RFI Mitigation in SAR images Exploiting Underlying Correlation Among Long-Term Multi-Temporal Observations

Unexpected radio frequency interference (RFI) artifacts in synthetic aperture radar (SAR) images may result in substantial loss of valuable Earth observations, largely due to the presence of numerous radio devices operating in adjacent frequency bands. Effective RFI mitigation relies on the distinct characteristics of interference relative to the targets’ signals, as well as the varied behaviors exhibited by different interference types. However, most methods do not make full use of the variation relationship between interference patterns in long-term observational periods. To solve this problem, an RFI characterization and mitigation method based on long-term multi-temporal SAR observations is proposed, in which interference awareness and mitigation are integrated. Firstly, the identification and detection of narrowband interference (NBI) and wideband interference (WBI) is implemented. Then, the RFI mask is extracted using Gabor filtering with different orientations, guided by the estimated prior interference characteristics. Finally, the interference-mitigated echo is subsequently obtained by notched filtering. Using real measured Sentinel-1 observations over the Gaza region in 2024, RFI characterization is estimated, and artifacts are effectively suppressed.