18 orbits in Eastern Pacific 10-22 August reprocessed by ARGANS using :-EAFFOV -only AFFOV-measurements in a square with Xi, Eta <0.25
Colocation with ARGO SSS at distance: +/-5days +/-50km
For each ARGO float, average colocated SMOS SSS weighted by the error of retrieved SSS and the resolution of the pixel:
<SSS>=(SSS/R2 )/(1/R2 )
Where R is the equivalent radius
Discriminate Ascending and Descending orbits
Compare EAFFOV and AFFOV SSS retrievals only on pixels common to AFFOV and EAFFOV retrievals (<=>compare only retrievals made in the central swath with more than 16 AFFOV meas. used in the retrieval)
Comparison of SMOS SSS with ARGO SSS under various Tb measurement selections
J. Boutin - N. Martin – X. Yin
TX-TXmodel
TY-TYmodel
std(TX-TXmodel)
std(TY-TYmodel)
Only SSS retrieved with more than 16 meas. in AFFOV are considered
EAFFOV
AFFOV
Ascending orbits
SSSsmos (psu) SSSsmos-SSSargo (psu) SSSsmos-SSSargo (psu)
32
37
-3
3
-3
3
AFFOV
EAFFOV
AFFOV
Ascending orbits
SSSsmos (psu) SSSsmos-SSSargo (psu) SSSsmos-SSSargo (psu)
32
37
-3
3
-3
3
AFFOV
Ascending orbits – 40S-20N to avoid spurious points – Model 1
AFFOV EAFFOV
AFFOV: std(SSSsmos-SSSargo)=0.65psuEAFFOV: std(SSSsmos-SSSargo)=0.54psu
AFFOV: r=0.80EAFFOV: r=0.86
Ascending orbits – 40S-20N to avoid spurious points in the South
Xi-Eta<0.25
Xi,Eta 0.25: r=0.84AFFOV: r=0.80EAFFOV: r=0.86
Xi, Eta 0.25: std(SSSsmos-SSSargo)=0.57psu AFOV: std(SSSsmos-SSSargo)=0.65psuEAFFOV: std(SSSsmos-SSSargo)=0.54psu
EAFFOV
Descending orbits
SSSsmos (psu) SSSsmos-SSSargo (psu) SSSsmos-SSSargo (psu)
32
37
-3
3
-3
3
AFFOV
EAFFOV
Descending orbits
SSSsmos (psu) SSSsmos-SSSargo (psu) SSSsmos-SSSargo (psu)
32
37
-3
3
-3
3
AFFOV
Descending orbits – 40S-10N to avoid spurious points in the North and South
AFFOV: std(SSSsmos-SSSargo)=0.74psuEAFFOV: std(SSSsmos-SSSargo)=0.59psu
AFFOV: r=0.76EAFFOV: r=0.83
Descending orbits – 40S-20N to avoid spurious points in the North and South
Xi-Eta<0.25
Xi,Eta 0.25 std(SSSsmos-SSSargo)=0.72psuAFFOV: std(SSSsmos-SSSargo)=0.74psuEAFFOV: std(SSSsmos-SSSargo)=0.59psu
Xi,Eta 0.25 r=0.73AFFOV: r=0.76EAFFOV: r=0.83
Better correlation and smaller rms differences between ARGO and SMOS SSS when EAFFOV measurements in the center of the track are considered
Similar behavior observed with models 2 and 3
Better results with Xi,Eta=0.25 than with AFFOV probably because spurious meas. in front of AFFOV are not considered
=> In the future, test a rectangular area of Xi Eta that include meas. in EAFFOV but exclude meas. at the far front and at the far back of EAFFOV
CONCLUSION (1)
Comparisons less scattered on ascending than on descending orbits
TY-TYmodel std(TY-TYmodel)
INFLUENCE OF THE NUMBER OF MEASUREMENTS USED IN THE RETRIEVAL ONTO SSSsmos-SSSargo (look at non averaged SSS differences)
Number of Tb used in the retrieval
Ascending orbits – AFFOV - 40S-20N to avoid spurious points – Model 1
SSSsmos-SSSargo
Number of colocations
INFLUENCE OF THE NUMBER OF MEASUREMENTS USED IN THE RETRIEVAL ONTO SSSsmos-SSSargo
AFFOV - Model 1
Ascending orbits
Descending orbits
Number of Tb used in the retrieval
50
Descending
INFLUENCE OF THE NUMBER OF MEASUREMENTS USED IN THE RETRIEVAL ONTO SSSsmos-SSSargo
Number of Tb used in the retrieval
Ascending
EAFFOV including Xi edges - Model 1
50
SSSsmos-SSSargo individual differences more scattered when the number of Tb used in the retrieval is small, especially when the EAFFOV measurements are used. A threshold of 50 measurements (instead of 16) as proposed by ARGANS seems reasonnable