Following work by Paz et al. , the relationship between NAWA and the Eastern Mediterranean (EM) winter climate is explored further. It is found that NAWA can be viewed as the regional signature of the hemispheric northern annular mode (NAM), at all time scales. While in-phase variability of winter NAM and NAWA increased after the so-called climate shift of 1976–77, largest correlation between the two indices is found during January. This is also the month when largest correlations are to be found between NAM and the NAWA poles. Correlations between NAWA and NAM are larger than that between NAWA and the North Atlantic Oscillation (NAO). Further evidence of direct relationship between NAM, NAWA, winter air temperature and rainfall in the EM is given.
 The Northern Annular Mode (NAM) displays regional patterns with significant climatic impacts in the northern hemisphere [Thompson and Wallace, 2001]. Over the North Atlantic, where the thermally driven subtropical is weak, the North Atlantic Oscillation (NAO) can be considered as a strong western Europe-Atlantic signature of NAM [Visbeck et al., 2003]. More specifically over the Atlantic sector, baroclinic waves have marked meridional excursions, which could explain the extent of the NAM signature there, and the attempt of recasting the NAO as an expression of an annular mode [Thompson et al., 2002]. While the Arctic Oscillation (AO) and NAM present some similarity [Thompson and Wallace, 1998], Cullen et al.  have shown that phases of the Arctic Oscillation (AO) are associated with precipitation variability over the Eastern Mediterranean (EM). Deser et al.  have also shown the relationship between Arctic sea-ice advection and thickness and phases of NAM.
Paz et al.  have recently isolated a sea level pressure winter index over the Mediterranean basin (NAWA - North Africa/Western Asia index), and shown its possible relationship with winter climate over the EM and with the AO [see also Rimbu et al., 2001]. In this paper, linkages between NAWA and NAM are demonstrated. NAWA being most correlated with NAM can be viewed as a NAM signature over the Mediterranean basin.
 In Figure 1 the normalized time-series for winter NAWA and NAM indices are plotted. The significant correlation at the 0.05 level using z-test, is of 0.66. Conspicuous overall linear trends are present in both time-series. Slightly lower correlation was found (0.56) when the time-series were both detrended as might be expected. The significant correlations between NAM and the poles of NAWA, NA and WA following Paz et al. , are of 0.71 and −0.47 (for original data) and 0.63 and −0.34 (for detrended data).
 Most of the time there is an in-phase relationship, with evidence of higher frequency variability prior to winter 1976–77. It was found that the NAWA-NAM cross-spectrum analysis displays large squared-coherency around 2.8- and 8-year periods (not shown). After the winter of 1976–77, when a so-called climate shift occurred in the northern hemisphere [Trenberth, 1990], NAM and NAWA display in-phase lower-frequency variability. The correlations between the two indices increased accordingly to 0.71 (original data) and 0.65 (detrended data) after 1976–77, when compared to correlations of 0.51 (original data) and 0.48 (detrended data) prior to winter 1976–77.
 It has been shown that quasi-biennial (QB) variability does impact the NAM variability [Baldwin et al., 2001]. This must have been particularly true prior to winter 1976–77. The correlations for each of the winter months (DJF) are the largest and most significant during January. In Figure 2 the normalized time-series for NAWA and NAM indices are then plotted for January: correlations of 0.69 and 0.65 (0.05 level, using z-test) are found for original and detrended data, respectively.
 The latter results can be explained as follows: while the NA pole of NAWA is correlated to the NAO (a regional signature of NAM) in January (0.60 and 0.61 for original and detrended data), the WA pole of NAWA is also anti-correlated to NAM in January (−0.50 and −0.42 for original and detrended data). This is further corroborated by results from Thompson et al. , showing a Euro-Atlantic signature of NAM, with extension of the NAM pattern over western Asia (see their Figure 1). As such, the January correlation between NAWA and NAM detrended time-series (0.65) is larger than the correlation between NAWA and NAO detrended time-series (0.55). This emphasizes the out-of-phase relationship between the two poles of NAWA as demonstrated by Paz et al. , and that NAWA, a regional signature of NAM, can then be used as a new, useful index for climate prediction over the EM. This is further discussed below.
3. Discussion and Conclusion
 It has been demonstrated that NAM has regional signatures over the Pacific ocean, the Atlantic ocean and the Euro-Atlantic sector, [Thompson et al., 2002; Visbeck et al., 2003]. While the AO and the NAO, intrinsic atmospheric modes [Gong et al., 2003], share common features, the AO has a weak signature in the North Pacific. It has been shown that the northern center of action of AO had shifted eastward during the mid-seventies, and that the Barents Oscillation in the extreme north Pacific and the AO are not independent [Tremblay, 2001]. In this paper, while the relationship between AO and NAM have still to be elucidated, as well as the atmospheric processes giving rise to NAM, evidence of a regional signature of NAM over the Mediterranean basin is given: the NAWA index. Several time scales (i.e., decadal, inter-annual, quasi-biennial) of variability, for both NAWA and NAM can be considered. This is also corroborated by recent findings by Hu and Tung , showing that trend, decadal and inter-annual variability of the NAM index are correlated to radiated arctic cooling, and an increase in greenhouse gases (GHG), rather than a result of a dynamical decrease of planetary wave activity. The low-frequency tendency in NAWA could then be interpreted as a regional signature of the hemispheric NAM and associated climate trends [Thompson and Wallace, 2001]. Although not unequivocally proven, NAWA trend could be due to enhanced concentration of greenhouse gases (GHG) in the northern hemisphere [Fyfe et al., 1999] and lesser snow extent, particularly over Siberia, as proposed by Gong et al. .
 Recurring drought processes in the EM and after the late 70s, have been identified by Paz et al.  and Xoplaki et al. , among others. This could be additional evidence of a Mediterranean climate change associated with hemispheric trends in GHG concentration [e.g., Shindell et al., 1999]. The apparent changes in frequencies and amplitudes of NAWA anomalies before and after winter 1976–77 (increased correlations between NAM and NAWA after winter 1976–77, when low-frequency signals seem to dominate), might be due to one of the so-called climate shifts during the 20th century [e.g., Trenberth, 1990; Minobe, 1999]. The quasi-biennial signal which does impact the NAM [Coughlin and Tung, 2001], and is visible in NAWA as well, was indeed most active prior to 1976–77.
 Linkages between winter NAWA and air temperature, and precipitation over the EM have been shown by Paz et al. [2003, Figure 3]. From the highly significant correlation between NAWA and NAM presented in this paper, additional correlations were computed between NAM, air temperature, and precipitation at stations in the EM (location of stations used in this study are displayed in Figure 3). The significant correlations are presented in Table 1 (for winter air temperature) and Table 2 (for winter precipitation):
Table 1. Significant Correlation Coefficients (α ≤ 0.05) During Winter (DJF) Between NAM and Air Temperature From Stations in the Eastern Mediterranean (EM)
Greece and Greek Islands
Table 2. Significant Correlation Coefficients (α ≤ 0.05) During Winter (DJF) Between NAM and Rainfall From Stations in the Eastern Mediterranean (EM)
Greece and Greek Islands
 The results presented in the two tables show that NAM has coherent regional signatures over the EM. This corroborates findings from the coherent composite SLP patterns and surface atmospheric circulation associated with the NAWA phases [Paz et al., 2003] or the AO phases [Rimbu et al., 2001]. It is also in full agreement with results from Thompson and Wallace , showing that NAM's polarities are associated with significant differences in distribution of precipitation in the Middle East.
 NAWA is better correlated to NAM than the winter NAO. This is certainly due to the fact that the two poles of NAWA are on the western side of the Mediterranean Basin (NA) and south-western part of Eurasia (WA), respectively. The northern hemispheric frequency of blocking and cold air outbreaks are reflected in NAM/AO phases. When the winter AO is in positive (negative) phase, the preceding fall has less snow extent than normal [Gong et al., 2003], the winter Siberian High is weaker (stronger) than normal (D. Gong and S. Wang, Arctic Oscillation and climate of China in winter, submitted to Advances in Atmospheric Sciences, 2000), meaning cyclonic anomalies there, with a tendency to bring northern cold air masses from Eurasia over the EM. The air flow originates from the North Atlantic sector, and its downward motion over Siberia is linked to the AO phases (J. Wang and B. Wu, The connection between the winter Arctic Oscillation and the Siberian High, the East Asian winter monsoon, and sea-ice extent, unpublished manuscript, 2002). The same could happen with the phases of NAM, associated with dynamics not unique to the North Atlantic sector. The tendency for positive phases of NAM after the mid-70s [Thompson et al., 2002, Figure 7] is present in NAWA. Following Paz et al. , this could further explain the increase in drought processes over the EM during the same period.
 Indeed, EM precipitations have different origins. It is well known that when the winter NAO is positive, winter storm tracks are increased in the North Atlantic. The reverse means more storms over at least the western Mediterranean Basin [Eshel et al., 2000]. Lack of winter precipitation over the EM could also be due to less water-vapor from the southeast. Our results tend to show that the two mechanisms are not separated, and in any case have similar effects on EM climate. More study is required to isolate the physical mechanisms at stake. While the above results can be applied for management and decision–making policies, a sound theory for the existence and variability of NAM remains to be established.
 Tourre and Paz thank Dr. Gérard Bégni, Director of MEDIAS-France for promoting and supporting this joint research project over the Mediterranean Basin. The authors thank Prof. H. Kutiel for his constructive comments and remarks. Paz would like also to thank the University of Haifa for its invaluable contribution. Tourre thanks also Dr. Mike Purdy, Director of LDEO of Columbia University for his continuing support. This is LDEO contribution # 6642.