Towards an Optimal SMOS Bayesian-Based Inversion Scheme for Salinity and Wind Speed Retrieval Purposes

Abstract

In this study, a simplified parallel version of the operational Soil Moisture and Ocean Salinity (SMOS) Level 2 data processor (L2OS) is used to assess the optimal configuration of both the SMOS Bayesian-based cost function and the corresponding Levenberg-Marquardt based multi-parametric minimization scheme for sea surface salinity (SSS) and sea surface wind speed (U10) retrievals. Within such a framework, realistically simulated brightness temperature measurements (TBs) and a post-launch derived semi-empirical forward model (FM) are used. As already anticipated in pre-launch studies, the results carried out in this work demonstrate that the SMOS cost function needs to be constrained by a priori information on the sea-surface related geophysical parameters, due to the low TB sensitivity with respect to both SSS and U10 changes. Furthermore, the present study, in which a variety of marine scenarios are simulated, confirms that the operationally used SMOS L2OS cost function configuration is optimal for SSS retrieval purposes. As a novelty, this study shows that the optimal SMOS cost function configuration for U10 retrievals is different from the one used for SSS retrieval purposes. This is mainly due to the TB sensitivity to SSS changes being different to the TB sensitivity to U10 changes. Current work focuses on the verification of the simulation results with SMOS real data, using collocated in situ observations and numerical weather prediction model output. The most relevant results of this study will be presented at the workshop