1. Andrea Kaiser-Weiss, Melbourne Joint GHRSST Workshop, 6 th March 2012 Experiences with SST profiles from near-surface Argo measurements A. Kaiser-Weiss

2. The near-surface measurements from Argo profiling floats provide a new data set that is highly valuable for independent validation of SST retrievals, models and analyses. Thus the GHRSST Science Team recommends that these data are reserved for independent validation. Near-surface Argo data (both pumped and unpumped) are available via the British Oceanographic Data Center: ftp://ftp.pol.ac.uk/pub/bodc/argo/NST/BODC/. At this stage the accuracy of the unpumped Argo measurements has not been verified. Andrea Kaiser-Weiss, Melbourne Joint GHRSST workshop, 6 th March 2012

3. GHRSST requirements: AST: Iridium floats typically record at 2 dbar vertical resolution; given the response of the T sensor (~½ s) it’s possible to sample every second (~0.1 dbar) but with the resolution of the P sensor (0.1 dbar) it is probably not worth sampling faster than every 2 s (~0.2 dbar spacing). Hence with Iridium floats it would be possible to meet the GHRSST ‘goal’ of 50 cm sampling from 5 to 15m depth and the ‘useful’ sampling of 50 cm for less than 5 m depth, but unlikely to be possible to achieve the 10 cm ‘goal’ accuracy above 5m. If depth less than 5 dbar : 12 non-pumped samples at intervals of every 6 seconds (~0.6 dbar) The additional data can be transmittted with 2 extra Argo messages. a vertical resolution of 10 cm from 0 to 3 m and 50 cm resolution from 3 m to 10 m Andrea Kaiser-Weiss, Melbourne Joint GHRSST workshop, 6 th March 2012

4. NST measurement cycle Justin Buck @ BODC Andrea Kaiser-Weiss, Melbourne Joint GHRSST workshop, 6 th March 2012 CDT pumps are turned off to avoid contamination of salinity sensor. NST continue sampling temperature at 0.6 dbar intervals

5. Main uncertainty: determining pressure (vertical co-ordinate) Temperature sensor is 0.2 dbar below pressure sensor Surface pressure not easy to determine (water in inlet) NST data used with current cycle surface pressure (and not the previous cycle) No worries about pressure sensor drift -> corrected by difference surface pressure

6. Good and bad NST data points Justin Buck | juck@bodc.ac.uk | Data Scientist @ BODC Safe approach: discard all data where Δp < 0.6 dbar, take this as “surface pressure” only know to +- 60 cm where we sample.

7. Which data can be interpreted? Safe approach: discard all data where Δp < 0.6 dbar to make sure last reading is in water, take this value as “surface pressure” Thus we might not sample the first 60 cm, and only know to +- 60 cm the depth we do sample.

8. Can we interpret, understand and model the NST profiles? ΔP (dbar) 6 4 2 0 24 2526 T (°C) 8 10 6 4 2 0 8 6 4 2 0 8 6 4 2 0 8 25.5 26 19 20 26 26.5

9. Which Argo NST Quality control? The automatic quality control of near surface data has not been defined yet and the existing automated quality control thresholds for the profiles data may need refining. 6 4 2 0 8 10 Transmission errors? Short circuits? Energy flu? ΔP (dbar) T (°C) 19 18

10. Quality control of NST data At 20 dbar, 15 dbar, 10 dbar and 5 dbar cross- calibration samples are collected which consist of a non-pumped PT immediately followed by a pumped PTS. Jon Turton (UK Metoffice) showed encouraging agreement between pumped and non-pumped data, analysis to be repeated with a larger sample of data

11. Comparison of 5db Argo with drifters Figure provided by Jon Turton, UK Metoffice

12. Can the DV models reproduce what Argo NST observe?

13. What can we learn from: Argo DVArgo no DV Satellite DVQuantitative agreement, maybe because of larger than local DV event Local effects, warming in other places than Argo sampled Satellite no DV Local effects, warming probably on smaller scale than satellite can detect Quantitative agreement, due to local conditions similar to mean over footprint Shape of the profile Probablity distribution Max T vs latitude, local time of day, season, wind speed, cloud cover … Regional effects

14. What can we ask NST data to get easier to use Timing of ascent ? According to max expected DV, according to satellite overpass? Uncertainty characterization, options to determine surface Quality control/spike supression Data provision (now at BODC ftp) Near surface temperature data from all five institutes are now converted and live on BODC