Xu et al. (2026) A global all-weather PWV retrieval model integrating multi-band satellite observations considering land cover types and NDVI

Identification

• Journal: Remote Sensing of Environment
• Year: 2026
• Date: 2026-03-26
• Authors: Yan Xu, Yubo Wang, Ning Jiang, Ao Guo, Tianhe Xu
• DOI: 10.1016/j.rse.2026.115387

Research Groups

• Institute of Space Sciences, Shandong University, Weihai, China
• China Electronics Technology Group Corp 22nd Research Institute Qingdao Research Center, Qingdao, China

Short Summary

This study developed an all-weather, high-resolution global model for retrieving precipitable water vapor (PWV) by integrating multi-band satellite observations (NIR, TIR, MW) with GNSS data, achieving a substantial reduction in global average RMSE from 15.19 mm to 5.37 mm compared to the MYD05 product.

Objective

• To develop an all-weather, high-resolution global model for retrieving precipitable water vapor (PWV) by integrating multi-band satellite observations (near-infrared, thermal infrared, and microwave data) with Global Navigation Satellite System (GNSS) observations as a reference, considering land cover types and Normalized Difference Vegetation Index (NDVI).

Study Configuration

• Spatial Scale: Global
• Temporal Scale: All-weather, high-resolution (aiming to overcome limitations of satellite revisit intervals)

Methodology and Data

• Models used: A novel all-weather, high-resolution global PWV retrieval model was developed, integrating multi-band satellite observations and considering land cover types and NDVI.
• Data sources:
  • Near-infrared (NIR) satellite observations
  • Thermal infrared (TIR) satellite observations
  • Microwave (MW) satellite observations
  • Global Navigation Satellite System (GNSS) observations (as reference)
  • Land cover types
  • Normalized Difference Vegetation Index (NDVI)

Main Results

• The proposed multi-band model reduced the global average Root Mean Square Error (RMSE) for PWV retrieval from 15.19 mm (MYD05 product) to 5.37 mm, representing a 64.66% improvement.
• The model significantly decreased spatial heterogeneity in PWV retrieval accuracy.
• In challenging environments such as forested and urban areas, the model improved RMSE from 17.90 mm and 16.08 mm (conventional algorithms) to 6.51 mm and 5.60 mm, respectively.
• The multi-band model demonstrated strong applicability and robustness across complex water vapor fields under cloudy, clear sky, and uncertain weather patterns, as well as on a global scale.
• The study elucidated the interactions between different spectral bands and land-atmosphere parameters in water vapor inversion, highlighting the pivotal role of a cyclic mechanism.

Contributions

• Development of a novel, dynamic, and adaptive all-weather, high-resolution global PWV retrieval framework that integrates multi-band satellite observations (NIR, TIR, MW) and GNSS data, accounting for land cover types and NDVI.
• Significant improvement in global PWV retrieval accuracy and reduction in spatial heterogeneity compared to existing products (e.g., MYD05).
• Enhanced accuracy and robustness of PWV retrieval in traditionally challenging environments (e.g., forested, urban areas) and under diverse meteorological conditions (cloudy, clear sky, uncertain weather).
• Provision of a comprehensive framework for global water vapor monitoring adaptable to varying terrain and meteorological conditions.

Funding

Not specified in the provided text.

Citation

@article{Xu2026global,
  author = {Xu, Yan and Wang, Yubo and Jiang, Ning and Guo, Ao and Xu, Tianhe},
  title = {A global all-weather PWV retrieval model integrating multi-band satellite observations considering land cover types and NDVI},
  journal = {Remote Sensing of Environment},
  year = {2026},
  doi = {10.1016/j.rse.2026.115387},
  url = {https://doi.org/10.1016/j.rse.2026.115387}
}