The North Atlantic Oscillation NAO

The NAO has long been recognized as a major mode of atmospheric variability. In its simplest definition, the NAO describes co-variability in SLP variations between the Icelandic Low (IL) and Azores High (AH), the two "centers of action" in the atmospheric circulation of the North Atlantic. When both are strong, the NAO is taken to be in its positive mode. When both are weak, the NAO is in its negative phase. This co-variability is associated with prominent high-latitude signals in SAT, cyclone activity, moisture transport, precipitation, SST and ocean heat transport. The NAO is best expressed during winter and research has naturally focused on this season. Nevertheless, the NAO can be identified any time of the year. As reviewed by van Loon and Rogers (1978), manifestations of the NAO have been recognized for centuries. For example, the missionary Hans Egede Saabye recorded in a diary kept in Greenland during the years 1770-78 that the Danes were well aware that when winters in Greenland were especially severe, the winters in Denmark tended to be mild, and vice versa. Description of this temperature see-saw associated with the NAO can be found in writings throughout the nineteenth century.

The first formal description of the NAO is credited to Sir Gilbert Walker. In a paper with E. W. Bliss (Walker and Bliss, 1932), a selected set of highly correlated SAT, SLP and precipitation time series at widely separated stations over eastern North America and Europe were combined to develop an NAO index. The same approach was used to identify the North Pacific Oscillation (NPO) and the Southern Oscillation (SO). Walker and Bliss (1932) described the NAO as "a tendency for pressure to be low near Iceland in winter when it is high near the Azores and southwest Europe; this distribution is of course associated with high temperatures in northwest Europe and low temperatures off the Labrador coast". The NAO hence describes the meridional gradient in SLP over the north Atlantic, which can also be interpreted in terms of the strength of the mid-latitude westerly (west to east) winds.

Walker and Bliss (1932) suggested that a simpler index of the NAO could be based on the pressure difference between Iceland and the Azores (roughly representing the climatological centers of the IL and AH). The utility of this simple index was confirmed by Wallace and Gutzler (1981) and has been widely adopted. There are a number of variants of the station-based NAO index using different stations and techniques. Most commonly it is based on the difference in standardized station SLP, the intent of the standardization being to reduce the dominance of the Iceland station. Rogers (1984) used Ponta Delgadas, Azores, and Akureyri, Iceland, while Hurrell (1995,1996) used Lisbon, Portugal, and Stykkisholmur, Iceland. By contrast, van Loon and Rogers (1978) used an index based on the east-west contrast in SLP across the subpolar north Atlantic.

A disadvantage of an SLP-based index using two stations is that it may not optimally capture the negative correlation between the subtropical and subpolar SLP centers. For example, under the positive (negative) NAO mode, the IL and AZ are not just stronger (weaker), but exhibit poleward (equatorward) shifts. The centers of maximum anticorrelation also shift seasonally. In recognition, Portis et al. (2001) developed a seasonally and geographically varying "mobile" index of the NAO, defined as the difference between normalized SLP anomalies at the locations of maximum negative correlation between the subtropical and subpolar North Atlantic SLP. They find that the subtropical center shifts westward and northward into the central North Atlantic as the annual cycle passes through spring and summer. When tracked with their mobile NAO index, the NAO nodal correlations are only slightly weaker in the summer half of the year as opposed to the winter half. Their index also shows that the NAO is strongest in March, followed by February and January.

Other investigations have made use of so-called empirical orthogonal function (EOF) or rotated principal component (PC) analyses. These approaches have been applied variously to a North Atlantic domain or the Northern Hemisphere, using grid-ded SLP or tropospheric height fields (e.g., Trenberth and Paolino, 1980; Wallace and Gutzler, 1981; Barnston and Livezy, 1987; Thompson and Wallace, 1998, 2000; Hurrell et al., 2003). The NAO emerges very clearly as the leading mode of variability in the North Atlantic study of Trenberth and Paolino (1980), accounting for 34% of the variance in winter SLP. The first EOF from that study, reproduced in Figure 11.8,

Nao Slp Eof

90W 60UV 30W 0 30E

Figure 11.8 The first EOF of winter (December-March) SLP for the North Atlantic sector over the period 1899-1977 based on Trenberth and Paolino (1980) (from Dickson et al., 2000, by permission of AMS).

90W 60UV 30W 0 30E

Figure 11.8 The first EOF of winter (December-March) SLP for the North Atlantic sector over the period 1899-1977 based on Trenberth and Paolino (1980) (from Dickson et al., 2000, by permission of AMS).

Figure 11.9 Normalized indices of the winter (December-March) NAO, based on: the difference in normalized SLP between Lisbon, Portugal, and Stykkisholmur/Reykjavik, Iceland; PC1 of Atlantic sector SLP (20-70° N, 90° W-40° E); and PC1 of Northern Hemisphere SLP (20-90° N). The last time series is also known as the Arctic Oscillation or Northern Annular Mode. The heavy solid lines represent the indices smoothed to remove fluctuations with periods less than four years (updated from Hurrell et al., 2003, by permission of AGU).

Figure 11.9 Normalized indices of the winter (December-March) NAO, based on: the difference in normalized SLP between Lisbon, Portugal, and Stykkisholmur/Reykjavik, Iceland; PC1 of Atlantic sector SLP (20-70° N, 90° W-40° E); and PC1 of Northern Hemisphere SLP (20-90° N). The last time series is also known as the Arctic Oscillation or Northern Annular Mode. The heavy solid lines represent the indices smoothed to remove fluctuations with periods less than four years (updated from Hurrell et al., 2003, by permission of AGU).

shows the dipole structure of the NAO. Similar results can be obtained through composite analysis of SLP fields for extremes of the NAO index. EOF or PC approaches are usually considered more "robust" than station-based indices in that they use information from the entire atmospheric field to extract the dominant modes of variability. On the other hand, from a practical viewpoint, individual station time series from which indices can be compiled are available for a longer time period than are gridded fields.

NAO time series for winter (December-March) 1864 through 2004, based on three different approaches using SLP data (station index, PC1 of the Atlantic sector, PC1 of the Northern Hemisphere) are presented in Figure 11.9. As mentioned, the advantage of the station-based index (top panel) is that a longer record can be obtained (back to 1864). The time series structure from the three approaches is very similar for the period of overlap. As discussed shortly, the PC1 for the Northern Hemisphere has come to be known as the AO or NAM. There is little evidence of any preferred time scale of variability. The studies of Hurrell and van Loon (1997) and Cook et al. (1998) document concentrations of spectral power in the NAO at 2.1, 8 and 24 years, but as noted by Hurrell et al. (2003) there are no significant peaks. Furthermore, large changes can occur from one winter to the next, and there is considerable variability within a given winter season.

While consistent with the notion that much of the variability in the NAO is from processes internal to the atmosphere (Hurrell etal., 2003), one can expect surface influences. One such forcing that could operate on seasonal and shorter time scales is sea ice. Indeed, two recent modeling studies (Alexander etal., 2004; Deser et al., 2004) find that when the sea ice margin in the Atlantic sector retreats, it invokes a local change in heating, which alters the North Atlantic storm track. This appears to involve at least a weak response of the NAO. However, most of the interest in surface forcings is with respect to the weak low-frequency variability in the NAO. While low-frequency variability does not stand out as significant in a spectral analysis, Portis et al. (2001) nevertheless identify four epochs of the NAO. There was an epoch from about 1870 through about 1900 when the NAO was mostly negative, followed by mostly positive values from 1900 to 1950. This was followed by a negative period from 1960 through 1980, and then a positive epoch from the 1980s through close to the present (see also Polyakov and Johnson, 2000). The last two epochs stand out. Over the record from the station-based time series, the NAO was at its most negative in the mid 1960s and at its most positive in the late 1980s. Accordingly, there is a strong upward trend between these two periods although the past decade shows a regression toward more neutral values.

There is growing evidence that low-frequency NAO variability involves large-scale ocean interactions. As reviewed by Hoerling et al. (2001), while the observed spatial pattern and amplitude of the NAO is typically well-simulated in AGCMs with fixed climatological annual cycles of solar radiation, trace gas composition and SST, such simulations do not reproduce interdecadal changes comparable in magnitude to those that are observed. This suggests a role of slowly varying SST. Rodwell et al. (1999) find that much of the multidecadal variability of the winter NAO over the past 50 years can be simulated when the observed temporal evolution of North Atlantic SST distributions is included. They argue that the SST characteristics are communicated to the atmosphere through evaporation, precipitation and atmospheric heating processes and that with knowledge of SSTs, it may be possible to predict European winter climate several years in advance.

Hoerling et al. (2001) counters that most evidence actually suggests that anomalous SST and upper-ocean heat content feedbacks to the atmosphere from the extratropics are rather small, implying that a strict focus on the North Atlantic is inappropriate. Their modeling study, which focuses on the recent upward trend, indicates a more important role of progressive warming of tropical SST, especially over the Indian and Pacific Oceans. Follow-on papers (Hoerling et al., 2004; Hurrell et al., 2004) strengthen this argument. These ocean changes alter the pattern and magnitude of tropical rainfall and atmospheric heating, which in turn has forced a trend toward the positive phase of the NAO. They suggest that the rise in tropical SST may contain an anthropogenic component. Cassau and Terray (2001) also focus on the tropics but provide a somewhat different view - they find links between the positive phase of the NAO and the cold phase of ENSO (La Nina).

0 0

Post a comment