Geolocation of RFIs by Multiple Snapshot Difference Method for Synthetic Aperture Interferometric Radiometer

Jin, Rong; Wu, Lidong; Li, Qingxia; Lu, Hanchao; Feng, Li

doi:10.1109/tgrs.2020.3045088

Cited by 8 publications

(1 citation statement)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(Park et al, 2016) has discussed MUltiple SIgnal Classification direction-of-arrival estimations for RFI detection. (Jin et al, 2022b) provided a finer accuracy by noticing the difference on multiple snapshots on the premise that the location of RFI sources remains unchanged, and further addressed the low to moderate contamination cases based on the point-ripple characteristic of RFI sources (Jin et al, 2022a). The above methods of tacking RFI in the very early stage of the signal processing chain, while theoretically feasible, are very likely to propagate errors through the data flow, leading to exacerbated inaccuracies on the retrieved SSS.…”

Section: Related Workmentioning

confidence: 99%

RFI mitigation for 2D Synthetic Aperture Interferometric Radiometers using combined theoretical and machine learning technique

Xu,

Li,

Yin

2023

Front. Mar. Sci.

View full text Add to dashboard Cite

Synthetic Aperture Interferometric Radiometer (SAIR) as one of the most advanced instruments for Sea Surface Salinity (SSS) observation, has been in service on SMOS mission for years and is planned on the Chinese Ocean Salinity Satellite in the near future. However, a lot of Radio Frequency Interference (RFI) emissions are found in SMOS views, which contaminate the brightness temperature measurements of the SAIR instrument, and further impede the retrieval of SSS fields. Concerning SAIR’s operating mode, this study proposes an RFI mitigation method comprising two algorithms for co- and cross-polarization, respectively. First, RFI signatures are identified based on a series of thresholds defined by radiation theory, and then mitigated through a simple machine learning technique of Support Vector Regression (SVR), leveraging either SAIR’s multi-angle measurements or sea surface roughness descriptors, depending on the specific polarization mode. Finally, the outputs of all polarizations are merged and written back to the Level 1C brightness temperature product as the final result. Using the proposed method, the notable outliers arose from RFI contamination are attenuated, and the variation of standard deviations over nearby snapshots is smoothed, as expected on a homogeneous ocean. Furthermore, with the official L2OS software implementing the SSS retrieval procedure from the rewritten Level 1C brightness temperatures, the data re-gain of SSS fields is achieved in some places that are not attainable for the current SMOS Level 2 SSS products, with a reasonable error compared to WOA2009 SSS, confirming the validity of the proposed method. Hopefully, this work could provide a practical solution to current and future SAIR observing predicaments.

show abstract

Section: Related Workmentioning

confidence: 99%