This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. IEEE GEOSCIENCE AND REMOTE SENSING LETTERS 1 Mitigation of RFI Main Lobes in SMOS Snapshots by Bandpass Filtering Justino Martínez , Verónica González-Gambau , and Antonio Turiel, Member, IEEE Abstract—Since the beginning of the soil moisture and ocean salinity mission, the pervading presence of radio frequency interferences (RFI) has been one of the most problematic issues. The effect of an RFI is not just a hot spot but also six tails along the three main axes, and the general presence of ripples which degrade the quality of L1 brightness temperature snapshots. The standard mitigation technique is to apply an apodization (Blackman), but such a low-pass filter leaves traces of the tails and spreads the signal of the main lobes. New RFI mitigation techniques, such as nodal sampling, are very effective in reducing the impact of tails and ripples, but in some cases they lead to the spread of the RFI main lobe, with a significant loss of data on the affected area. In this letter, we propose a new technique to reduce their spread by an adaptive thresholding on a bandpass filtered version of the snapshot, with a significant recovery of data. Index Terms— Imaging, interferometry, remote sensing, radio frequency interferences (RFI), signal processing, soil moisture and ocean salinity (SMOS). I. I NTRODUCTION T HE scientific goal of the soil moisture and ocean salinity (SMOS) mission to monitor surface soil moisture over the continents and sea surface salinity over the oceans [1], [2]. The payload carried by SMOS is the so-called microwave imaging radiometer by aperture synthesis. This instrument has full polarimetric capabilities and by means of a Y-shape array of antennas produces 2-D images of brightness tempera- ture (T B ) in the microwave L-band (1.413 GHz). Nevertheless, due to the finite extent of the instrument and the fixed locations of the antennas, the signal is only sampled at certain spatial frequencies. Therefore, the impulse response of the instrument is affected by high-frequency artifacts (Gibbs phenomena). For that reason, any sharp transition, such as radio frequency interferences (RFI) sources, the Sun alias, land/sea/ice transi- tions, and even the Earth surface/sky transition, spawn artifacts with high-spatial frequency components. This contamination appears in form of ripples and sidelobes (tails) in the bright- ness temperature scenes [3]. The standard SMOS L1 processor uses a Blackman win- dow [4] to reduce the Gibbslike contamination. The use of the Manuscript received November 29, 2017; revised February 14, 2018; accepted March 20, 2018. This work was supported in part by the Ministry of Economy and Competitiveness, Spain and in part by FEDER EU through the National R+D Plans PROMISES Project under Grant ESP2015- 67549-C3-2-R and L-Band Project under Grant ESP2017-89463-C3-1-R. (Corresponding author: Justino Martínez.) The authors are with the Barcelona Expert Center, Institut de Ciències del Mar, CSIC, 08003 Barcelona, Spain (e-mail: justino@icm.csic.es; vgonzalez@ icm.csic.es; turiel@icm.csic.es). Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LGRS.2018.2818285 Blackman window reduces the amplitude of high-frequency components, at the cost of globally reducing the resolution of the image and producing a wider main lobe; it is essentially a low-pass filter. However, this approach only partially mitigates the problem [5], and it is quite unsatisfactory in the case of significant RFI sources. In past three years, a new image reconstruction method, designed to mitigate the Gibbslike contamination, has been developed for SMOS. Nodal sampling (NS) [6], [7] starts from the hypothesis that the measured signal is composed by a geophysical component and a perturbation fast oscillating in frequency. If the geophysical signal presents a slow variation as compared to the perturbation of the scale of the spatial resolution of the instrument, then the T B snapshots can be sampled at those points where the perturbation has a minimum contribution to the signal. These points are called nodal points, and can be found iteratively by spatial oversampling of the measured signal. This adaptive method has been shown to be a good methodology for the correction of ripples and sidelobes. In some cases, the main lobe of perturbations gets spread. The widening occurs where the iterative scheme is unable to attain an adequate equilibrium, typically in zones having very spatially concentrated high values of T B , resulting in a larger extension of the RFI [7]. Nevertheless, this is not a systematic effect and seems to be linked with the position of the RFI source inside of the SMOS pixel. The aim of this letter is to reduce the main lobe extension of RFI sources, no matter how the image reconstruction is carried out—using NS or standard processing—and mark as invalid the points that are directly corrupted by the presence of the RFI source and are thus unreliable. II. RFI SPREADING MITIGATION When RFI spreading takes place in NS, it mainly occurs for oversampling points located initially inside the zone affected by the RFI main lobe. That is, when the main lobe in the raw image has already some significant spread. In such a case, the actual nodal points lie in a more marginal zone of the oversampled pixel and the algorithm is not able to converge to those points. Therefore, it could be possible to obtain less contaminated and better representative choices of nodal points if the RFI main lobe is detected in the original image, i.e., when no apodization function is applied. Then, their T B values are changed to values similar to the surrounding ones and flagged as bad data. Finally, the oversampled points that converge to flagged positions are discarded as unreliable. The procedure is similar to the standard processing method. That is, to apply the Blackman window just after changing the 1545-598X © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.