Water Detection in Urban Areas from GF-3

Area of Science:

• Earth Observation
• Remote Sensing Technology
• Urban Hydrology

Background:

• Accurate urban water detection is vital for management, river monitoring, and flood assessment.
• Existing methods struggle to differentiate water from building shadows in single Synthetic Aperture Radar (SAR) images.
• High-spatial-resolution SAR imagery offers potential for detailed urban environmental analysis.

Purpose of the Study:

• To develop an effective method for urban water detection using GaoFen-3 (GF-3) SAR images.
• To address the challenge of discriminating water bodies from building shadows in SAR data.
• To enhance the accuracy and reliability of water detection in complex urban environments.

Main Methods:

• A novel approach was developed to remove building shadows by identifying their correspondence with buildings.
• The method utilizes high-spatial-resolution GF-3 SAR images for urban water mapping.
• Experimental validation was conducted using six GF-3 SAR image datasets.

Main Results:

• The proposed method successfully removed building shadows, enabling precise water detection.
• Achieved a Probability of Detection (PD) of 98.36% and a Probability of False Alarm (PFA) of 1.91%.
• Significantly outperformed the traditional thresholding method, which had a maximum PD of 72.62% at a comparable PFA.

Conclusions:

• The new method effectively detects urban water bodies with high precision by eliminating building shadow interference.
• Demonstrates the significant potential of high-spatial-resolution GF-3 SAR imagery for urban water resource management and disaster assessment.
• Offers a robust solution for improving water detection accuracy in challenging urban SAR datasets.