Towards a North Atlantic SMOS SSS product using a numerical model of the North Atlantic Ocean

Abstract

In a recently published work (Hoareau et al., 2013), we have shown that the salinity field resulting from assimilating SMOS Level 3 data (i.e. binned maps) into an ocean model has less noise than the input data. Moreover, by using the ocean model as a dynamical interpolator, the resulting fields have no datavoids and increased geophysical coherence. The region of study had been the Macaronesian Region (the Northeast subtropical Atlantic Gyre), where SMOS data is prone to large errors (about 0.50 in the practical salinity scale) due to the vicinity of the large continental masses and the presence of artificial Radio Frequency Interferences (RFI). Our work proved that a technique as simple as the Newtonian Relaxation (nudging) is able to assimilate SMOS in such a challenging region, provided that the spatial structure of the SMOS data is taken into account. To further investigate the ability of data assimilation to produce a basin-wide salinity product, we plan to assimilate SMOS data in a numerical simulation of the North Atlantic Ocean. The NATL025 DRAKKAR configuration is an eddy permitting simulation ranging from 20ºS to 80ºN. It has a sea-ice climatology in the Northern boundary to help maintaining a realistic Meridional Overturning Circulation. It has a free surface with z-coordinates on the vertical. The ocean model is coupled with the Louvain La Neuve Ice Model (LIM2.0). The horizontal grid is a Mercator grid with resolution of 1/4º. The bathymetry is the ORCA025-G70, and the vertical structure of the simulation is given by 45 geopotential levels, with a vertical thickness of 6 m at the surface and 250 m at the bottom. In a first phase towards the data assimilation of SMOS, the variability of the salinity and temperature fields provided by the model is characterized via first and second order statistics, empirical orthogonal functions and power spectra. The time period of our simulation covers the years 2001 to 2012. Com- parison against Argo data is also done to estimate the spatial structure of the model error. Finally, the spatial structure of the error of SMOS data in the North Atlantic region is estimated in order to ensure the proper application of the nudging algorithm