Josh K. Willis Jet Propulsion Laboratory

1. Can In-Situ Floats and Satellite Altimeters Detect the Impact of Global Warming on Atlantic Ocean Overturning? Josh K. Willis Jet Propulsion Laboratory Physical Oceanography This research powered by

2. Argo

3. Subsurface velocity level of known motion Profile data provide geostrophic shear Argo Floats Limited by 2000 m isobath

4. Profile data ~27,000 profiles excluding those from WHOI floats with pressure errors

5. Argo Trajectories Density is always computed first! Displacements from different depths are combined using vertical shear to estimate 1000 db velocity

6. Altimeter Data

7. AVISO SSH Data Snapshots of SSH from AVISO

8. Combining profile data with Altimetry Regression Coefficient between SSH and density 500 m

9. Float Displacements and Altimetry “Pseudo-displacements” computed from AVISO Regression btwn SSH and displacements

10. 2004 thru 2006 Time Mean • Willis and Fu, JGR, 2008 • Combined Altimeter data with • Argo profiles => 3-D density • Argo displacements =>1000 db dynamic height • 3-D Geostrophic velocity using density & 1000 db dynamic height as reference level Willis and Fu, JGR, 2008

11. 2004 thru 2006 Time Mean Density Willis and Fu, JGR, 2008

12. 2004 thru 2006 Time Mean Circulation @ 1000 db Willis and Fu, JGR, 2008

13. 2004 thru 2006 Time Mean SSH Can we integrate, west to east? Geostrophic Velocity

14. Northward flow of surface water Difference dynamic height at 2000 m isobath NADW Return flow Boundary Current Separated 2004 thru 2006 Time Mean Steep Topography

15. Error ~ e12 + e22 Transport Errors No Argo data in the shallow regions

16. ECCO 2 – 18 km Run

17. Errors Due to Lack of Shallow Data 0 – 1000 m Northward Transport from 18-km ECCO run with 2000 m isobath mask At these lats, lack of deep data doesn’t cause big errors RMS error in 3-month ave: 1.1 Sv

18. Temporal Variability • Again altimeter & Argo data => (Altimeter * regression coeff.) provide initial guess • 3-month time averages of 3-D dynamic height • Error bars => both mapping error & 1.1 Sv error due to lack of shallow data Willis and Fu, JGR, 2008

19. Data distribution near 40° N 3-month periods June, 2004 June, 2002 June, 2008 June, 2006

20. Time Series at 41°N

21. Trend: ~ 2.4 Sv increase Time Series at 41°N Mean: 15.5 Sv RMS: 2.4 Sv

22. Summary • AMOC at 41°N ~ 16 Sv • 2.4 Sv RMS variability • ~3 Sv error bar • Variability Mostly Ekman • No significant trend, 2002 to present • SSH => 2.4 Sv increase over 16 years • Argo and Altimetry complement other types of AMOC observations

23. Other Stuff

24. Time Series of Density Maps June, 2002 June, 2004 0/1000 db steric height from density June, 2006

25. Time Series of 1000 db Velocity June, 2002 June, 2004 June, 2008 June, 2006

26. Time Series of Velocity at 39°N June, 2002 June, 2004 June, 2006 June, 2008

27. Mapping Density Covariance function for density Noise, length scale from fit

28. Mapping Displacements Covariance Functions Noise: .0013 (m/s)2 ~3.6 cm/s >> 0.2 cm/s

29. Float Displacements and Altimetry Regression Correlation

30. Computing subsurface displacements x x RMS error: 0.2 cm/s Park et al., JTECH, 2005

31. Computing Transport Errors 3-Year Mean Difference between independent estimates Same, normalized by (1-skill)

32. 0 by definition for time mean Variance reduction in density from using a x SSH Mapping Density ρestimate = { ρprofile – α × SSH } + α × SSH

33. Mapping Density Covariance function for density 2-scale covariance function: <ρiρj> = ½ (1 + r + 1/6 r2 – 1/6 r3) exp(–r) + ½ exp(-R2) + n δij, r = sqrt[ (xi – xj)2 / lx2 + (yi – yj)2 / ly2 ], R = sqrt[ (xi – xj)2 / Lx2 + (yi – yj)2 / Ly2], From Rio and Hernandez, JGR-Oceans, 2004

34. Mapping Displacements 2-scale covariance function: <pipj> = [½ (1 + r + 1/6 r2 – 1/6 r3) exp(–r) + ½ exp(-R2)] × exp(-Ω)+ n δij, Ω = (ζi – ζj)2 / (ζi2 + ζj2) <uu>, <uv>, and <vv> derived from: As in Lavender et al., DSR-I, 2008