Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 The spatial extent of the North Atlantic Oscillation (NAO) is investigated. The results suggest that the spatial structure of the NAO differs significantly according to the phase of the solar cycle. During the solar maximum phases, the NAO has a hemispherical structure extending into the stratosphere, which is similar to the Arctic Oscillation (AO) except for the Pacific sector. However, during the minimum phase, the NAO is confined to the eastern Atlantic sector in the troposphere. Whether or not the NAO is modulated by the solar cycle, these results may shed new light on the controversy of the physical reality of the two modes of variability, the NAO and the AO.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The North Atlantic Oscillation (NAO) is a seesaw of surface pressure in the eastern part of the North Atlantic between the high latitudes and subtropics [e.g., van Loon and Rogers, 1978]. It has long been known that European weather conditions are strongly influenced by the NAO [Walker and Bliss, 1932]. More recently, Hurrell  demonstrated a statistical linkage between the increasing trend in the winter NAO index and the Northern Hemisphere (NH) surface temperature. It is not clear, however, how a regional phenomenon such as the NAO can have hemispheric influence. Thompson and Wallace  responded to this question by introducing the concept of Arctic Oscillation (AO), which is a hemispheric mode of variability. The spatial pattern of the AO is similar to the NAO over the Euro-Atlantic region, but it is characterized by a zonal structure. Hence, the AO is also referred to an “annular mode” [Thompson and Wallace, 2000].
 The AO is usually defined as the leading mode of the Empirical Orthogonal Function (EOF) of the sea level pressure (SLP) over the northern hemisphere (NH). If, however, EOF analysis is applied to the Atlantic sector alone, or Rotated EOF (R-EOF) analysis is used, the result would be a more sectorial structure, such as the NAO. Therefore, the hemispheric structure of the AO may be simply due to an involved mathematical constraint [Deser, 2000; Ambaum et al., 2001]. However, if a hemispheric mode of variability really exists, does the application of the R-EOF exaggerate the local structure because of the converse mathematical constraint?
 If two modes exist and do not occur simultaneously, it would be possible to separate them by dividing the data by time. In the present study, data are grouped according to the phase of the solar cycle. The question of why the NAO should be modulated by the solar cycle will be discussed elsewhere and is not examined here.
 The NAO index used in the present study was calculated by Hurrell  as a difference in the normalized monthly mean SLP between Lisbon and Stykkisholmur. Following Hurrell, an extended winter (DJFM) mean NAO index is used as a representative standard throughout the present study. Surface temperature (combined land surface air temperature and sea surface temperature) data were compiled by Jones . All other meteorological variables are from the NCAR/NCEP reanalysis dataset [Kalnay et al., 1996]. For a measure of the solar activity, the winter (DJF) mean 10.7 cm solar radio flux is used.
Figure 1 shows a standardized (a) DJFM-mean NAO index and (b) a DJF-mean 10.7 cm solar radio flux for the 39 years from 1958/1959–1996/1997. Visual inspection easily reveals that there is no correspondence between the peaks of the NAO and solar cycle. In fact, spectral analysis [Hurrell and van Loon, 1997] shows that the quasi-periodic variability in the NAO is about 8 years and shorter than the 11-year solar cycle. Years of higher or lower solar activity compared to the average values are indicated by solid and open circles, respectively. They are also designated as solar maximum and minimum phases, respectively.
 The mean spatial feature of the NAO can be seen in a correlation map between the NAO index and the SLP. Figure 2 shows the correlation coefficients between the DJFM-mean NAO index and the SLP at each grid point calculated from a total of 39 years. The contour interval is 0.1, and absolute values below 0.5 are omitted. The correlation map exhibits a familiar dipole pattern over the North Atlantic. This average feature is significantly different between the solar maxima and minima.
Figure 3 shows the same correlation maps as Figure 2 but calculated separately from (a) 17 winters during the solar maxima and (b) 22 winters during the solar minima. If the data of each winter are independent, the correlation coefficients corresponding to 95%-significant levels are 0.48 and 0.42 for the maximum and minimum cases, respectively. To show only the significant part of the variability, the contour lines of absolute values less than 0.5 are suppressed. The correlation pattern shows a compact north-south seesaw over the Atlantic Ocean during the minimum phase. This regional characteristic of the NAO fits the nomenclature of “the North Atlantic Oscillation” quite well. However, during the maximum phase, the NAO-related variability extends all over the NH, except for the Pacific sector.
 Correlation coefficients between the NAO index and surface temperature at each grid point are also displayed in the bottom panels in Figure 3. During the solar minima (Figure 3d), temperature anomalies exhibit a seesaw between Europe and Greenland, which is a well-known feature of the NAO. However, in the case of the solar maxima (Figure 3c), although similar temperature anomalies are found over the Atlantic sector, the largest signal appears over the Eurasian continent.
 Another large difference between the solar maxima and minima is found in the vertical structure. Figure 4 shows the correlation coefficients between the DJFM-mean NAO index and (top) the zonal-mean zonal winds at each grid point and (bottom) the zonal-mean temperature at each grid point. During the maximum phases (Figure 4a), zonal winds show high correlation around 55°–60°N latitudes extending from the surface to the stratosphere. In the temperature field (Figure 4c), a seesaw between the polar and equatorial temperatures is found in the lower stratosphere-tropopause region, while, near the surface, a cooling in the subtropics contrasts a warming in the mid-latitudes. During the solar minima, a significant correlation is limited only near the surface around 55°N latitude in the zonal wind field (Figure 4b), and no correlation is found in the zonal temperature field (Figure 4d).
 Separation of the dataset according to the phase of the solar cycle reveals a quite different spatial structure associated with the winter NAO index. Leaving for future studies the problem of whether this is really caused by the sun, the present results suggest the existence of two distinct modes of variability. Both have a similar horizontal structure over the North Atlantic sector, but one is a synoptic (S) scale and the other is a planetary (P) scale extending over the NH (For convenience, hereafter, they are referred to as types S and P, respectively).
 Because types S and P have a similar structure within the Atlantic sector, both exert influence on the NAO index. It would be impossible to separate two such modes of variability by applying either hemispheric EOF or R-EOF analysis: hemispheric EOF analysis would preferentially produce a large-scale feature, while R-EOF creates a local one, but neither is obtained at once [Dommenget and Latif, 2002]. In this respect, it is interesting that the “non-linear principal component analysis” [Monahan et al., 2000] reveals two modes of variability over the Euro-Atlantic sector. One is a blocking-type circulation over the East Atlantic, and the other has a more hemispheric extent and is referred to as an “Arctic-Eurasia Oscillation.” These two modes of variability can be compared to the two types found in the present study.
 The spatial structure of the type P is essentially the same as a “stratosphere-troposphere coupled mode” or the “Polar/ EurAsia (PEA) pattern” [Kodera et al., 1996], which is characterized by changes in zonal flow and meridional propagation of stationary waves in the troposphere. From this, we speculate that the interaction between the planetary waves and zonal flow is important for type P. Concerning type S, it appears preferentially during the minimum phases. For instance, in the winter of 1996, during a solar minimum phase, the NAO index took a large negative value (Figure 1), and Europe experienced a cold winter [Kushnir, 1999] due to a frequent blocking situation. However, no important impact was observed within the Eurasian continent. It should be remembered that two possible mechanisms are proposed for the generation and maintenance of the NAO. One is the interaction between transient eddies and regional stationary eddies [e.g., Hurrell, 1995], and the other is the interaction between the zonal flow and stationary waves [DeWeaver and Nigam, 2000]. Thus, the difference in the horizontal and vertical structure between types S and P could be attributed to the difference in the involved waves.
 Currently, active discussions are in progress regarding whether the NAO or the AO has a physical reality [Wallace, 2000; Ambaum et al., 2001]. The present result suggests that there are two different modes of variability in types S and P. If type S is called the NAO by its regional characteristic and type P the AO by its hemispheric feature, then both the NAO and AO could have physical reality. It should be remembered that the AO is also called the “Northern Hemisphere Annular Mode” (NAM). The correlation pattern in Figure 3a, however, exhibits a complete lack of a signature over the Pacific sector, and the “annular pattern” is difficult to recognize. With the exception of the Pacific sector, the horizontal and vertical structure of type P is very similar to that associated with the NAM [cf. Figures 3 and 4 with Figures 6 and 7 of Thompson and Wallace, 2000]. This suggests that the change in the polar vortex is crucial for the coupling of the stratosphere and the troposphere but the annular structure in the mid-latitudes of the troposphere is not essential.
 The present analysis suggests two phenomena having a similar structure over the Euro-Atlantic region but a different vertical and horizontal extent. It is speculated that different spatial characteristics are due to the difference in the involved waves: planetary waves for a hemispheric one extending to the stratosphere (type P) and transient waves for a regional one trapped in the troposphere (type S). This needs to be demonstrated in future studies.
 In the present study, data are divided according to the phase of the solar cycle. If the two different modes of variability really exist, why should they be modulated by the solar cycle? Because the present analysis covers only four solar cycles, the relationship with the sun could be a coincidence. It should be noted, however, that the lagged correlation between the NAO index during the solar maxima reveals a downward propagation of zonal winds (figure not shown) similar to that found in Kuroda and Kodera . On the other hand, zonal wind anomalies produced by solar activity in the stratopause region in early winter can propagate downward into the troposphere through interaction with planetary waves [Kodera, 1995; Shindell et al., 2001]. Thus, modulation effects by solar cycle could originate in the upper stratosphere. This also needs to be investigated in future studies.
 The author thanks E. C. Weatherhead for valuable comments and encouragement. The surface temperature dataset was originally created by P. D. Jones, and updated data were obtained from T. Michel, JISAO. The NAO index and NCEP reanalysis data were obtained from NCAR and CPC, respectively. Solar radio flux data were obtained from the NOAA National Geophysical Data Center.