1. 1 Met Office, United Kingdom AATSR Meteo product: global SST validation at the Met Office Lisa Horrocks Jim Watts, Roger Saunders, Anne O’Carroll Envisat Validation Workshop 9-13 December 2002

2. Envisat Validation Workshop / 2 Outline Met Office contribution to AATSR validation Data chain and Meteo product availability Preliminary results: –Meteo product SSTs –Validation against buoys –Validation against MOHSST –Validation against HadISST Conclusions, plans and recommendations

3. Envisat Validation Workshop / 3 Met Office commitment to AATSR validation Validation of the AATSR Meteo product SST against in situ observations and analyses –Gross checks on instrument performance in NRT Validation activities: –comparisons against buoy SST (daily/weekly) –comparisons against MOHSST (each month) »5° gridded in situ –comparisons against HadISST (each month) »1° globally-complete monthly analysis Web pages showing routine validation results

4. Envisat Validation Workshop / 4 AATSR monitoring website Password required Updated daily at 0900 Monitoring plots provided two days behind time Buoy matchups updated weekly Monthly summary plots http://www.metoffice.com/research/nwp/satellite/infrared/aatsr/index.html

5. Envisat Validation Workshop / 5 Skin – bulk SST differences AATSR is sensitive to radiative skin temperature Validation data are measures of bulk SST (>1 m) Expect skin – bulk differences arising from: –skin effect –diurnal thermocline Skin always cooler than “sub-skin” by >0.1 K –conduction through molecular surface layer –at night, sub-skin = bulk Sub-skin can get warmer than bulk by several K –strong insolation, low wind => thermal stratification

6. Envisat Validation Workshop / 6 ground station Level 1b processing Level 2 processing: –Cloud detection –Spatial averaging –Retrieval of skin SST Meteo product extraction Second retrieval of skin SST Skin effect model Quality control Diurnal thermocline model Intercompariso n of SSTs ESA Met Office Data chain BUFR encodin g FTP serve r Buoy matchup AATSR data averaging

7. Envisat Validation Workshop / 7 Near-real time data availability Routine service from 19 August 2002 ~8–10 orbits per day (Kiruna ftp server) ~4 blind orbits per day (until Svalbard scenario, end-Nov) Some data gaps Require better communication of data supply changes or problems for operational service

8. Envisat Validation Workshop / 8 AATSR Meteo Product SST

9. Envisat Validation Workshop / 9 AATSR Meteo product coverage Kiruna-only scenario: Atlantic “blind” at night

10. Envisat Validation Workshop / 10 Dual-view algorithm comparison n = 38338 20 October 2002 (night) dual-3 minus dual-2 SST Night time brightness temperature data Retrieve SST using 3 channels and 2 channels Results show dual-2 ~0.2K cooler than dual-3 Implications for day-night biases in Meteo SST Similar discrepancy seen with ATSR-2 “Correction” can be determined D3 – D2 / K frequency

11. Envisat Validation Workshop / 11 Validation against buoy SSTs Quality control of buoy data –weekly buoy - NWP background test –gross check of each reported SST against climatology Matchup criteria: –Collocated to within 10 arc minutes »(BUT caveat L2 AST lat/lon calculation error) –Coincident within ± 3 hours Weekly mean and SD of AATSR-buoy differences Monthly subset of matchup data uploaded to NILU

12. Envisat Validation Workshop / 12 Meteo SST - buoy SST coverage so far =1860 weekly coverage ~130 Means for period 19 Aug - 27 Nov: all data = 0.03 K (s.d. 0.53 K) night only = 0.02 K (s.d. 0.42 K) day only = 0.03 K (s.d. 0.60 K) Moored buoys Drifting buoys

13. Envisat Validation Workshop / 13 Skin effect using buoy SSTs 19 Aug to 27 Nov 764 night time matchups AATSR SST minus buoy SST vs wind speed expect skin-buoy negative delta T compare Fairall skin effect model Dual-view 3-channel SST slightly too warm ?

14. Envisat Validation Workshop / 14 12 September “anomaly” Daily mean Meteo – buoy SST = – 0.68 K (1.90) Anomaly traced to 2 out of (only) 10 matchups Located off coast of San Francisco Buoy SSTMeteo SSTNWP SST 286.40282.22286.77 287.20282.96290.06 “Truth” confirmed by NWP background AATSR SST more than 4 K too cool Likelihood of undetected stratocumulus ?

15. Envisat Validation Workshop / 15 Validation against MOHSST Gridded in situ data at 5° resolution. October mean. Courtesy of N. Rayner, Hadley Centre for Climate Prediction and Research AATSR bulk SST minus MOHSST / K

16. Envisat Validation Workshop / 16 Validation against HadISST Globally complete SST analysis at 1° resolution. October mean.

17. Envisat Validation Workshop / 17 Validation against HadISST Skin - HadISSTBulk - HadISST Sep (1-30)-0.07 (0.79)0.08 (0.79) Oct (1-31)-0.08 (0.77)0.06 (0.77) HadISST available 10 days after end of month –September and October completed –global statistics (K): Regional statistics also computed –not yet enough data to spot patterns

18. Envisat Validation Workshop / 18 Conclusions From preliminary results, the Meteo product validates well against buoys and climate data: SSTs close to, or within, expectation Interalgorithm differences ~0.2 K (as ATSR-2) –Dual-view 3-channel SSTs may be slightly warm Evidence for undetected cloud in difficult areas Longer validation record required to detect regional or seasonal trends Larger buoy matchup dataset required for detailed investigation

19. Envisat Validation Workshop / 19 Future work and Recommendations Plans for continued and expanded validation –continue routine monitoring and comparison in NRT –AATSR/ATSR-2 –AATSR/microwave SSTs/AIRS Recommend –more timely and more detailed communication from PDS on data supply issues (changes or gaps) –product format amendments »BUFR version to include number of pixels (in Meteo) –work to correct dual-view interalgorithm bias –forum for discussion of ~1 year results