Wang et al. (2026) PES-UNet: A Physics-Inspired Enhanced Hybrid Network for Cloud Classification from FY-4A Satellite Data

Identification

• Journal: IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
• Year: 2026
• Date: 2026-01-01
• Authors: B. Xinhua Wang, Huitang Li, Wei Cheng, Qinghong Sheng, Yang Du, Jun Li, Xiao Ling
• DOI: 10.1109/jstars.2026.3664458

Research Groups

Not available in the provided text.

Short Summary

This paper introduces PES-UNet, a physics-inspired enhanced hybrid deep learning network, designed to improve cloud classification accuracy using data from the FY-4A geostationary meteorological satellite.

Objective

• To develop and evaluate PES-UNet, a physics-inspired enhanced hybrid network, for accurate cloud classification from FY-4A satellite data.

Study Configuration

• Spatial Scale: Global/regional (implied by FY-4A geostationary satellite, specific extent not detailed in provided text).
• Temporal Scale: Not specified in provided text (likely related to FY-4A observation periods).

Methodology and Data

• Models used: PES-UNet (a physics-inspired enhanced hybrid network), likely building upon the UNet architecture.
• Data sources: FY-4A Satellite Data (geostationary meteorological satellite).

Main Results

• Specific quantitative results are not available in the provided text, but the study aims to demonstrate improved cloud classification performance using PES-UNet.

Contributions

• Introduction of PES-UNet, a novel physics-inspired hybrid network architecture, for advanced cloud classification, specifically leveraging FY-4A satellite imagery.

Funding

• Not available in the provided text.

Citation

@article{Wang2026PESUNet,
  author = {Wang, B. Xinhua and Li, Huitang and Cheng, Wei and Sheng, Qinghong and Du, Yang and Li, Jun and Ling, Xiao},
  title = {PES-UNet: A Physics-Inspired Enhanced Hybrid Network for Cloud Classification from FY-4A Satellite Data},
  journal = {IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing},
  year = {2026},
  doi = {10.1109/jstars.2026.3664458},
  url = {https://doi.org/10.1109/jstars.2026.3664458}
}