Quality analysis and improvement of polarimetric synthetic aperture radar (SAR) images from the GaoFen-3 satellite using the Amazon rainforest as an example

Abstract

ABSTRACT As China’s first civil synthetic aperture radar (SAR) satellite, the data quality of the GaoFen-3 (GF-3) satellite has been widely studied. This study analyzes the data quality of this satellite from a polarimetric perspective and recalibrates the GF-3 images using the Amazon rainforest. Four calibration approaches, including the Ainsworth method, the Quegan method, a method based on rotation symmetry (referred to as the rotation method), and a newly proposed polarimetric calibration method using only the rainforest (referred to as the PCMRF method), are used to improve the SAR image quality. For the channel imbalance correction, the four methods demonstrate similar performances. For cross-talk correction, the proposed method exhibits a better performance than the other three calibration methods. These four methods provide complementary assessments with respect to the evaluation and the calibration performance. The test results using the rainforest data indicate that the channel imbalances are reduced from 0.48 dB to 0.1 dB in amplitude, from 16° to 4° in phase, and from −36 dB to −46 dB in cross-talk. The calibration effects are further validated using several points and distributed targets in different scenes, and the suggested calibration parameters are given in this study to increase calibration viability.