SMOS L1v620-L2v613 versus L1v505-L2v550 validation May 2011

1. SMOS L1v620-L2v613 versus L1v505-L2v550 validationMay 2011 Nicolas Martin, Jacqueline Boutin LOCEAN 26 May 2014

2. All the metrics SMOS-ARGO and SMOS-ISAS, as well as a complete descriptions of flags, are available on:http://www.locean-ipsl.upmc.fr/~martin/work/L1Cv620_L2OPv613 Observations (at >800km from coast; stats for whole ocean including <800km from land available on web site): -Careful must be taken as L1 AND L2 versions are diffferent -SMOS-ARGO colocations in radii of 9days-100x100km2 are much noisier than SMOS-ISAS monthly comparisons and hence conclusions less clear => in this ppt, we show only SMOS-ISAS -Ascending orbits (next slide) : mean bias is reduced (possibly an effect of daily OTT in new L2) but a positive bias in the northern Pacific, we do not see what in L2 change could create this; likely to be a remaining drawback of L1; larger RFI pollution in Northern Atlantic in v6 than v5: L1 RFI flags not sufficient, RFI flags in L2 need to be optimized, -Descending orbits: improvement of north-south bias likely an effect of TEC retrieval (not an L1 effect) => not shown in this ppt as it does not concern L1 -Land effect: still there (<0 bias) + larger std in v6: possibly RFI sorting less efficient!

3. S[SMOS A]-S[ISAS] - L1Cv620_L2OPv613 Ascending orbits : mean bias is reduced (possibly an effect of daily OTT in new L2) but a positive bias in the northern Pacific, we do not see what in L2 change could create this; likely to be a remaining drawback of L1; larger RFI pollution in Northern Atlantic in v6 than v5: L1 RFI flags not sufficient, RFI flags in L2 need to be optimized (they have change from L2 v5 to L2v6), S[SMOS A]-S[ISAS] - L1Cv504_L2OPv550

4. Mean difference and std of the difference per region– red: v6; blue: v5 Ascending orbits Mean bias is reduced BUT STD is not improved (even slightly worse on ARGO comparisons see www page); only large scale bias improved

5. Effect of land vicinity (D<800km) : Ascending orbits; still <0; larger std, RFI?

6. Effect of land vicinity (D<800km) : Descending orbits; still <0; larger std, RFI?

7. Descending orbits results Be careful that TEC was retrieved in L2 v6 but not in L2 v5 => latitudinal biases are reduced due to TEC retrieval (see last slides of this presentation shown at QWG 13)

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

11. SMOS SSS - ISAS SSS: descending PCTVAR<80% (exclude regions with no ARGO) and 700km far from land L2V550 OTT A3TEC on - L1cTEC prior –TEC dwell retr. L2V550 OTT L1cTEC – L1cTECprior – TEC dwell retr. L2V611b, OTT L1cTEC – A3TEC ‘prior’ -no TEC dwell retr. L2V611b, OTT A3TEC on- A3TEC prior – no TEC dwell retr. ‘OTT A3TEC on- A3TEC prior – no TEC dwell retr.’ give best retrieved SSS compared with ISAS SSS.

12. SMOS SSS - ISAS SSS: descending Metrics L2V550 OTT A3TEC on - L1cTEC prior –TEC dwell retr. L2V550 OTT L1cTEC – L1cTECprior – TEC dwell retr. Global: mean=0.106 (std=0.389) [45S45N;180W180E] : mean=0.0535(std=0.320) [30S30N;180W180E] : mean=-0.006 (std=0.288) mean=-0.034 (std=0.408) mean=-0.084 (std=0.343) mean=-0.163 (std=0.306) L2V611b, OTT L1cTEC – A3TEC ‘prior’ -no TEC dwell retr. L2V611b, OTT A3TEC on- A3TEC prior – no TEC dwell retr. Global: mean=0.188 (std=0.442) [45S45N;180W180E] : mean=0.112 (std=0.362) [30S30N;180W180E] : mean=0.019 (std=0.319) mean=0.002(std=0.355) mean=-0.010 (std=0.286) mean=-0.016(std=0.270) For descending orbits, when ISAS-PCTVAR<80% and SMOS at 700km further from land, ‘OTT A3TEC on- A3TEC prior – no TEC dwell retr.’ give lowest differences (mean and std) compared with ISAS SSS over the global ocean.