MONITORING OF RFI LOCALIZATIONS FOR THE SMOS MISSION: SEASONAL VARIATIONS AND SYSTEMATIC ERRORS Yan Soldo (1,2) , Ali Khazaal (1) , Ewa Słomińska (3) , François Cabot (1,2) , Rémy Fieuzal (1) , Yann H. Kerr (1,2) (1) CESBIO, 18 av. Edouard Belin, Toulouse, France (2) CNES, 18 av. Edouard Belin, Toulouse, France (3) Space Research Centre PAS, 18 A. Bartycka, Warsaw, Poland ABSTRACT Artificial sources emitting in the protected part of the L- band are polluting the retrievals of ESA’s Soil Moisture and Ocean Salinity (SMOS) satellite. Detection and localization of such sources are of interest for the exploitation of science products as well as for the identification of the emitters. A simple and fast method that provides snapshot-wise information is presented. From a statistical analysis of the results, some systematic errors are reported along with their potential causes and an approach to mitigate them. In the case of sources at high geomagnetic latitudes a seasonal variation of the localization error is also noticed; the origin of such phenomenon is still under investigation. Index Terms— SMOS, RFI, detection, localization, systematic errors 1. INTRODUCTION Since the launch of ESA’s Soil Moisture and Ocean Salinity (SMOS) [1] satellite in November 2009, a large number of Radio Frequency Interferences (RFI) in the protected part of the L-Band (1400-1427 MHz) have been detected. These artificial sources pollute the geophysical thermal noise that is measured by the passive interferometric radiometer MIRAS (Microwave Imaging Radiometer with Aperture Synthesis), that constitutes the sole payload aboard SMOS. The extent of RFI contamination is so large that it represents one of the major sources of error in the retrieval of both soil moisture and ocean salinity, and in some region hardly any observations can be used [2]. For this reason an effort has been made to identify the emitters [3] and to compensate for the signal introduced by the interferences [4, 5]. In the present contribution a simple and fast algorithm for early detection and localization of such sources is presented, along with its performances on both simulated and real data. With this method a statistically consistent dataset of localization was created thus allowing the investigation of tendencies and systematic errors. 2. INSTRUMENT FIELD OF VIEW SMOS’ snapshots are produced in the region of the Field Of View (FOV) that is inside the hexagon (where reconstruction is done) and outside Earth’s aliases; this region is called the Extended Alias-Free FOV (EAFFOV). Nevertheless RFIs (or their aliases) may be present outside the EAFFOV and they must be taken into account when coping with RFIs, since particularly bright point sources anywhere in the FOV will have a non-negligible effect in all other grid points of the FOV [5]. To have an idea of the proportion between the region that is seen by the instrument and the region inside the EAFFOV, consider that SMOS swaths are about 1000 km wide [6], and that in the unit circle is represented almost 4% of Earth surface. 3. DETECTION AND FAST LOCALIZATION The method presented in this contribution evaluates first the BTs Fourier Components inside the hexagon, in order to provide a view of the whole scene. Then clusters are formed around the local maxima according to criteria on the BT distribution and on the BT gradients in the principal directions of the grid. The barycenters of these clusters will be used as first guess for the RFI position on the (ξ, η) plane. With an optimization process we define a more precise position which corresponds to the point of maximum BT. Note that the optimization stops when a tolerance of 1E-6 (distance in (ξ, η) units) is reached, so that the resulting position will not correspond to a point of the sampling grid. Note also that the sum of the natural thermal emission and the artificial emission can vary, in extreme cases, as much as 2.5 orders of magnitude. As a consequence a particularly strong source can hide weaker ones. Since no mitigation is done, for each snapshot only the sources that have similar intensities (less than 1 order of magnitude) will be identified. The sources are then localized on ground only when lying in the EAFFOV, to avoid confusion between real sources and their aliases. 1912 978-1-4799-1114-1/13/$31.00 ©2013 IEEE IGARSS 2013