1. The meridional coherence of the North Atlantic meridional overturning circulation Rory Bingham Proudman Oceanographic Laboratory

2. Introduction: The thermohaline circulation The global conveyor belt: Redistributes heat (approx. 50%) Helps maintain the relatively mild European climate Highly idealised

3. Introduction: The thermohaline catastrophe Air temperature from a HADCM3 model experiment. Change due to a THC shutdown. Greenland ice-sheet melting Freshening of the North Atlantic surface waters Inhibiting convection and deep water production THC shutdown Global warming may lead to the following scenario:

4. Introduction: The thermohaline catastrophe Hollywood style BUT, at least it puts the ocean into climate! Sea-level rises 30m in a matter of hours! A global ice-age within a week!

5. Introduction: Palaeo evidence 231Pa/230Th indicates much reduced MOC in at least the deglacial events, but evidence is sparse for earlier events. (Gherardi et al., EPSL, 2005) Differential settling rate for protactinium and thorium allows for the reconstruction of past MOC strength from sediment cores

6. Introduction: The RAPID-MOC array Objective: To provide a timeseries of absolute meridional transport as a function of depth Why 26N? Gulf Stream determined by cable voltage measurements Latitude of maximum heat trasnport Historical record on hydrographic surveys

7. Introduction: Early results from the 26.5N array Cunningham et al., Science, 2007

8. Introduction: Propagating signals Fig. 1. Shallow-water model with moving surface layer and infinitely deep, motionless, lower layer. Thermohaline overturning is represented by a prescribed outflow from the surface layer on the northern boundary. Model domain is a sector ocean 50° wide, extending from 45°S to 65°N Fig. 2. Surface layer thickness after a thermohaline overturning of 10 Sv is switched on at time t = 0 in the northwest corner of an ocean initially at rest. There is no wind forcing, and the surface layer is initially 500 m deep. The contour interval is 2 m, and thicknesses less than 499 m are shaded. Note that the thickness anomaly on the western boundary is much greater than that in the interior (the layer thickness is approximately 350 m in the northwest corner of the domain; see Fig. 3 for details), but extra contours are not plotted here. The southernmost 10° of the domain comprises the sponge region, where we might expect the dynamics to be somewhat unrealistic Changes in the deep water formation rate at high latitudes are rapid communicated to lower latitudes via coastally trapped waves (Kawase,1987; Johnson and Marshall, 2002)

9. Introduction: The WAVE arrays W-line B-line (Halifax) A-line (Grand Banks Objective: To detect propagating signals associated with changes in the THC

10. Introduction: The WAVE arrays

12. The most likely culprit for the loss of the WAVE moorings? Corrosion of an anchor link after one year in the water at 26N

13. Presentation Outline Statistical analysis on meridional transport coherence Local dynamics of meridional transport variability at a given latitude Dynamical origins of meridional differences OCCAM: 0.25 ° eddy permitting resolution 66 vertical levels ECMWF 6hrly forcing 5 day mean fields 1985-2003 period after spin up

14. Upper layer meridional transport variability OCCAM North Atlantic MOC streamfunction (1985-2003) This picture is suggestive of an MOC that varies as a coherent entity Must be the case at long enough timescales Short according to some theories of MOC adjustment (eg Johnson and Marshall 2003)

15. Depth integral of MT (100-1000m) Upper layer meridional transport variability OCCAM North Atlantic MOC streamfunction (1985-2003) Low freq. dominates High freq. dominates Poleward of approx. 40N a interannual mode is clearly visible. (Also reported by Marsh et al, 2005) To the south higher frequency variability more dominant Short lived meridionally coherent signals apparent Radon transform indicates south propagation at 1.8ms -1

16. How well does interannual MT variability at one latitude correlate with the variability at other latitudes? Statistical analysis: Cross correlation analysis 0-1000m 100-1000m 0-100m (Ekman) For the 0-1000m MT integral clear separation at 40N. Mutually correlated north and south of 40N. Due in part to meridional structure of zonal wind stress over NA Excluding Ekman transport improves overall correlation between latitudes north and south of 40N, but still low Suggests an underlying mode of interannual MT variability

17. Is there a coherent underlying mode of MT variability? Statistical analysis: Empirical Orthogonal Functions Dominant interannual mode is a single overturning cell More intense to the north of 40N where it accounts for most of variance Becomes weaker and accounts for less of the variance to the south Represent meridionally coherent MT fluctuations of 0.8Sv RMS 1 st mode (29%) 2 nd mode (11%) TF1: Red TF2: Blue Contour int. = 0.2SV

18. Meridional transport anomaly between 100m and 1000m depth OCCAMHadCM3 Model intercomparison: HadCM3 vs. OCCAM Contour int. = 0.2SV Leading EOF of interannual meridional transport variability

19. Story so far Models show a distinction in the nature of meridional transport variability north and south of 40N: –Low frequency variability dominates to the north –Higher frequency variability dominates to the south Underlying meridionally coherent mode of MOC variability Suggests caution when interpreting “MOC” measurements from one latitude. Worth monitoring transport north of 40N. But a 26N-type array not feasible Can the RAPID-WAVE array can be used to determine MT variability?

20. The meridional transport at 42N Absolute transport Anomalous transport (100m-1000m)

21. At depths below the Ekman layer, the zonally-integrated northward mass flux is given, using geostrophy, by Dynamics: Geostrophic balance p w high p w low inc. southward flow inc. northward flow Pressure relative to the eastern boundary

22. Dynamics: BP and SSH variability at 42N BP SSH

23. Dynamics: Low frequency BP and SSH variability at 42N bottom pressure sea level EB WB MAR 300m 1300m 3000m

24. At depths below the Ekman layer, the zonally-integrated anomalous northward mass flux is given, using geostrophy, by where the approximate equality holds for changes to the flow, on the assumption that the eastern boundary pressure varies much less than that at the west. Dynamics: Geostrophic balance p w high p w low inc. southward flow inc. northward flow ?

25. Dynamics: The geostrophic calculation at 42N Actual Residual (T * -T) Using boundary east-west pairs only Using all east- west pairs Using eastern boundary points only Using western boundary points only

26. Dynamics: The geostrophic calculation at 42N Using all east- west pairs Using boundary east-west pairs only Using western boundary points only Upper layer (100-1000m) Lower layer (1000-3000m)

27. Slope average pressure signal at 42N Slope average BP: East West MT (0-100m) -MT (100-bot) Depth av p e - p w

28. Dynamics: The geostrophic calculation at 42N Upper layer transport RMS error: 0.28Sv Lower layer transport RMS error: 0.31Sv Actual Inferred from western boundary pressure

29. Leading EOFs of interannual sea-surface height and bottom pressure BP EOF1 SSH BP SSH EOF1Strong association between leading bottom pressure and sea-level EOFs and low frequency MOC mode Pressure signal strongly constrained by bathymetry Consistent with geostrophic relationship Both account for most of variance on shelf and upper slope but little in the deeper ocean. Signal weakens to the south

30. Observed low frequency SSH variability at 42N Eastern boundary Western boundary Mid-Atlantic Ridge

31. Observational evidence: Leading EOFs of interannual sea-surface height AltimetryOCCAM Altimetry OCCAM

32. Story so far Interannual MT variability largely determined from pressure on the western boundary Basin-scale pressure signals mean that the slope average pressure must be removed – Ekman compensation transport lost Some supporting evidence from altimetry What drives w.b. pressure variability?

33. Depth integral of MT (100-1000m) Dynamical origins of meridional differences

34. Dynamics: The geostrophic calculation at 50N Upper layer (100-1000m) transport; RMS error: 0.39Sv Actual Inferred from western boundary pressure Lower layer (1000-3000m) transport; RMS error: 0.39Sv

35. x10 -4 Kgm -3 High-low density composite MOC dynamics at 50N Upper layer transport at 50N Low frequency mode has clearest expression at 50N -> examine dynamics at this latitude x10 -4 Kgm -3 Density profile Strong association with density on the western boundary. Increased density leads to increased MT. Negligible signal on eastern boundary.

36. MOC dynamics at 50N Western boundary density profile Anomalous bottom pressure (eq. cm) on the western boundary Density changes on the western boundary drive changes in bottom pressure Anomalous meridional transport Through geostrophy changes in the east-west pressure difference across that basin are associated with meridional transport variations

37. Origin of meridional differences: Evolution of boundary density Path following the 1000m isobath along the western slope

38. Origin of meridional differences: Evolution of boundary density P1 P2 P3 P1 P2 P3 Anomalous density along the 1000m isobath Advection Convection + advection + waves 50N 42N Advection + waves advection 0.9cms -1 wave: 1.8ms -1 Seasonal cooling events associated with NAO are integrated to give low frequency mode clear at 50N 50N signal advected to lower latitudes, and degraded along the way Timing follows LSW life-cycle described by Yashayaev et al, 2007

39. Summary Interannual MT transport variability determined by wb pressure Western boundary pressure signal results from results from cooling in the western subpolar gyre Low frequency density signal results form seasonal cooling events associated with strong NAO Density signal advected along the western boundary but diminishes in amplitude Corresponds with the weakening of the underlying MOC mode at lower latitudes and changing nature of the MT with latitude

40. WAVE BPRs from 27 Dec 2004 Time (hrs from midnight)

41. (E2) Model resolution: 1.4 degrees Forcing: winds and surface fluxes from ECMWF (E3) Model resolution: 0.23 degrees Forcing: monthly climatological winds and surface fluxes from ECMWF, repeating each year. (E1) Model resolution: 0.23 degrees Forcing: winds and surface fluxes from ECMWF Are the results robust to different model formulations and forcing scenarios? Statistical analysis: Isopycnal model experiments