02.06.2013 Views

SMOS L2 OS ATBD - ARGANS

SMOS L2 OS ATBD - ARGANS

SMOS L2 OS ATBD - ARGANS

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

64<br />

ICM-CSIC<br />

LOCEAN/SA/CETP<br />

IFREMER<br />

<strong>SM<strong>OS</strong></strong> <strong>L2</strong> <strong>OS</strong><br />

Algorithm Theoretical<br />

Baseline Document<br />

Doc: SO-TN-ARG-GS-0007<br />

Issue: 3 Rev: 9<br />

Date: 25 January 2013<br />

Page: 64<br />

where ais the azimuth angle between wind direction (w) and the azimuthal observation<br />

angle of radiometer look direction (r): aw-r with all these angles counted<br />

counterclockwise, with origin on x axis (mathematical convention).<br />

Tabulations of Th0, Th1, Th2, Tv0, Tv1, Tv2, U1, U2, V1 and V2 are provided, as functions<br />

of incidence angle, SSS, SST, wind speed assuming a neutral atmosphere, WSn. We will<br />

assume that Tbrough can be entirely described using WSn in future versions, depending on:<br />

-accuracy of ECMWF WS<br />

-possibility of deducing a reliable U* from ECMWF WS and Tair-sea<br />

-sensitivity of Tbrough to WS/U*<br />

we could add dependency on WS.<br />

It is proposed to use WSn components WSx and WSy from ECMWF to initialize the<br />

inversion. Wind speed and direction used to initialize Tb mod values are deduced from WSx<br />

(positive eastward) and WSy (positive northward) as:<br />

2<br />

WSn WSx WSy<br />

w = arctan(WSy/WSx)<br />

2<br />

The tabulation is provided for SSS between 30 and 39 psu, SST between 0 and 30°C, WSn<br />

between 2 and 30 ms -1 , and an incidence angle between 0 and 68 degrees. All the Tb rough<br />

are set to 0 for WSn=0. For the interpolation of the table, a switch allows to choose between a<br />

linear interpolator (following the method described in TGRD) and a Hermite interpolator.<br />

The Hermite interpolation results in a continuous theoretical error on SSS, but is<br />

computationally heavy. The linear interpolation is faster, but it results in a discontinuity of<br />

the theoretical error on SSS, linked to the non-linearity of the model for wind speeds of 7<br />

m/s.<br />

If one of the prior values of the retrieved geophysical parameters is out of the LUT range, or<br />

if any retrieved geophysical parameters goes out of LUT range during the retrieval, different<br />

flags (Fg_OoR_Rough_dim1, Fg_OoR_Rough_dim2, Fg_OoR_Rough_dim3,<br />

Fg_OoR_Rough_dim4) are raised. No extrapolation is done and the boundary value is taken.<br />

4.2.1.3. Error budget estimates (sensitivity analysis)<br />

The advantage of retrieving WSn is that it is comparable to scatterometer derived wind speed.<br />

4.2.2. Practical considerations<br />

4.2.2.1. Calibration and validation<br />

The algorithm described above was derived from L1c data v344 and the new LUT has been<br />

implemented in <strong>L2</strong><strong>OS</strong> v500.<br />

4.2.2.2. Quality control and diagnostics<br />

Values outside the min/max ranges given in the Tbrough tabulations should be deduced from<br />

a linear extrapolation of the two edge values of the tabulations (since most of the

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!