What is the NAO and why is it important?
The climate of the Atlantic sector and surrounding continents exhibits considerable variability on a wide range of time scales. Improved understanding of this variability is essential to assess the likely range of future climate fluctuations and the extent to which these fluctuations are predictable, and to assess the potential impact of climate change due to anthropogenic forcing.
A substantial portion of the climate variability over the Atlantic basin is associated with the NAO, which is a dominant pattern of atmospheric circulation variability. The NAO refers to a meridional oscillation in atmospheric mass with centers of action near Iceland and over the subtropical Atlantic from the Azores across the Iberian Peninsula. When the NAO is in its positive phase, low pressure anomalies over the Icelandic region and throughout the Arctic combine with high-pressure anomalies across the subtropical Atlantic to produce stronger-than-average westerlies across middle latitudes. This phase of the oscillation is, consequently, associated with cold conditions over the northwest Atlantic and warm weather over Europe, as well as wet conditions from Iceland through Scandinavia and dry conditions over southern Europe. This pattern of climate anomalies, which has been recognized at least since Walker and Bliss (1932), is most pronounced during winter when atmospheric teleconnection patterns such as the NAO are strongest.
A remarkable feature of the NAO that has motivated much recent study is its trend toward a more positive phase over the past 30 years. In fact, the magnitude of this recent trend appears to be unprecedented in the observational record (Hurrell 1995a), and based on reconstructions using paleoclimate data, perhaps over the past several centuries as well (Stockton and Glueck 1999). The most pronounced anomalies have occurred since the winter of 1989 (Hurrell 1995a; Walsh et al. 1996; Thompson and Wallace 1998; Watanabe and Nitta 1999) when record positive values of an index of the NAO have been recorded. Moreover, the trend in the NAO accounts for several remarkable changes recently in the climate and weather over the middle and high latitudes of the Northern Hemisphere, as well as in marine and terrestrial ecosystems. Among these changes are:
_ Strengthened subpolar westerlies from the surface to the lower stratosphere (Thompson et al. 1999).
_ Milder winters in Europe downstream across Asia juxtaposed against more severe winters over eastern Canada and the northwest Atlantic (Hurrell 1995a; Wallace et al. 1995; Hurrell 1996; Shabbar et al. 1997; Thompson and Wallace 1998).
_ Pronounced regional changes in precipitation patterns (Hurrell 1995a; Hurrell and van Loon 1997; Dai et al. 1997) resulting in the advance of some northern European glaciers (Hagen 1995; Sigurdsson and Jonsson 1995) and the retreat of Alpine glaciers (Frank 1997).
_ Changes in sea-ice cover in both the Labrador and Greenland Seas as well as over the Arctic (Chapman and Walsh 1993; Maslanik et al.1996; Cavalieri et al. 1997; Parkinson et al. 1998; McPhee et al. 1998; Deser et al. 1999).
_ Pronounced decreases in mean sea level pressure (SLP) over the Arctic (Walsh et al. 1996).
_ Changes in the physical properties of Arctic sea water (Sy et al. 1997; Morison et al. 1998; McPhee et al. 1998; Dickson 1999; Dickson et al. 1999a,b).
_ Changes in the intensity of convection in the Labrador and the Greenland-Iceland Seas (Dickson et al. 1996; Houghton 1996) which in turn influence the strength and character of the Atlantic meridional overturning circulation.
_ Stratospheric cooling over the polar cap (Randel and Wu 1999), and total column ozone losses poleward of 40oN (Randel and Wu 1999; Thompson et al. 1999).
_ Changes in storm activity and the shifts in the Atlantic storm track (Hurrell 1995b), changes in within season variability such as blocking (Nakamura 1996).
_ Trend in North Atlantic surface wave heights (Kushnir et al. 1997).
_ Changes in the production of zooplankton and the distribution of fish (e.g., Fromentin and Planque 1996).
_ Changes in the length of the growing season over Europe (Post and Stenseth 1999), and changes in the population dynamical processes of several terrestrial species (Post et al. 1999; Stenseth et al. 1999).
All these appear to be strongly related to the recent trend in the NAO.
This unprecedented behavior of the NAO in recent decades and, more generally, its pronounced low-frequency behavior over the longer record have added to the debate over our ability to detect and distinguish between natural and anthropogenic climate change. Hurrell (1996) has shown, for example, that the recent upward trend in the NAO accounts for much of the observed regional surface warming over Europe and Asia, as well as the cooling over the northwest Atlantic over the past several decades. The NAO accounts for about one-third of the hemispheric interannual surface temperature variance over the past 60 winters. Since global average temperatures are dominated by temperature variability over the northern landmasses, a significant fraction of the recent warming trend in global surface temperatures can thus be explained as a response to observed changes in atmospheric circulation. In particular, changes over the North Atlantic are associated with the NAO (see also Graf et al. 1995; Thompson et al. 1999). Since the NAO is a natural mode of atmospheric variability, one could argue that much of the recent warming is not related to the build-up of greenhouse gases in the atmosphere over the past century. This viewpoint, however, ignores the possibility that anthropogenic climate change might influence modes of natural variability, perhaps making it more likely that one phase or another of the NAO be preferred over the other phase (Palmer 1999; Corti et al. 1999). Understanding the physical mechanisms that govern the NAO and its intraseasonal-to-interdecadal variability, and how modes of natural variability such as the NAO may be influenced by anthropogenic climate change remain, therefore, central research questions.
What are the mechanisms which govern NAO variability?
Although the NAO is a natural mode of variability of the atmosphere, surface, stratospheric or even anthropogenic processes may influence its phase and amplitude. At present there is no consensus on the process or processes that are responsible for observed low frequency variations in the NAO, including its unprecedented upward trend over past 30 years.
There is ample evidence which shows that much of the atmospheric circulation variability in the form of the NAO arises from internal atmospheric processes. Atmospheric general circulation models (AGCMs) forced with climatological annual cycles of solar insolation and sea surface temperature (SST), and fixed atmospheric trace-gas composition, display NAO-like fluctuations (e.g., Kitoh et al. 1996; Saravanan 1998; Osborn et al. 1999; Shindell et al. 1999). The governing dynamical mechanisms are eddy mean flow interaction at the exit region of the Atlantic storm track and eddy-eddy interaction between baroclinic transients and low-frequency variability (Wallace and Lau 1985; Lau and Nath, 1991; Ting and Lau, 1993; Hurrell, 1995a). Such intrinsic atmospheric variability exhibits little temporal coherence so that the low-frequency variations evident in the ~150-year observational record of the NAO could be interpreted as sampling variability. Wunsch (1999), for instance, has argued that the observed NAO record cannot be easily distinguished from a random stationary process. Indeed, the spectral density of the NAO is nearly white with only slight broadband features near biennial and decadal time scales (Hurrell and van Loon 1997; Jones et al. 1997). However, paleoclimate evidence suggests that NAO variability is highly intermittent and does not exhibit a preferred time scale (Appenzeller et al. 1998a). Nonetheless, the climate system is not stationary, so alternative hypotheses to stationarity also need to be posed (Trenberth and Hurrell 1999). The trend evident in the NAO index over the past 30 years, for instance, exhibits a high degree of statistical significance relative to the background interannual variability (Thompson et al. 1999), and it exceeds the interdecadal variability during the first 100 plus years of the instrumental record. While paleoclimate evidence suggests that prolonged positive and negative NAO phases have occurred in the past (Appenzeller et al. 1998a; Cook et al. 1998; Mann 1999; Luterbacher et al. 1999; Schmutz et al. 2000), the extreme positive values of the index evident since the late 1980s may be unprecedented over the past 5 centuries (Stockton and Glueck 1999). Either the recent trend is a reflection of natural variability occurring on multi-decadal or longer time scales, or it is a component of the global response to external forcing (Corti et al. 1999).
Recently, Thompson and Wallace (1998; 1999) suggested that the NAO might be more appropriately thought of as an annular (zonally symmetric) hemispheric mode of variability characterized by a seesaw of atmospheric mass between the polar cap and the middle latitudes in both the Atlantic and Pacific Ocean basins. A very similar structure is also evident in the Southern Hemisphere. They name this mode the Arctic Oscillation (AO) and showed that, during winter, its vertical structure extends deep into the stratosphere. Similar findings have previously been recognized in the context of tropospheric-stratospheric coupling (Baldwin et al. 1994; Perlwitz and Graf 1995; Cheng and Dunkerton 1995; Kitoh et al. 1996; Kodera et al. 1996). During winters when the stratospheric vortex is strong, the AO (and NAO) tends to be in a positive phase. Baldwin and Dunkerton (1999) suggest that the signal propagates from the stratosphere downward to the surface, so that the recent trends in the tropospheric circulation over the North Atlantic could be related to processes which affect the strength of the stratospheric polar vortex. For instance, tropical volcanic eruptions (Robock and Mao 1992; Kodera 1994; Kelly et al. 1996), ozone depletion (Volodin and Galin 1999), and changes in greenhouse gas concentrations resulting from anthropogenic forcing (Graf et al. 1995; Shindell et al. 1999) all may act to cool the polar stratosphere and strengthen the polar vortex.
On the other hand, it has long been recognized that fluctuations in SST and the strength of the NAO are related (Bjerknes 1964), and there are clear indications that the North Atlantic Ocean varies significantly with the overlying atmosphere. The leading mode of SST variability over the North Atlantic during winter consists of a tri-polar pattern with a cold anomaly in the subpolar region, a warm anomaly in the middle latitudes centered off of Cape Hatteras, and a cold subtropical anomaly between the equator and 30N (e.g., Deser and Blackmon 1993, Kushnir 1994). The emergence of this pattern is consistent with the spatial form of the anomalous surface fluxes associated with the NAO pattern (Cayan 1992). The strength of the correlation increases when the NAO index leads the SST, which indicates that SST is responding to atmospheric forcing on monthly time scales (Battisti et al. 1995; Delworth 1996; Deser and Timlin 1997). But SST observations also display a myriad of long-term (interannual and decadal) responses (Kushnir 1994; Hansen and Bezdek1996; Sutton and Allen 1997; Visbeck et al., 1998), which allows for the possibility that decadal and longer-term variations in the state of the ocean surface imprint themselves back on the atmosphere.
While intrinsic atmospheric variability may exhibit temporal incoherence, the ocean can respond to it with marked persistence or even oscillatory behavior. The time scales imposed by the heat capacity of the upper ocean, for example, can lead to low frequency variability of both SST and lower tropospheric air temperature (Frankignoul and Hasselmann, 1977; Barsugli and Battisti 1998). Basin-wide, spatially-coherent atmospheric modes such as the NAO may also interact with the mean oceanic advection in the North Atlantic to preferentially select quasi-oscillatory SST anomalies with long time scales (Saravanan and McWilliams 1998). Stochastic atmospheric forcing may also excite selected dynamical modes of oceanic variability that act to redden the SST spectrum (Griffies and Tziperman 1995; Frankignoul et al. 1997; Capotondi and Holland 1997; Saravanan and McWilliams 1997; 1998; Saravanan et al. 1999). These theoretical studies are supported by observations of winter SST anomalies born in the western subtropical gyre that spread eastward along the path of the Gulf Stream and North Atlantic Current with a transit time of roughly a decade (Sutton and Allen 1997). Moreover, the SST anomalies reflect anomalies in the heat content of the deep winter mixed layers which, when exposed to the atmosphere in winter (Alexander and Deser 1995), could provide the forcing to drive the NAO on the advective time scale of the gyre (McCartney et al. 1996).
A key question is the sensitivity of the middle latitude atmosphere to changes in surface boundary conditions, including SSTs, sea-ice, and/or land. Robertson et al. (1999a) report that changing the SST distribution in the North Atlantic affects the frequency of occurrence of different regional low-frequency modes and substantially increases the interannual variability of the NAO simulated by their AGCM. The experiment by Rodwell et al. (1999) also points to SST in the North Atlantic as having a marked effect on NAO variability. By forcing their AGCM with observed SST patterns and sea ice distributions, they successfully captured much of the multi-annual to multi-decadal variability in the observed NAO index since 1947, including about 50% of the observed strong upward trend over the past 30 years. However, many other AGCM experiments have led to rather confusing and inconsistent conclusions (Kushnir and Held, 1996). Additionally, several recent studies conclude that NAO variability is closely tied to SSTs over the tropical South Atlantic (Xie and Tanimoto 1998; Rajagopalan et al. 1998; Robertson et al. 1999b). Variations in the tropical Atlantic are substantial and involve strong interannual and decadal variations of meridional SST gradients. Such variations, which most likely affect the Hadley circulation, potentially modulate North Atlantic middle latitude atmospheric variability through an atmospheric bridge mechanism akin to that acting over the Pacific (e.g., Lau and Nath 1996; Trenberth et al. 1998). The response of the atmosphere to changes in tropical, middle and high latitude SST distributions within the Atlantic Basin remains a problem that needs to be addressed.
The role of sea ice in producing atmospheric variability is also not well understood. Changes in sea-ice cover in both the Labrador and Greenland Seas as well as over the Arctic appear to be well correlated with the NAO (Deser et al. 1999). The relationship between the sea level pressure (SLP) and ice anomaly fields is consistent with the notion that atmospheric circulation anomalies force the sea ice variations (Prisenberg et al. 1997). Feedbacks or other influences of winter ice anomalies on the atmosphere have been more difficult to detect, although Deser et al. (1999) suggest that a local response of the atmospheric circulation to the reduced sea ice cover east of Greenland in recent years is also apparent.
Watanabe and Nitta (1999) have suggested that land processes are responsible for decadal changes in the NAO. They find that the change toward a more positive wintertime NAO index in 1989 was accompanied by large changes in snow cover over Eurasia and North America. Moreover, the relationship between snow cover and the NAO was even more coherent when the preceding fall snow cover was analyzed, suggesting that the atmosphere may have been forced by surface conditions over the upstream land mass. Watanabe and Nitta (1998) reproduce a considerable part of the atmospheric circulation changes by prescribing the observed snow cover anomalies in an AGCM.
Several recent studies suggest that both the oceanic wind forced gyre circulation and the thermohaline circulation can actively interact with atmospheric flow to produce coupled decadal and interdecadal climate variability. Latif and Barnett (1996) and Grîtzner et al (1998) argued that when positive SST and sub-surface heat content anomalies in the central North Atlantic, are created by an enhanced subtropical ocean gyre circulation. The response in the atmosphere is an anticyclonic circulation pattern and a weakened storm track, which locally enhances the SST anomalies. The atmospheric response, however, also consists of a wind stress curl anomaly that spins down the subtropical gyre, thereby reducing the northward transport of heat and eventually creating negative SST and sub-surface heat content anomalies. This lag between positive and negative feedback between the ocean and the atmosphere leads to oscillatory behavior on decadal time scales. Other studies have suggested a coupled mode of variability involving the thermohaline circulation. The modeling results of Timmermann et al. (1998), for instance, suggested that an anonymously strong thermohaline circulation produces positive SST anomalies over the North Atlantic. The atmospheric response is a strengthened NAO, which, in turn, produces anomalous fresh water fluxes, and Ekman transport off Newfoundland and the Greenland Sea. The resulting reduction in sea surface salinity is advected by the subpolar gyre and eventually reduces the convective activity south of Greenland, thereby weakening the strength of the thermohaline circulation. The outcome is reduced poleward oceanic heat transport and the formation of negative SST anomalies, which completes the phase reversal and results in multi-decadal variability. It must be recognized, however, that the presence of periodicity or correlated behavior between the atmosphere and ocean in both observations and models does not necessarily imply the existence of a coupled mode. Either damped modes of the uncoupled ocean that are stochastically excited by atmospheric variability (e.g., Saravanan and McWilliams 1998) or unstable modes of the uncoupled ocean that express themselves spontaneously (e.g., Jin 1997; Goodman and Marshall 1999; Weng and Neelin 1998) can also produce such behavior.