2. Climate Stability and Instability: Transition from Flywheel to Driver? Jochem Marotzke School of Ocean and Earth Science Southampton Oceanography Centre Southampton, SO14 3ZH United Kingdom

4. NOAA Global SST Analysis, 4 - 9 November 2002

5. North Atlantic warmer than North Pacific  NADW formation not a simple forced response to stronger cooling by atmosphere: If it were, NA should be colder than NP. Ocean circulation active in setting fundamental properties High North Atlantic sea surface salinity (SSS) crucial for NADW formation Ocean circulation can, in principle, maintain NA SSS greater than NP SSS without bias in forcing such as Atlantic-to-Pacific atmospheric water vapour transport (Marotzke & Willebrand, 1991). True in reality? - “without bias in forcing”? Coupled GCMs give equivocal answers (e.g., Manabe & Stouffer, 1999).

6. Is there another circulation mode that the MOC could attain?

7. Could transitions to another mode be abrupt?

8. Discuss intricacies using the example of ocean mixing Conceptual, mostly steady-state; illustrated w/ simple GCMs Flywheel or Driver? Is there another circulation mode that the MOC could attain? Confirmation requires continuous MOC observations How can this be done? Could transitions to another mode be abrupt? Would an MOC transition be a passive response to external forcing, or be self-driven, possibly following a trigger?

9. Mixing in Stratified Waters (I): Sandström (1908, 1916; see Colin de Verdière 1993): Heating below cooling is required so that fluid can act as a heat engine (buoyancy-driven flow exists) Jeffreys (1926): Expansion below contraction is crucial, which is possible in presence of mixing even if heating & cooling occur at the same pressure Munk (1966): Mixing heats upwelling deepwater Weyl (1968): Mixing converts turbulent kinetic energy into potential energy, which is needed to drive flow Munk and Wunsch (1998): Energy for mixing derives significantly both from tides and from wind

10. Mixing in Stratified Waters (II): GCMs with fixed diffusivity: MOC increases with density gradient (e.g., Scott, thesis 2000) With fixed amount of energy available for mixing, MOC might decrease with density gradient (Walin 1990, Lyle 1997, Huang 1998, Nilsson & Walin 2001, Oliver, thesis in prep.) Series of GCM experiments: Nilsson & Walin (submitted): Mixing and MOC: Flywheel or Driver - Meaningless question?

11. Expect mixing to matter mainly over very long timescales Time-dependent situations? Kevin Oliver (UEA, thesis in prep.): Considers transient behaviour in isopycnic box model with energy-dependent mixing (Nilsson & Walin, 2001)

12. Oliver (Thesis, UEA, in prep.)

13. F F increased from 0.3 to 0.4 Sv F F decreased from 0.4 to 0.3 Sv

14. Wang et al. 1999, idealised global model: “NADW” collapses under doubling of FW forcing within 1000 years NB: Collapse timescale unpredictable within factor 2 BUT: Steady-state: NADW increases with FW forcing NADW consistent with Rooth (1982) box model Total nearly constant

15. Convective mixing & sinking are different processes: Mauritzen (1996): DSOW derives from gradually sinking Atlantic Water, not convection in central Greenland Sea gyre Marotzke & Scott (1999): Sinking possible without convective mixing; sinking expected near boundaries Spall & Pickart (2001): Convective mixing & sinking co-located near sloping topography

16. If convective mixing is unimportant, why do we pay so much attention to its fate in the North Atlantic?

17. If high-latitude salinity is so important in the North Atlantic, why is the freshwater part of the surface buoyancy flux so small? Schmitt et al., 1989

18. Large & Nurser, 2001 Blue: Ocean heat loss Red: Ocean water gain Red: Ocean density gain

19. Pole-to-equator (and top-to-bottom) density contrast is dominated by temperature: The pycnocline is a thermocline

20. Water is dense because it is cold (from high latitudes) Which high latitudes ventilate deep ocean depends on SSS Density contrasts between high latitudes (competing DW formation sites) much smaller than between pole & equator Cross-equatorial coupling between high latitudes crucial Cooling dominates buoyancy flux in DW formation region Interhemispheric (& interocean?) dynamics central

21. Tziperman 1997 Wang et al. 1999 Klinger & Marotzke 1999

22. Convective mixing determines dominant high latitudes but not global deepwater formation rate Interhemispheric (& interocean?) dynamics central Diapycnal mixing works on overall density contrast Controls global rate of upwelling deepwater Efficiency of convective mixing unimportant for global rate Distribution over competing high latitudes depends on surface density, hence SSS High latitudes with deepest convective mixing dominate (Needs to be qualified: Topography, overflows etc.)

23. Convective mixing determines dominant high latitudes but not global deepwater formation rate Cooling dominates buoyancy flux in DW formation region Interhemispheric (& interocean?) dynamics central Summary Part I: Mixing and MOC: Flywheel or Driver - Meaningless question? Timescales critical in dependence on mixing and FW forcing Oceanic and atmospheric processes linked inextricably

24. Confirmation (of hypotheses of what controls MOC and its variability) requires continuous MOC observations as a starting point How can this be done?

25. 26.5°N MOC Monitoring Proposal PIs: Jochem Marotzke, Stuart Cunningham, Harry Bryden (SOC) Submitted to NERC RAPID Programme (which is funded with £20M over 6 years) Requested: £4.7M over 5 years Would support 2 Post-docs, 1 Research Assistant, 1 Ph.D. Student Funding decision expected 25/26 November

26. Why 26.5°N? Near Atlantic heat transport maximum - captures total heat transport convergence into North Atlantic South of area of intense heat loss ocean  atmosphere over Gulf Stream extension MOC dominates heat transport at 26.5°N Heat transport variability dominated by velocity fluctuations (Jayne & Marotzke, 2001) Florida Strait transport monitored for >20 years (now: Johns, Baringer & Beal, Miami, collaborators) 4 modern hydrographic occupations

27. Approach: Integrated thermal wind (geostrophy) Ekman contribution to MOC included Surface layer Ekman transport assumed to return independent of depth

30. Model-based experiment design: Funded through NERC prior to conception of RAPID Joël Hirschi (post-doc), Johanna Baehr (M.Sc. student) “Deploy” antenna in high-resolution models, OCCAM (1/4°; SOC, Webb et al.; Hirschi), FLAME (1/3°; IfM Kiel, Böning et al.; Baehr ) See Hirschi et al. poster

32. Blue: Covered Red: MOC Blue: Recon- struction

33. Red: MOC Blue: Reconstruction Black: OCCAM Heat Transport Green: Reconstruction OCCAMFLAME

34. Red: MOC Blue: Reconstruction Cyan: 300 realisations with random error (1 Sv Florida Strait; 0.01 kgm -3 ) OCCAM

35. Blue: Reconstruction Cyan: Thermal Wind Green: Ekman FLAME OCCAM

36. Transition from Flywheel to Driver: Importance of mixing in MOC dynamics Nature and location of mixing matter but are unknown (interior & boundary mixing; base of SO mixed layer; energetics) 1. What have we learned during the WOCE period? MOC could reorganise Dynamics of convection

37. Transition from Flywheel to Driver: DBE visualised inhomogeneity of mixing Deep Indian Ocean MOC: Well studied in WOCE projects (despite lack of WOCE 32S section); considerable deep mixing required to balance inflow. 2. What specifically was the WOCE contribution? Hydrographic sections gave accurate global estimate of MOC

38. Transition from Flywheel to Driver: Continuous observations of MOC drivers (heat & FW budgets of convection areas) Estimates of global distribution of mixing 3. What is required in the future (I)? Continuous observations of the MOC at selected latitudes

39. Transition from Flywheel to Driver: 3. What is required in the future (II)? Model-based experiment design for climate time series: Rational resource allocation Ocean (and coupled) models that represent coupled nature of mixing Improved (or development of) conceptual understanding of interaction between high latitudes (within and across oceans)