1. Observed influence of North Pacific SST anomalies on the atmospheric circulation Claude Frankignoul and Nathalie Sennéchael LOCEAN/IPSL, Université Pierre et Marie Curie, Paris, France SST anomaly

2. Air-sea interactions in the North Pacific SST Early attempts: cause and effect were confused in the observations (e.g. Namias 1963), as the main interaction is the forcing of the ocean by the atmosphere (Davis 1976, Frankignoul and Hasselmann 1977)

3. Air-sea interactions in the North Pacific SST Early attempts: cause and effect were confused in the observations (e.g. Namias 1963), as the main interaction is the forcing of the ocean by the atmosphere (Davis 1976, Frankignoul and Hasselmann 1977) Lagged correlation between the atmospheric forcing and SST in the passive case » SST leads SST follows

4. Air-sea interactions in the North Pacific SST Early attempts: cause and effect were confused in the observations (e.g. Namias 1963), as the main interaction is the forcing of the ocean by the atmosphere (Davis 1976, Frankignoul and Hasselmann 1977) Lagged correlation between the atmospheric forcing and SST in the passive case » SST leads SST follows Negative feedback

5. Davis (1978) Namias (1976): the strength of the Aleutian low in fall is related to N.Pacific SST anomalies in the antecedent summer Davis (1978): it can also be predicted by the summer SLP

6. Davis (1978) Namias (1976) showed that the strength of the Aleutian low in fall is related to North Pacific SST anomalies in the antecedent summer Davis (1978) found that it could also be predicted by the summer SLP The fall and winter SLP signal of Namias (1976) was related to previous SST in the equatorial Pacific (Barnett 1981)

7. In the late 90s, AGCM sensitivity studies showed a weak but significant response to North Pacific SST anomalies: function of the background climatology and the interaction of the baroclinic response to low-level heating and synoptic eddies (Peng et al. 1997) the equivalent-barotropic eddy-forced component resembles patterns of natural variability (Peng and Robinson 2001) North Atlantic SST anomalies have a small influence on the NAO (Czaja and Frankignoul 1999) the atmospheric response can be coarsely decomposed into a baroclinic response to diabatic heating and an equivalent barotropic mode forced by transient eddy fluxes (Deser et al. 2004)

8. In the late 90s, AGCM sensitivity studies showed a weak but significant response to North Pacific SST anomalies: function of the background climatology and the interaction of the baroclinic response to low-level heating and synoptic eddies (Peng et al. 1997) the equivalent-barotropic eddy-forced component resembles patterns of natural variability (Peng and Robinson 2001) North Atlantic SST anomalies have a small influence on the NAO (Czaja and Frankignoul 1999) the atmospheric response can be coarsely decomposed into a baroclinic response to diabatic heating and an equivalent barotropic mode forced by transient eddy fluxes (Deser et al. 2004) No observational evidence of a large-scale atmospheric response to extratropical SST anomalies in the North Pacific Nonaka and Xie (2003), Chelton, and collaborators: observed modulation of the atmospheric boundary layer by mesoscale eddies and fronts.

9. Latif and Barnett (1994): the atmosphere in their coupled model responds in such a way that it enhances the SST anomaly via surface heat flux but damps it after a delay via the gyre response to the wind stress curl, yielding decadal oscillations 500 mb geopotential height Wind stress curl AGCM response

10. Solomon et al. (2003): atmospheric response to extratropical SST anomalies is key to decadal variability of the North Pacific subtropical cell and thus ENSO in a simple ocean model coupled to a statistical atmosphere

11. Detecting the influence of the ocean on the atmosphere Two time scales Not valid for strongly coupled equatorial modes like ENSO Consider 1-dim. case linearized SST influence R TZ (  ) = R Tz (  ) + a R TT (  ) R TZ (  ) =  R TT (  ) for  << -  atmosphere

12. Detecting the influence of the ocean on the atmosphere Two time scales Not valid for strongly coupled equatorial modes like ENSO Consider 1-dim. case linearized SST influence R TZ (  ) = R Tz (  ) + a R TT (  ) R TZ (  ) =  R TT (  ) for  << -  atmosphere ENSO teleconnections ENSO time behavior ≠ 0 even if  = 0 ENSO signal must be subtracted

13. Basic variables for the MCA: monthly anomalies in SST, SLP, geopotential height Z at 850,, 700, 500 and 250 mb If atmospheric variables are in NDJ, SST is in NDJ (lag 0), OND (lag -1), … Only MCA mode 1 has robust significance when the ocean leads Lagged Maximum Covariance Analysis (SVD) of the 1977-2004 NCEP reanalysis in the North Pacific

14. Percentage of removed variance (ENSO) Second order polynomial trend and seasonally varying, asymmetric ENSO signal removed

15. Significance level: O 1% o 5% o 10% 20% SST lags SST leads no lag

16. Significance level: O 1% o 5% o 10% 20%

17. Summer signal (SST, Z250 in ASO) Contour interval 7 m

18. ASO AMJ C=0.58

19. ASO

21. HF is positive upward

23. Winter signal (SST, Z250 in NDJ) Contour interval 7m

27. Same for heat flux, positive upward SST projected forward to NDJ

28. Based on lag-2

30. Cross-validation results ASO: 29% of the variance of the atmospheric signal is predictable 4 months in advance OND and NDJ: 19% of the PNA-like signal is predictable 2 months in advance

31. Conclusions North Pacific SST anomalies have a significant hemispheric influence on the atmosphere in two seasons, independently from ENSO Summer signal: linked to SST anomalies in the Kuroshio extension region and in the eastern N. Pacific. It is mostly equivalent barotropic (20 - 35 m K -1 at 250 mb); may affect western Europe. Late fall - early winter signal: PNA-like (35 - 70 m K -1 at 250 mb) linked to a quadripolar SST anomaly that propagates eastward, eventually resembling the SST anomaly forced by the PNA via an active air-sea coupling throughout the fall The winter signal modulates the strength of the subtropical and subpolar gyres as in Latif and Barnett (1994), but no support for the hypothesized feedback loop. The two signals might influence the North Pacific subtropical cell, but no support for Solomon et al. (2003)