The influence of sea surface temperature anomalies on low-frequency variability of the North Atlantic Oscillation

Abstract

The influence of sea surface temperature anomalies (SSTA) on multi-year persistence of the North Atlantic Oscillation (NAO) during the second half of the twentieth century is investigated using the Center for Ocean-Land-Atmosphere Studies (COLA) Atmospheric GCM (AGCM) with an emphasis on isolating the geographic location of the SSTA that produce this influence. The present study focuses on calculating the atmospheric response to the SSTA averaged over 1988-1995 (1961-1968) corresponding to the observed period of strong persistence of the positive (negative) phase of the decadal NAO. The model response to the global 1988-1995 average SSTA shows a statistically significant large-scale pattern characteristic of the positive phase of the NAO. Forcing with the global 1961-1968 average SSTA generates a NAO of the opposite polarity compared to observations. However, all large-scale features both in the model and observations during this period are weaker in magnitude and less significant compared to 1988-1995. Additional idealized experiments show that over the northern center of the NAO the non-linear component of the forced response appears to be quite important and acts to enhance the positive NAO signal. On the other hand, over the southern center where the model response is the strongest, it is also essentially linear. The 1988-1995 average SSTA restricted to the western tropical Pacific region produce a positive NAO remarkably similar in structure but stronger in magnitude than the model response to the global and tropical Indo-Pacific 1988-1995 forcing. A 200-hPa geopotential height response in these experiments shows a positive anomaly over the southern center of the NAO embedded in the Rossby wave trains propagating from the western tropical Pacific. Indian Ocean SSTA lead to much weaker positive NAO primarily through the effect on its northern center. SST forcing confined to the North Atlantic north of equator does not produce a response statistically different from the control simulation, suggesting that it is not strong enough to significantly affect the phase of the decadal NAO. Inclusion of the South Atlantic north of 45° south does not change this result.