Sea Surface Wind Speed Retrieval from Marine Radar Image Sequences Based on GLCM-Derived Texture Features

Area of Science:

• Marine Meteorology
• Remote Sensing
• Oceanography

Background:

• Sea surface wind speed is crucial for marine safety and offshore operations.
• Existing radar wind speed retrieval methods lack adaptability and accuracy.
• Empirical models struggle with generalization and precision.

Purpose of the Study:

• To develop a novel sea surface wind speed retrieval method using X-band marine radar.
• To leverage Gray-Level Co-occurrence Matrix (GLCM) texture features for improved accuracy.
• To enhance the environmental adaptability and applicability of wind speed estimation.

Main Methods:

• Utilized X-band marine radar image sequences for wind speed retrieval.
• Applied a three-stage preprocessing pipeline: noise suppression, geometric correction, and interpolation.
• Extracted wind streaks and analyzed GLCM texture features (energy and entropy).
• Developed a segmented dual-parameter wind speed model based on identified features.

Main Results:

• The proposed GLCM texture-based models significantly outperform traditional empirical models.
• Energy and entropy models reduced mean errors by approximately 49.3% and 16.7%, respectively.
• The energy stable model achieved a high correlation coefficient (0.89); the entropy model excelled at low wind speeds.

Conclusions:

• The novel GLCM texture-based method offers a more accurate and efficient solution for sea surface wind speed retrieval.
• The segmented dual-parameter model enhances robustness across varying environmental conditions.
• This approach provides a valuable advancement for marine meteorology and related applications.