Atmospheric response to the North Atlantic Ocean variability on seasonal to decadal time scales

Abstract

The NCEP twentieth century reanalyis and a 500-year control simulation with the IPSL-CM5 climate model are used to assess the influence of ocean-atmosphere coupling in the North Atlantic region at seasonal to decadal time scales. At the seasonal scale, the air-sea interaction patterns are similar in the model and observations. In both, a statistically significant summer sea surface temperature (SST) anomaly with a horseshoe shape leads an atmospheric signal that resembles the North Atlantic Oscillation (NAO) during the winter. The air-sea interactions in the model thus seem realistic, although the amplitude of the atmospheric signal is half that observed, and it is detected throughout the cold season, while it is significant only in late fall and early winter in the observations. In both model and observations, the North Atlantic horseshoe SST anomaly pattern is in part generated by the spring and summer internal atmospheric variability. In the model, the influence of the ocean dynamics can be assessed and is found to contribute to the SST anomaly, in particular at the decadal scale. Indeed, the North Atlantic SST anomalies that follow an intensification of the Atlantic meridional overturning circulation (AMOC) by about 9 years, or an intensification of a clockwise intergyre gyre in the Atlantic Ocean by 6 years, resemble the horseshoe pattern, and are also similar to the model Atlantic Multidecadal Oscillation (AMO). As the AMOC is shown to have a significant impact on the winter NAO, most strongly when it leads by 9 years, the decadal interactions in the model are consistent with the seasonal analysis. In the observations, there is also a strong correlation between the AMO and the SST horseshoe pattern that influences the NAO. The analogy with the coupled model suggests that the natural variability of the AMOC and the gyre circulation might influence the climate of the North Atlantic region at the decadal scale.

Appendix: The removal of ENSO from North Atlantic SST and geopotential height

The variability of ENSO is assessed using the first two PCs of the SST in the equatorial Pacific Ocean (12.5°S–12.5°N, 100°E–80°W). The first (second) modes represent between 62 and 66 % (7 and 11 %) of the total variance depending on the season. The ENSO variability is rather low in IPSL-CM5, with an incorrect phase locking of the ENSO variability to the annual cycle, even if its frequency spectrum is relatively similar to that of the observations.

Here, the relations between ENSO and the North Atlantic variability are assessed using simultaneous regressions of the SST (K) and Z500 (m) onto the first normalized PC of the Equatorial Pacific SST. In observations (Fig. 14), the main ENSO effect is to shift the subtropical jets equatorwards during El Niño phase, thereby inducing the same SST anomalies as a negative NAO in the Atlantic subtropical domain (Seager et al. 2003), while the subpolar North Atlantic SST anomalies are much weaker and not significant. In IPSL-CM5 (Fig. 15), the subtropical SST anomalies in response to ENSO are weaker, which is consistent with an underestimation of the SST variability off the coast of Africa. A positive phase of ENSO (El Niño) also warms the subpolar gyre, and the overall SST pattern is similar to the SST tripole associated to a negative NAO. This is consistent with the AMO (see Fig. 5, lower panels), that shows a link between the North Atlantic subpolar region and the equatorial Pacific Ocean in IPSL-CM5, but not for HadISST-LIM. This might be related to the incorrect phase locking of ENSO, which was demonstrated to alter the teleconnection with the Equatorial Pacific.

Fig. 14

Regression of the JFM (left panel) SST, in K, and (right panel) Z500, in m, onto the normalized ENSO index, in NCEP twentieth century and HadISST-LIM. The ENSO index is the PC1 of the JFM SST in the Equatorial Pacific. The color shades are suppressed when the significance, given by Monte Carlo analysis, is below 5 %

Fig. 15

Same as Fig. 14, but for IPSL-CM5 during JFM

The winter Z500 related to ENSO SST anomalies has strong anomalies over North America, as the Pacific-North American pattern is strongly modulated by ENSO. Over the North Atlantic the Z500 anomalies are roughly similar to the NAO, an El Niño phase causing a negative phase of the NAO in both model and observation as in Alexander and Scott (2002). The anomalies are also similar to those of OrtizBeviá et al. (2010), but the strong non-linearity of the ENSO teleconnections is neglected in this study.

In both the model and observations, ENSO has an impact onto SST and Z500 similar to the local influence of SST anomalies (compare Figs. 14 and 15 with Figs. 3 and 4, for lags larger than 3–4 months). Unlike in the observations of Frankignoul and Kestenare (2005), the removal of ENSO turned out to reduce the statistical significance of squared covariance and correlation of the first MCA mode when the ocean leads compared to other studies Czaja and Frankignoul (1999); Czaja and Frankignoul (2002), where ENSO was not removed from the SST and Z500. For example, when ENSO is not removed in observations, the squared covariance of the first MCA mode is 5 % significant up to lag 6 months when the ocean leads for Z500 in FMA, while the statistical significance is similar for Z500 in NDJ.