Polarimetric calibration of linear dual-pol SAR when corner reflectors are unavailable

Abstract

Dual-pol synthetic aperture radar (SAR) has a higher resolution and broader swath than the quad-pol system. However, calibration of dual-pol SAR is challenging because only two-channel observations exist: HH&VH and HV&VV. In external calibration, the timely and periodic deployment of corner reflectors (CRs) increases costs when calibration factors vary over time. The effectiveness of such calibration can be increased when the target scattering in the scene is known. This study proposes a novel method that considers a distortion-free image as a benchmark to calibrate the receiving distortion of other products even within several years. The high-correlation targets in the time-series images present stable scattering mechanisms and provide a robust reference to estimate the radiometric calibration factor and channel imbalance (CI). The vegetation with azimuth symmetry is proposed to evaluate the crosstalk (XT) further. The novel technique is tested on dual-pol data of Gaofen-3 and Sentinel-1A. Results confirm that the residual XT is lower than − 35 dB, and the CI error is better than 0.25 dB with a 5° phase error after calibration. Our novel method can calibrate dual-pol data without using CR and reduce the workload in external calibration. The source codes are provided on the website for benchmarking and reproducibility purposes (http://jszy.whu.edu.cn/sl/zh_CN/jxzy/428540/list/index.htm).