Comparison of Orbit-Based and Time-Offset-Based Geometric Correction Models for SAR Satellite Imagery Based on Error...

Area of Science:

• Remote Sensing
• Geomatics Engineering
• Photogrammetry

Background:

• Geometric correction is crucial for accurate Synthetic Aperture Radar (SAR) satellite imagery geolocation.
• Selecting precise geometric correction models and parameters is essential for achieving sub-pixel accuracy.
• Current challenges exist for general users in adopting suitable and precise geometric correction models.

Purpose of the Study:

• To evaluate and compare the performance of orbit-based and time-offset-based geometric correction models for SAR imagery.
• To assess the suitability of these models under varying levels of satellite orbit accuracy.
• To determine the best model for achieving sub-pixel geolocation accuracy with different SAR datasets.

Main Methods:

• Error simulation was employed to evaluate geometric correction models.
• Radarsat-1 imagery, known for orbit errors, and TerraSAR-X imagery, with high-accuracy data, were used for evaluation.
• Performance was assessed based on achieved geolocation accuracy.

Main Results:

• For Radarsat-1 imagery, the orbit-based geometric correction model demonstrated superior performance compared to the time-offset-based model.
• Both orbit-based and time-offset-based geometric correction models exhibited similar performance for TerraSAR-X imagery.
• Sub-pixel geolocation accuracy was achievable with both models for TerraSAR-X data.

Conclusions:

• The choice of geometric correction model is critical and depends on the characteristics of the SAR imagery, particularly satellite orbit accuracy.
• Orbit-based models are recommended for SAR imagery with significant satellite orbit errors, such as Radarsat-1.
• For high-accuracy satellite data like TerraSAR-X, both evaluated models can effectively achieve sub-pixel geolocation accuracy.