Xu et al. (2025) A Generalized New Method for Anomalous Phased Array Radar Echo Image Restoration Based on Generative Adversarial Network

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

Identification

• Journal: Earth and Space Science
• Year: 2025
• Date: 2025-11-27
• Authors: Jinyan Xu, Ling Yang, Xiaoqiong Zhen, Yan Fu, Zhendong Yao, Chong Wu, Chao Chen
• DOI: 10.1029/2025ea004262

Research Groups

Not specified in the provided abstract.

Short Summary

This paper proposes a novel deep learning model, GCD, for restoring X-band phased array radar echo images, effectively addressing various data quality issues like echo voids and radial obstructions. The GCD model significantly improves restoration quality, particularly for strong echoes, and drastically reduces processing time compared to traditional methods.

Objective

• To develop a radar echo image restoration model (GCD) based on adversarial generative networks that effectively addresses various data quality problems (echo voids, abnormal radials, radial obstructions, and irregular missing echoes) in X-band phased array radar PPI images, with a focus on improving strong echo restoration and processing efficiency.

Study Configuration

• Spatial Scale: Radar image pixel level (applied to PPI images).
• Temporal Scale: Per radar scan/image (for initial data processing and single-image testing).

Methodology and Data

• Models used: Radar echo image restoration model (GCD) based on color correction and detail enhancement of adversarial generative networks with a dual-stream encoder-decoder. It incorporates a multi-scale Feature Alignment (FA)-based strong echo color correction module and a Local Detail Enhancement Module.
• Data sources: X-band Phased array radar PPI images.

Main Results

• The GCD model is applicable to all types of radar PPI images.
• It significantly reduces the time required for initial data processing, achieving a speedup ratio of 361.28 for single-image testing compared to the traditional Sliding Window Filling method.
• For radial obstructions, GCD achieves an improvement of 10.15 in PSNR and a notable decrease of 18.921 dBZ in the absolute difference of reflectivity (|dBZ|), resolving the issue of false enhancement of meteorological echo edges.
• Compared to a base model, GCD further enhances the restoration of strong echoes, with a decrease in |dBZ| by 0.654.

Contributions

• Proposes a novel deep learning model (GCD) for comprehensive radar echo image restoration, addressing multiple data quality issues in X-band phased array radar PPI images.
• Fills a gap in the field of radar data quality control by applying advanced image processing techniques.
• Significantly improves restoration quality (especially for strong echoes and radial obstructions) and processing speed compared to traditional methods.
• Promotes the development and application of artificial intelligence in radar data quality control.

Funding

Not specified in the provided abstract.

Citation

@article{Xu2025Generalized,
  author = {Xu, Jinyan and Yang, Ling and Zhen, Xiaoqiong and Fu, Yan and Yao, Zhendong and Wu, Chong and Chen, Chao},
  title = {A Generalized New Method for Anomalous Phased Array Radar Echo Image Restoration Based on Generative Adversarial Network},
  journal = {Earth and Space Science},
  year = {2025},
  doi = {10.1029/2025ea004262},
  url = {https://doi.org/10.1029/2025ea004262}
}