Towards a coherent sea surface salinity product from SMOS radiometric measurements and ARGO buoys

The SMOS (Soil Moisture and Ocean Salinity) Mission is the second of the ESA\´s Living Planet Programme Earth Explorer Opportunity Missions and it is scheduled for launch in 2008. Its objective is to provide global and frequent soil moisture (SM) and sea surface salinity (SSS) maps. SMOS\´ single payload is the Microwave Imaging Radiometer by Aperture Synthesis (MIRAS) sensor, an L-band two-dimensional aperture synthesis interferometric radiometer. To help in the retrieval process, auxiliary data will be used in combination with the brightness temperatures measured by MIRAS. In the salinity retrieval case, the main data sources are the sea state, the sea surface temperature, and the salinity values (even at low density sampling), which would allow a more accurate retrieval and a better quality product. One of the obvious candidates of auxiliary data is the ARGO buoy array. In this study, two different algorithms are proposed to use the ARGO salinity measurements in the retrieval procedure and provide coherent SSS maps with these measurements: The first algorithm is referred to as the "external brightness temperature calibration" [1] and aims at eliminating the bias introduced by the instrument errors and the image reconstruction process at the brightness temperature level; The second algorithm is referred to as the "external salinity calibration", and aims at correcting the biases introduced by the retrieval algorithm [2] as well, due to an imperfect dielectric constant model and an imperfect sea state parameterization in the emissivity model. Using the SMOS End-to-end Performance Simulator (SEPS) [3] as a source of SMOS-like brightness temperatures and the ARGO data as the source of auxiliary parameters, a series of simulations have been performed both in open ocean, and in coastal regions using a SMOS Level 2 Processor Simulator (L2PS) developed by the Universitat Politecnica de Catalunya (UPC). The performances of both algorithms are compared, and simulation resu- lts are presented and discussed.