Toward an Optimal SMOS Ocean Salinity Inversion Algorithm

As part of the preparation for the European Space Agency´s Soil Moisture and Ocean Salinity (SMOS) satellite mission, empirical sea-surface emissivity (forward) models have been used to retrieve sea-surface salinity from L-band brightness-temperature (T _B) measurements. However, the salinity inversion is not straightforward, and substantial effort is required to define the most appropriate cost function. Various Bayesian-based configurations of the cost function are examined, depending on whether a priori information is used in the inversion. A sensitivity analysis of T _B to several geophysical parameters has been performed and has shown that the instrument has low sensitivity to the parameters that modulate the T _B (including salinity). The SMOS end-to-end simulator is used to test the accuracy of different cost-function configurations. Currently, the general opinion in the SMOS community is that a partially constrained cost function, in which the salinity constraint is effectively removed, is the most appropriate for salinity retrieval. The purpose of this letter is to show that we found no evidence that such a configuration performs better than a fully constrained or a nonconstrained one. Moreover, in contrast to previous results, we found that the fully constrained inversion does not converge to the reference or auxiliary salinity value and produces the most accurate salinity retrievals of the tested configurations. Therefore, such a configuration should not be disregarded for future tests.