The RSS/NASA SMAP Salinity Version 3 Release:
Resampling and Correction for Land Contamination
[27-Aug-2018] Wentz, F., Meissner, T., and Manaster, A.
Presented at the 2018 Ocean Salinity Science Team and Salinity Continuity Processing Meeting
This presentation discusses the resampling and land contamination correction in the RSS SMAP Salinity Version 3 release.
SMAP V3 provides Level 2 and Level 3 SSS products at two different spatial resolutions: 40 km and 70 km. In both cases the sampling uses a Backus-Gilbert type Optimum Interpolation (OI) onto a fixed 0.25° Earth grid with a synthetically defined target footprint. The targets of the 40-km product are the same elliptical cells as in the L1B brightness temperature that are produced by the SMAP L1 algorithm and that are used as the basic sensor input in the salinity retrievals. They are different for fore and aft looks. The target of the 70-km is a circular footprint with a 70 km diameter. Both products have well-defined spatial response and a well-defined noise reduction. The noise in the 40-km L2 salinity product is about 0.9 psu. In the 70-km L2 salinity product the noise gets significantly beaten down to close to 0.5 psu. Further noise reduction in both products can be achieved by temporal averaging to the 8-day or monthly Level 3 products.
SMAP V3 uses an upgraded sidelobe correction for the correction for land contamination. The land correction in SMAP Version 2 had tried a shortcut by assuming an exact 8-day repeat orbit with fixed equatorial crossing times and longitudes. It turned out that this assumption is not valid accurately enough for the real SMAP orbit. This lead to inaccuracies in the land correction in V2. In V3 the land correction is derived using the full degrees of freedom without making assumptions about the equatorial time and location of the SMAP orbit. As for Aquarius V5, the SMAP V3 land correction is based on a pre-computed correction table. At a given location in the ocean near the coast we compute the difference between the "real" antenna temperature (TA), which is obtained by integrating over the SMAP antenna gain pattern including the sidelobes that fall over land, and the TA from the main beam. This difference is stratified and tabulated as a function of longitude, latitude, part of the orbit (ascending or descending), the look azimuth of the SMAP antenna, and the time (month of the year). The step sizes are 0.5° for longitude and latitude and 12° for the azimuth angle. The computation is done for each month using climatology values for soil moisture and land surface temperature from NCEP GDAS as input to the land emissivity computation.
We will provide an analysis showing the degradation of the SMAP salinity retrievals as function of distance from the coast with and without applying the land contamination correction, for both the 40-km and the 70-km salinity products.