The atmospheric variability over the North Atlantic sector is dominated, on interannual to decadal timescales, by the North Atlantic Oscillation (hereafter NAO). As shown in Fig. 1a, the NAO modulates the sea level pressure difference between Iceland and the Azores, and so is a measure of the strength of the surface atmospheric circulation over the North Atlantic. As the sea level pressure anomaly (hereafter SLP) shown in Fig. 1a is shifted compared to the climatology, the NAO characterizes not only a modulation in the strength of the surface circulation but also its latitudinal displacement. Years during which the atmosphere is in a high NAO phase have stronger storms, penetrating further to the North than normal, with a stronger and more poleward path of the Jet. Conversely, years during which the atmosphere is in a low NAO phase have a weaker and more zonal storm track / Jet system [1990]. The NAO exerts a dominant influence on temperatures, precipitation, storms, fisheries and eco-systems of the Atlantic sector and its surrounding continents (see the recent review by Marshall [2000b]). Understanding the NAO and its time dependence, therefore, appears to be of primary importance.
The time evolution of the NAO index, i.e. the normalized seal level pressure difference between Lisbon and Stykkisholmur averaged over northern hemisphere winter [1995], is shown in Fig. 1b. Superimposed on pronounced interannual fluctuations, one sees longer timescale components, like the downward trend from the 1920s to the 1960s, followed by a positive trend over the last 30 years. Superimposed on these are `decadal' fluctuations. Spectral analysis of the NAO index of Hurrell [1995] by Wunsch [1999] suggests that it has much of its energy at high frequencies, but with a broad band peak near the decadal period (although not statistically significant).
In the the ocean, reports of low-frequency variability in SST reveal preferred decadal timescales, both in observations (e.g. Deser [1993]; Dickson [1996]; Sutton [1997]; Curry [1998]; Tourre [1999] ), and coupled models (e.g. Grötzner [1998]; Timmermann [1998]; Selten [1999]; Christoph [2000] ). Because of its slower evolution and its capacity to store and transport heat, the ocean may be involved in the evolution of mid-latitude SST anomalies on long timescales (periods longer than a few years) and, in turn, influence the NAO, as was first suggested by Bjerknes [1964].
The canonical model of how the ocean
interacts with the midlatitude atmosphere is that of
Frankignoul [1977] (in the following [FH77])
is a damping timescale
(of order of a season) and N is the surface forcing
(primarily the surface turbulent heat flux) associated
with, in our context, the NAO. In [FH77]
N is taken to be a white noise, so that (1) predicts
a red spectrum for SST on timescales shorter than 
,
and a flattening on longer timescales. Obviously, the
model (1) omits any role of ocean circulation
in modulating SST in response to fluctuations in the NAO.
The goal of this lecture is, based on the analysis of observations, to develop a simple model that allows for an active role of the ocean circulation (wind driven and thermohaline) in shaping the NAO-related variability of SST in the Atlantic. Dominant patterns of air-sea fluxes are reviewed (section 2), and oceanic processes that may account for a coupled response are explored (section 3). A simple model comprising stochastic, wind-driven, and thermohaline processes is then formulated (section 4). Statistical signatures of the impact of ocean circulation on SST predicted by the model are established, and compared to a long observational record of SST and SLP (section 5). The possibility that SST may not only passively respond to the NAO, but also impact in return the NAO, is also discussed. Essential ingredients to this lecture summary are the work of J. Marshall, H. Johnson and J. Goodman [2000] and A. Czaja and J. Marshall [2000].
