Calibration-Validation Activities

SMOS BIASES over the Ocean   A key goal of the European Space Agency’s (ESA) Soil Moisture and Ocean Salinity (SMOS) mission, launched in November 2009, is to produce global maps of ocean surface salinity with an accuracy of 0.1 (on the Practical Salinity Scale 1978) over a time scale of 10 days and at a spatial resolution of about 200 km using measurements of upwelling L-band radiation obtained from the spaceborne L-band interferometric radiometer MIRAS. This is a challenging objective for several reasons. First, the sensitivity of L-band brightnes temperatures to variations in SSS,  is at best about 1 K per psu for total power (or about 0.5 K per psu for each polarization) in the tropics. This sensitivity is very weak given that spatial and temporal variability in open-ocean SSS does not exceed several psu and that the reconstructed brightness temperature noise for an integration time of 1.2 s is typically several kelvin. Second, as we shall see below, surface roughness emission and land contamination impact the reconstructed brightness temperatures at a level very significant for SSS retrieval. Third, the scene brightness model used to derive SSS is plagued with inaccuracies, especially with respect to surface excess emission, sun glint and scattered galactic noise. As the reconstructed scene brightness temperature noise is quite high, significant averaging is required to achieve the desired accuracy in retrieved SSS, and so we must pay particular attention to biases that may depend upon the scene or evolve over time. Even in the open ocean far from land the reconstructed brightness temperatures (which are actually averaged over synthetic beams for which the beamwidths are determined by the frequency coverage of the instrument and the Blackman spatial filter applied to the brightness temperature Fourier coefficients) exhibit significant bias with respect to our forward model, and this bias varies significantly over the field of view of the instrument. Examples of the spatial patterns in the bias for the two orthogonal linear polarizations X and Y (the instrument polarization basis) are shown in Figs. 2(a) and 2(b). The bias reaches several kelvin even in the alias-free field of view, which translates to a bias of several psu according to the sensitivity.   Figure 2: (a) and (b): Examples of bias patterns (the mean of data minus model in X and Y polarizations evaluated using snapshots between 50◦S and 10◦N latitude) over the MIRAS field-of-view plotted in director cosine coordinates; (c) Variation of the scene-averaged bias (averaged over the alias-free domain only) between data and model (Tx + Ty)/2 as a function of time and latitude and (d): Same as (c) except the difference between descending and ascending passes. The mean bias averaged over both the alias-free and extended alias-free domains and over each half-orbit (averaged between 50◦S and 10◦N latitude), shown in Fig.3, drifted significantly from the beginning of April through May 2010. Moreover, Fig.3(c) shows that during this same period there is a consistent bias (about 0.5 K) between ascending and descending passes (shown in Fig.3(c)), and this is reflected in SSS retrievals shown later. The sudden drop in the biases near June 9, 2010 is associated with a change in the Level 1 Operational Processor, which performs the image reconstruction after correcting for biases in the visibilities. Given the typical sensitivity of total power to changes in SSS, the bias trends translate into bias changes of several psu and will therefore impact any conclusions regarding SSS trends unless corrected.

Č

Updating...

ć

Contrib_NR_REVEX.pptx

(26402k)

Salinity CERSAT,

Nov 23, 2010, 2:13 AM

v.1

ď

Comments