1. Examining the relationships between low-frequency upper ocean temperature and AMOC variability in ECCO v4 solutions Martha W. Buckley and Rui Ponte (AER) Thanks to the ECCO group, in particular Gael Forget and Patrick Heimbach

2. Low-frequency Atlantic SST variability Observations indicate Atlantic SSTs exhibit significant low-frequency variability (Bjerknes 1964; Kushnir 1994; Ting et al. 2009). Impacts of Atlantic SST variability include : –Temperature, precipitation over adjacent landmasses (Zhang and Delworth, 2006, 2007; Pohlman et al, 2006) –Changes in frequency/intensity of Atlantic hurricanes (Zhang and Delworth, 2006) However, the origin of SST anomalies is not understood. –Passive (local) response to atmospheric forcing (Seager, 2000). –Wind and/or buoyancy forced baroclinic Rossby waves (Sturges and Hong, 1995, 1998; Qiu 2002; Piecuch and Ponte, 2012). –Large scale changes in ocean heat transport due to changes in the AMOC (Kushnir 1994, etc.) and gyre circulations. –Lozier (2010): most significant question concerning the AMOC is role of AMOC in creating decadal SST anomalies.

3. Outline What is the relationship between interannual AMOC and upper-ocean heat content (UOHC) variability? I.ECCO version 4 state estimate A. Model details (G. Forget) B. Comparison to observations 1.Comparison of mean temperature to in-situ observations. 2.Comparison of SST variability to satellite SST observations. 3.Comparison of mean AMOC/OHT and AMOC and OHT variability to RAPID estimates. II.Interannual UOHC and AMOC variability A.UOHC variability in ECCO B.Associated AMOC variability C.Causality?

4. ECCO version 4 state estimate MITgcm least squares fit to observations, 1992-2010 For details, see G. Forget’s talk or Forget et al, in prep. new global grid LLC_90 –includes the Arctic –telescopic resolution to 1/3 o near the Equator –meridionally isotropic in mid-latitudes 50 vertical levels with partial cells forcing using ERA-Interim state-of-the-art dynamic/thermodynamic sea ice model nonlinear free surface + real freshwater flux B.C.s third-order advection scheme removal of C-D scheme for Coriolis terms use of diffusion operator (Weaver & Courtier, 2001) for in-situ obs. all satellite data are daily along-track internal model parameters are part of the control space

5. Comparison of ECCO & mean in-situ temperature G. Forget et al, in prep

6. Variability: comparison of ECCO & satellite SST SST variability: rms(modeled – observed) (G. Forget)

7. Realism of ECCO: AMOC and mean OHT Estimate of Atlantic OHT from ECMWF reanalysis: Trenberth and Caron (2001)

8. Realism of ECCO: comparison to RAPID Correlation 0.73 No systematic offset Correlation improves with time Correlation: 0.81 ECCO systematically underestimates OHT Note: Neither Florida current transport or RAPID AMOC transport estimates are used as constraints in ECCO

9. ECCO: Upper-ocean temperature (UOT) variability

10. UOT variability  AMOC variability AMOC variability UOT variability ?? Causality? AMOC  UOT OHT anomalies associated with AMOC variability lead to UOT variability UOT  AMOC UOT anomalies reach ocean boundaries and lead to AMOC variability UOT  AMOC UOT variability results in AMOC variability which feeds back onto UOT

11. Conclusions ECCO v4 estimate reasonably captures mean ocean state and large-scale ocean variability. –Atlantic OHT matches Trenberth & Carone (2001) well, although somewhat too low compared to direct ocean estimates. –Mean and variability of AMOC strength at 26 o N similar to RAPID –OHT variability at 26 o N captured; mean smaller than MOCHA estimate. ECCO exhibits low-frequency upper ocean temperature (UOT) variability: large anomalies over Gulf Stream path and in subpolar gyre. Large-scale AMOC variability associated with UOT variability. Causality of relationship to be determined. –UOHC budget analyses: role of air-sea heat fluxes, advection (separately consider Ekman transports, local response to winds).

12. Extra slides

13. Baroclinic pressure: modal decompostion Equations of motion linearized about a resting mean state, flat bottom Separation of variables  eigenvalue problem for vertical structure (Gill, 1982). Vertical structure for pressure: Solve for F(z) at each horizontal location using observed N(z). Modes = complete, orthonormal basis What portion of the variability is captured by 1 st baroclinic mode? Majority of baroclinic pressure variability in upper ocean is explained by 1 st baroclinic mode