SMOS L2 Ocean Salinity Level 1 fixed cal reprocessing analysis (presentation by J. Tenerelli) – Drift (impact of TP7 correction) – Sun impact LO decimation study (TN/presentation by J. Font) Radiometric accuracy & sun flags (presentation by X. Yin) OTT drift and quality analysis OTT sensitivity study: – impact of galaxy signal, num. snapshots, time window (BEC PM#20) Outliers detection to identify RFI Upgrading roughness models – Rough 1: foam effect added, change in surface spectrum – Rough 2: going to an empirical roughness correction – Rough 3: improving LUT, going to fully empirical model Scientific status
SMOS L2 Ocean Salinity L2OS status 27 December January February 2011 L2OS 0316 enters its 9 th month of processing in DPGS...
SMOS L2 Ocean Salinity 317 history – delivered 22 November 2010; FAT passed 7 December 2010 – January installation delayed (L1OP not available) – new roughness model 3 LUT & OTTs for January calibration delivered early February – management meeting 9 February approves installation 317 – February IDEAS reject 317 due to AUX_BULL_B validity period issues (40+ s) – go-ahead for 317 to be installed in DPGS March... Issues: – L2OS ESL lose 2+ months pre-reprocessing DPGS data analysis – L2OS processor development/testing for summer delivery still using L1OP 34x no L1c 35x with correct RFI flags yet! 500 history? L2OS status
SMOS L2 Ocean Salinity L1 & L2 RFI flags L2 fm_l1c_rfi set if any of L1c flags set: – fml1c_rfi_mitigation – fml1c_rfi_strong – fml1c_rfi_point L2OS 0317 sets fm_valid = false if fm_l1c_rfi set Some orbits may have L1 RFI flags set over a wide area, eg SM_TEST_MIR_SCSF1C_ T080748_ T090119_350_001_0 So L2OS 0317 will find too few measurements for many grid points, leading to missing salinity retrievals
SMOS L2 Ocean Salinity Rain Rate ECMWF rain rate is metres/hour (10 3 litres per m 2 per hour), not mm/hr Max values are typically in storms. AUX_CNFOSD/F sets Tg_max_rainfall = 2, so Fg_sc_rain is never set ECMWF Rain_Rate Fg_sc_rain (Tg_max_rainfall = 0.002)
SMOS L2 Ocean Salinity Presentation by J. Tenerelli: land contamination Radiometric accuracy & sun flags (presentation by X. Yin) RFI detection (presentation by P. Spurgeon)
SMOS L2 Ocean Salinity Validation of current DPGS L2OS Using L3 operationally generated by CP34 – SMOS SSS (10 days L3 product) – WOA2005 SSS (climatology for February)
SMOS L2 Ocean Salinity – SMOS SSS (10 days L3 product) – Argo at -7.5 m (same period) Bias 0.26 psu STD 0.81 psu Still a lot to improve!
SMOS L2 Ocean Salinity Towards reprocessing L2OS upcoming improvements for (May/June) delivery (v500) – RFI detection/mitigation Fg_ctrl_suspect_RFI (no impact on schema), Dg_num_RFI_outliers (UDP schema change) switch RFI detection on/off, add threshold(s) to AUX_CNFOSD/F – roughness model & configuration tuning ICM empirical model without flat sea - analysis and algorithm design: CRR impact assessment N/A – fixed PRs: outlier detection ATBD/DPM mis-match – detected 1 February mis-use of L1 RFI flags – detected 27 January wrong ECMWF rain rate units & Tg_max_rainfall/Fg_sc_rain – detected 4 February BULL_B validity issues – detected 18 February On-going studies – errors in ECMWF WS (compared to SSMI) lead to bias in SSS – work continuing – investigate sun/galactic glint models (mismatch with SMOS data)
SMOS L2 Ocean Salinity ECMWF/SSMI WS: example of difference
SMOS L2 Ocean Salinity ECMWF/SSMI WS: Average SSSsmos-SSSargo in classes of 1m/s of WSecmwf-WSssmi WSecmwf-WSssmi (m/s)
SMOS L2 Ocean Salinity Time series & site statistics site monitoring: S. Pacific
SMOS L2 Ocean Salinity Site monitoring: S. Pacific (ascending)
SMOS L2 Ocean Salinity Site monitoring: S. Pacific (descending)
SMOS L2 Ocean Salinity Site monitoring: S. Pacific SSS1/2 correlation
SMOS L2 Ocean Salinity SMOS Model 1 - ARGO SSS (3-31 August 2010; asc orbits) versus wind speed (center of orbit)
SMOS L2 Ocean Salinity OTT/drift analysis Objective – determine OTT drift & required update frequency Analyse DPGS L2OS 316 salinity drift in stable South Pacific zone (0-30S) – retrieved salinity with sigma_SSS error – SSS anomaly (retrieved – climatology) Generate new OTT with 0317 for 0, +1, +2, +4, +7, +13, +28, +43 days – 02/08/2010, 03/08/2010, 04/08/2010, 06/08/2010, 09/08/2010, 15/08/2010, 30/08/2010, 14/09/2010 Compare DPGS L2OS 0316 SSS drift with new OTT drift Reprocess reference orbit 02/08/2010 with 0317 & new OTT – plot SSS & anomaly drift
SMOS L2 Ocean Salinity DPGS 0316 drift, Aug-Dec 2010 Selected South Pacific orbits, 0-30S, xswath ±300km, unfiltered, with sigma_SSS
SMOS L2 Ocean Salinity DPGS day drift Selected South Pacific orbits, SSS – climatology, 0-30S, xswath ±300km, unfiltered
SMOS L2 Ocean Salinity OTT drift OTT1 delta for selected South Pacific orbits 0-30S, 03/08/2010 to 29/11/2010
SMOS L2 Ocean Salinity Correlation between OTT & DPGS drift OTT delta versus mean SSS for selected South Pacific orbits 0-30S
SMOS L2 Ocean Salinity Reprocessing ref orbit with new OTT orbit reprocessed 0317 with OTT from later dates
SMOS L2 Ocean Salinity OTT/drift analysis discussion Purpose of OTT: – correct L1c TBs so L2OS can retrieve good quality salinity – secondary role: diagnostic for L1c quality Issues – Wait for L1OP 350 & 317 in DPGS to analyse OTT correction? – Wait for L1PP 500 & L2OS 500 June 2011 & analyse drift/OTT impact? – Perform analysis on fixed cal reprocessed commissioning data? Analysis strategies – Routine generation of OTT drift statistics? – UDP SSS anomalies or other metrics (eg time-series statistics, collocation match-ups) – Other? OTT strategies – One per day/week/month etc (eg mid-week – good for commissioning data) from a single orbit average of several orbits – eg one ascending & one descending abrupt transition between OTTs or smoothed – running average? – One size fits all – single OTT for entire reprocessing?
SMOS L2 Ocean Salinity Algorithm validation Once new L1 500 and L2OS 500 are implemented in DPGS, we will check: – Effectiveness of OTT strategy – Ascending/descending differences – Land contamination – RFI properly removed – Impact of sun and galactic signals – Performance of new roughness models. Comparison of SMOS to Argo and climatology Timetable?
SMOS L2 Ocean Salinity Decisions proposed to QWG #4 Increasing LO calibration frequency RFI outlier detection at L2OS Removing sun correction at L1 Extend L1 SUN_POINT flag to a circle around sun Flagging suspenders at L1
SMOS L2 Ocean Salinity 38 Remembering SSS requirements R The accuracy of the ocean salinity data products with a 50 km spatial resolution shall be <= 1.2 PSU. Note: This OS Level-2 accuracy requirement corresponds to : · snapshot images (100) of one pixel taken during one single pass, · usage of the narrow swath, · full spatial coverage, · ocean temperature of 277 K G The accuracy of the ocean salinity maps outside coastal areas, after averaging over 200 km x 200 km x 10 days space-time domain, shall be <= 0.1 PSU, after corrections by using vicarious calibration, namely by periodical imaging of one or more well-known reference sources.
SMOS L2 Ocean Salinity 39 On statistics During 10 minutes 500 SMOS snapshots are acquired Calibrating LO every 10 min means loosing 5 snapshots, then the error in building an L3 map by averaging L2 products (wrt no calibration) increases by sqrt (500)/sqrt(495) = Calibrating every 2 min, 25 snapshots are lost, the L3 error increases by sqrt(500)/sqrt(475) = Going back to one orbit, the increase of usable min compared min implies an error reduction of sqrt(495)/ sqrt(475) = 1.02 This 2% reduction on the required SSS error means 0.02 psu We showed that min instead min implies an error reduction by 0.2 psu (clean Pacific), 0.5 psu in global average