1 / 85

In L2 v6 => latitudinal biases are reduced wrt L2 v5 due to TEC

About L2OS v6 improvement wrt L2OS v5 N. Martin – J.L. Vergely - J. Boutin Descending orbits results. In L2 v6 => latitudinal biases are reduced wrt L2 v5 due to TEC. Slides from QWG 13. TEC retrieval in L2 reduces latitudinal biases in descending orbits (tests conducted using same L1c).

tanek
Download Presentation

In L2 v6 => latitudinal biases are reduced wrt L2 v5 due to TEC

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. About L2OS v6 improvement wrt L2OS v5N. Martin – J.L. Vergely - J. BoutinDescending orbits results In L2 v6 => latitudinal biases are reduced wrt L2 v5 due to TEC

  2. Slides from QWG 13 TEC retrieval in L2 reduces latitudinal biases in descending orbits (tests conducted using same L1c)

  3. SMOS SSS - ISAS SSS: descending PCTVAR<80% (exclude regions with no ARGO) and 700km far from land The TEC retrieval in L2OS improves retrieved SSS compared with ISAS SSS. L2V550 OTT A3TEC on - L1cTEC prior –TEC dwell retr. L2V550 OTT L1cTEC – L1cTECprior – TEC dwell retr. L2V550 OTT L1cTEC – L1cTECprior – TEC dwell retr. Global: mean=0.11 (std=0.39) [45S45N;180W180E] : mean=0.05 (std=0.32) [30S30N;180W180E] : mean=-0.01 (std=0.29) L2V611b, OTT A3TEC on- A3TEC prior – no TEC dwell retr. L2V611b, OTT L1cTEC – A3TEC ‘prior’ -no TEC dwell retr. L2V611b, OTT A3TEC on- A3TEC prior – no TEC dwell retr. Global: mean=0.00(std=0.36) 45S-45N: mean=-0.01 (std=0.29) 30S-30N: mean=-0.02 (std=0.27)

  4. Suntail flag in L1c v620X. Yin1,2, J. Boutin2, J. Vergely3 ,P. Spurgeon1 1. ARGANS 2. LOCEAN 3. ACRI-ST Thanks to L1 team for providing L1C v620 product. Aims: • Validate the new suntail flag • Define the methodology at L2 to be used to flag data affected by sun

  5. 1) Data • 5 days ascending and descending orbits in May 2011 and November 2012, L1C v620 and L2OS v612 • L2OS without Suntail filtering processed by ARGANS • L2OS with Suntail filtering processed by LOCEAN 2) Method 2.1 Part1 : Validation of TB radiometric accuracy std(Tbsmos-Tbmodel) wrt radiometric accuracy The grid points in the FOV with number of measurements less than 30 are discarded 2.2 Part 2: Validation of SSS and Chi2p flagging Comparisons of monthly SSS w.r.t ISAS, and intercomparisons between SSS (CTRL_Chi2p flag) with or without filtering with L1 Suntail flag Consider only Tbs 1000km away from coasts to avoid land contamination SMOS SSS flags used CONTROL FLAGS : check to be set to 1 : CTRL_ECMWF check to be clear to 0 : CTRL_MANY_OUTLIERS;CTRL_SUNGLINT;CTRL_MOONGLINT;CTRL_REACH_MAXITER;CTRL_MARQ;CTRL_CHI2_P;CTRL_SUSPECT_RFI SCIENCE FLAGS : check to be set to 1 : SC_LOW_WIND;SC_LAND_SEA_COAST1 check to be clear to 0 : SC_ICE;SC_SUSPECT_ICE +Dg_af_fov>130

  6. Part1 Validation of TB radiometric accuracy Ratio of std(Tbsmos-Tbmodel) wrt theoretical radiometric accuracy computed between 0N and 30N The grid points in the FOV with number of measurements less than 30 are discarded

  7. Descending orbits in the northern hemisphere are strongly affected in November 2012 (eclipse period)[std (TB-Tbmodel)/rad_acc.] of 1 orbit (20121118) in open ocean between 0N and30N No filtering Suntail + Sun point Suntails in FOV Suntail and Sun point flag remove some regions with strong Sun contamination. However, vertical stripes outside suntails are still seen in FOV (due to imperfect sun correction?).

  8. Same as previous slide except for 10 orbits Long integration snapshots with filtering Theoretical Radiometric accuracy i.e. Ra Std(TBsmos-TBmodel) i.e. StdTB Ratio: StdTB/Ra X pol. Y pol. Vertical stripes outside Suntails can be seen in FOV (due to imperfect sun correction?).

  9. Same as previous slide except for 10 orbits short integration snapshots with filtering Theoretical Radiometric accuracy i.e. Ra Std(TBsmos-TBmodel) i.e. StdTB Ratio: StdTB/Ra X pol. Y pol. Vertical stripes outside Suntails can be seen in FOV (due to imperfect sun correction?).

  10. Long integration snapshots, Ratio: StdTB/Ra, 10 orbits With filtering Without filtering X pol. Y pol. Suntail flag and Sun point flag remove some data contaminated by Sun, but they are not enough to flag ripples outside the tails.

  11. Ascending orbits in May 2011 (Sun alias in bottom left of FOV)TB of 1 orbit (20110526) in the OTT region (45S and 5S by black box) No filtering Suntail + Sun point Suntails in FOV The impact of Sun is much smaller in May than in November. Suntail and Sun point flag remove regions with contamination (maybe too much).

  12. Same as previous slide except for 10 orbits Long integration snapshots with filtering Theoretical Radiometric accuracy i.e. Ra Std(TBsmos-TBmodel) i.e. StdTB Ratio: StdTB/Ra X pol. Y pol. Suntail and Sun point flag remove regions with contamination.

  13. Same as previous slide except for 10 orbits short integration snapshots with filtering Theoretical Radiometric accuracy i.e. Ra Std(TBsmos-TBmodel) i.e. StdTB Ratio: StdTB/Ra X pol. Y pol. Suntail and Sun point flag remove regions with contamination.

  14. Conclusion and discussion of Part 1 (TB validation) 1. For ascending orbits in May with Sun alias in the bottom left of FOV, Suntail and Sun point flag remove regions with contamination but maybe too much. Question: Do we need to filter TB flagged by Suntail before SSS retrievals for ascending orbits given that impact of Sun is much smaller than the descending orbits in November? Be careful about the vertical tail in the FOV due to the Suntail flag. No or few TB can be used for SSS inversion along dwell lines near the vertical tail! More details in part 2. 2. For descending orbits in November with Sun alias in the left side of FOV, vertical stripes outside suntails, especially for X(H) pol., can be seen in FOV (due to imperfect sun correction?). The Suntail flag itself is not enough to flag those vertical stripes.

  15. Part 2 SSS and Chi2p 1. Comparisons of SMOS SSS w.r.t ISAS (monthly in situ interpolated maps) 2. Intercomparisons between SSS maps obtained with or without filtering L1c TB with Suntail flag and after cheching the L2OS Chi2P flag (chi2P : good TB fit if chi2P high. Chi2P > 0.05 in current processor. )

  16. One ascending orbit in May 2011 (Sun alias in bottom left of FOV)SSS of one orbit (20110526) SSSwithfilt-SSSISAS SSSnofilt-SSSISAS SSSwithfilt-SSSnofilt With L1c suntail filtering Without L1c suntail filtering Suntails in FOV 1. When applying Suntail filtering, no SSS are retrieved along the vertical Suntail. Also, SSS near the vertical tail are quite different to those retrieved in other parts of the swath. 2. Nothing strange can be seen in SSS along the vertical tail when no Suntail filtering is applied.

  17. Comparison of monthly SMOS SSS with ISAS (May 2011) SSSnofilt-SSSISAS, mean=-0.03, std=0.65 SSSwithfilt-SSSISAS, mean=-0.06, std=0.72 SSSnofilt-SSSwithfilt , mean=0.03, std=0.36 1. When applying Suntail filtering, SSS near the vertical tail are quite different to those in other parts of the swath 2. Nothing strange can be seen in SSS along the vertical tail when no Suntail filtering is applied.

  18. One descending orbits in November 2012 (eclipse period) SSS of one orbit (20121118) SSSwithfilt-SSSISAS SSSnofilt-SSSISAS SSSwithfilt-SSSnofilt Suntails in FOV When applying L1c Suntail filtering, less dwell lines are filtered by the L2OS Chi2P flag => more SSS are retrieved in the northern hemisphere ... but they suffer from strong impact of Sun (low biases).

  19. SSSnofilt-SSSISAS, mean=-0.02, std=0.79 SSSwithfilt-SSSISAS, mean=-0.22, std=0.86 SSSnofilt-SSSwithfilt , mean=0.05, std=0.38 When applying Suntail filtering, more SSS are retrieved in the northern hemisphere (middle panel) than when the L1c Suntail flag is not filtered,  but they are systematically biased (more details in slide 21).

  20. Extra slides: In case Sun alias does not appear in the FOV(descending orbits in May 2011) The vertical Suntail still appears in the FOV even if the Sun alias is not in the FOV. There is few valid TB near the vertical tail (figure left). SSS near the tail are strange (middle panel in the maps below). SSSnofilt-SSSISAS, mean=0.03, std=0.65 SSSwithfilt-SSSISAS, mean=0.00, std=0.70 SSSnofilt-SSSwithfilt , mean=0.03, std=0.33

  21. Extra slides: In case Sun alias does not appear in the FOV(ascending orbits in November 2012) SSS near the tail are strange (middle panel in the maps below) when applying Suntail filtering. SSSnofilt-SSSISAS, mean=-0.18, std=0.72 SSSwithfilt-SSSISAS, mean=-0.14, std=0.76 SSSnofilt-SSSwithfilt , mean=-0.04, std=0.37

  22. Conclusion and discussion of Part 2 (SSS): should we use L1c suntail flag in L2OS 1. All ascending orbits and descending orbits during the non eclipse period. For ascending orbits in May with Sun alias in the bottom left of FOV, Sun contamination is weak. When applying Suntail filtering, SSS near the vertical tail appear noisy and biased. This is not observed when no Suntail filtering is applied. The sam e is observedfor the period when the Sun alias is outside EAFFOV (vertical tail still present in the FOV). We recommend not to apply the L1c Suntail flag in the L2OS

  23. Conclusion and discussion of Part 2 (SSS) 2. Descending orbits during the eclipse period: When applying L1c suntail flag, more SSS are retrieved in the northern hemishpere but they are strongly biased compared with ISAS SSS (-0.66), while SSSsmos-SSSisas of ascending orbits is -0.27. We recommend not to apply the Suntail flag in the L2OS descending orbits during the eclipse period because the L2OS Chi2P flag better removes biased retrieved salinities when no L1c suntail filtering is applied 0-40N: SSSasc,nofilt -SSSISAS, mean=-0.27, std=0.72 0-40N: SSSdesc,withfilt -SSSISAS, mean=-0.66, std=0.88

  24. L2OS threshold optimisation 20 June 2014, PM26 JL Vergely, J. Boutin, P. Spurgeon ACRI-ST,LOCEAN, ARGANS

  25. RFI/outlier detection Aim : • To improve the thresholds of the L2OS processor to be applied on TB measurements in order to remove outliers. About thresholds : • should be independent on L1c quality products

  26. Thresholds to be tested test for outlier detection (dwell test) nsig test for out of range TB detection (FOV test) Tm_out_of_range_affov Tm_out_of_range_eaffov Tm_out_of_range_stokes3_affov Tm_out_of_range_stokes3_eaffov Tm_out_of_range_stokes4_affov Tm_out_of_range_stokes4_eaffov test for oscillation TB detection (FOV test) Ts_std Ts_std_stokes3 Ts_std_stokes4 Other tests : max of iteration

  27. Tests conditions |X_swath| < 400 km (& sig_SSS<1.35) Coast > 1000 km -40° < lat < 40° SSS ref: ISAS near real time (MyOcean) PCT_var < 80 Day : 1,2,3,4,5/5/2013, L1C v550 L2OS proc : v600 (CATDS processing chain)

  28. Indicators / SSS quality filter • chi2P : good TB fit if chi2P high. Chi2P > 0.05 in current processor. • mean(SSS SMOS – SSS Coriolis) and std(SSS SMOS – SSS Coriolis) • X = (SSS SMOS – SSS Coriolis)/SSS_error. X should be close to a Gaussian law with mean(X)=0 and std(X)=1. Does not depend on SSS accuracy (close to the ratio between empirical error and theoretical error).

  29. Chi2P and RFI Percentage of RFI contamination : january 2012, asc Chi2P, 5/5/2013, asc

  30. nsig : outlier test at ‘nsig’ sigma Outlier detection. Dwell test. TB removed if : |TBsmos – TBmodel – DA| > nsig.rad_noise DA = mean dwell correction Current value : 5 Tested values : 2, 3, 4, 5

  31. nsig full ocean Good fit Bad fit

  32. nsig full ocean 4 sigmas test Queue distribution : outliers Expected distribution Centred reduced variable

  33. nsig full ocean Mean and std(X) nsig = 2 No Chi2P filter in L2OS Chi2p > 0.05

  34. nsig full ocean Mean and std(X) nsig = 3 No Chi2P filter in L2OS Chi2p > 0.05

  35. nsig full ocean Mean and std(X) nsig = 4 No Chi2P filter in L2OS Chi2p > 0.05

  36. nsig full ocean Mean and std(X) nsig = 5 No Chi2P filter in L2OS Chi2p > 0.05

  37. nsig full ocean nsig=2: Many outliers at 4 sigmas

  38. nsig : coast (1000km)

  39. nsig coast Expected distribution

  40. nsig coast Mean and std(X) nsig = 2 No specific filter Chi2p > 0.05 & sigSSS < 1.35

  41. nsig coast Mean and std(X) nsig = 3 No specific filter Chi2p > 0.05 & sigSSS < 1.35

  42. nsig coast Mean and std(X) nsig = 4 No specific filter Chi2p > 0.05 & sigSSS < 1.35

  43. nsig coast Mean and std(X) nsig = 5 No specific filter Chi2p > 0.05 & sigSSS < 1.35

  44. nsig coast nsig=2: Many outliers at 4 sigmas nsig=2 : very biased !!

  45. 5 day processing 1,2,3,4,5/05/2013; nsig = 2, 3, 4, 5 Dwell test. TB removed if : |TBsmos – TBmodel – DA| > nsig.noise DA = mean dwell correction Current config : nsig = 5

  46. nsig=2 Without LS mask. chi2P > 0.05

  47. nsig=3

  48. nsig=4

More Related