural Network Modeling Method Based on Reanalysis Data},
  journal = {Atmosphere},
  year = {2025},
  doi = {10.3390/atmos16101194},
  url = {https://doi.org/10.3390/atmos16101194}
}