North Africa-West Asia (NAWA) sea-level pressure patterns and their linkages with the Eastern Mediterranean (EM) climate



[1] Monthly mean SLP anomalies are analyzed from the 1958–1997 record covering the Mediterranean Basin. From point-correlation technique, a significant winter anomalous SLP oscillation between North Africa (NA) and West Asia (WA) or NAWA, is identified. Insight into the inter-annual variability of NAWA is obtained from the detrended and composited patterns. During P/N, positive/negative values of the index, higher/lower SLP values over the Mediterranean and lower/higher values east of the Caspian Sea are displayed respectively. In both cases, maximum SLP gradients are found over the Eastern Mediterranean (EM). During P/N phases, decrease/increase in winter rainfall amounts and cooler/warmer temperatures are found and could be associated with enhanced northerlies/southeasterlies over the EM. The original and undetrended NAWA indices display mostly P phases during the second half of the period. The regional trend of NAWA index could explain increased drought processes in the EM after the late 70s, in relation with northern hemispheric circulation.

1. Introduction

[2] Several atmospheric sea-level pressure (SLP) or geopotential indices and composited patterns associated with climate variability in the Mediterranean regions have already been identified: the Southern Oscillation Index or SOI [Türkeş, 1998; Kahya and Karabörk, 2001], the North Atlantic Oscillation or NAO [Delitalia et al., 2000; Ben-Gai et al., 2001], the Eastern Atlantic pattern or EA [Wallace and Gutzler, 1981], the Southern Europe-North Atlantic index or SENA, [Kutiel and Kay, 1992], and more recently the East-Atlantic/West Russia index or EA/WR [Krichak et al., 2002]. The regional Mediterranean Oscillation or MO has been documented by Conté et al. [1989]. Finally Kutiel and Benaroch [2002] depicted the North Sea-Caspian pattern or NCP.

[3] While most of the above diagnostic studies dealt with large scale circulation patterns, very few examined the regional signatures of such oscillations or patterns. It is well recognized that distant teleconnections render difficulties in identifying the regional physical mechanisms involved. While influences of large scale circulation on local climatic parameters around the Mediterranean have been presented in several studies [Maheras et al., 1998, 1999; Kutiel and Paz, 1998; Kutiel et al., 2002; Krichak et al., 2002], none included SLP anomalous conditions over North Africa.

[4] The current study deals with climate variability over the eastern Mediterranean Basin (EM), a unique area where both mid-latitudes and equatorial atmospheric systems and processes play significant roles [Krichak et al., 2002]. A new regional SLP index, between North Africa (NA) and West Asia (WA) or NAWA is analyzed during the 1958–1997 period. While relationships between inter-annual variability of the NAWA index, rainfall amounts and averaged temperature in the EM are highlighted, possible long-term linkages with the hemispheric Arctic Oscillation (AO) are also discussed. Indeed, during the second half of the study period the hemispheric winter Arctic Oscillation (AO) index displayed positive values [Thompson and Wallace, 1998] like NAWA. Moreover, composited patterns have been found in proxy records of Red Sea coral and were linked to AO precipitation signatures in the EM [Rimbu et al., 2001]. Since the EM is located between the “two poles” of the NAWA index, it is believed that winter rainfall regime over the EM, already linked to SLP North Atlantic patterns [Eshel at al., 2000], is also linked to the more regional NAWA index. The results presented here should help improve long lead forecasting rainfall schemes developed by Eshel et al. [2000]. In addition, these results combined with modeling activities, should contribute to a better understanding of the mechanisms that modulate climate variability in the EM, such as rainfall decrease during the last two decades [Paz et al., 1998; Xopalki et al., 2000].

2. Data and Methods

[5] A monthly normalized and gridded SLP dataset (5° latitude by 10° longitude, between 0°–80°E and 15°N–50°N), from the Climate Research Unit (CRU) of University of East Anglia [Jones, 1987], is analyzed for a 40-year period (1958–1997). For each season, linear correlations were calculated between SLP time-series at each grid point/cells. Time-series were detrended to highlight inter-annual variability. As expected, correlation coefficients for adjacent cells were highly positive, and decreased with increasing distance. In order to identify if sub-regions had possible out-of-phase SLP behavior with local climate signature, only statistically significant negative correlations were kept (following Wallace and Gutzler, 1981; Conté et al., 1989; Kutiel and Benaroch, 2002). Subsequently, two statistically significant regions over North Africa and West Asia were thus isolated.

[6] Besides evaluating lead-lag correlation between detrended and averaged time-series from the above two regions, a normalized SLP index, describing atmospheric circulation between North Africa and Western Asia, or the NAWA index, is computed. The index allowed identifying positive (P) and negative (N) phases. SLP compositing according to these phases respectively displayed significant and coherent climate patterns. Relationships between these patterns, seasonally averaged rainfall amounts and temperatures for 27 stations covering the EM basin [Paz, 2000], were subsequently examined.

3. Results and Discussion

[7] Significant negative correlation coefficients at the 0.05 level (−0.32 > r > −0.73) were found between the following two regions: the Maghreb-Sahara desert-central Sahel and west Asia. Wallace and Gutzler [1981], Conté et al. [1989] and Kutiel and Benaroch [2002], defined significant and potentially teleconnected areas based upon yearly persistence of negative correlation (for geopotential and/or SLP). Accordingly two teleconnected areas, NA and WA were identified, while NA is the mean SLP over eight North African grid cells and WA is the mean SLP over seven West Asian grid cells (see Figure 1). The subsequent southwest-northeast normalized SLP index, or NAWA, is constructed as follows:

equation image



= a single year


= Standard deviation for NA or WA with:


= equation image (15°N–0°, 20°N–0°, 30°N–0°, 35°N–0°E, 40N°–0°E, 15°N–10°E, 20°N–10°E, 25°N–10°E)


= equation image (40°N–70°E, 45°N–60°E, 45°N–70°E, 45°N–80°E, 50°N–60°E, 50°N–70°E, 50°N–80°E)

Figure 1.

Location of the selected grid points/cells and the two poles NAWA pattern.

[8] Lead-lag correlations were calculated between the NA and WA regions. The most significant results were found during the winter season. The NAWA index (original and detrended) is then displayed in Figure 2. It is worth noting that Wallace and Gutzler [1981] had identified, besides the now classic Pacific and Atlantic NPO and NAO northern hemisphere winter patterns, less marked and less documented out-of-phase geopotential signatures over the Asia/Mediterranean areas.

Figure 2.

Winter time-series (dashed line) of the NAWA index (original data, top and detrended data, bottom). The solid line is a Gaussian running mean.

[9] The winter season is the most meaningful period in the EM when precipitation and temperature variability are being considered. Moreover, the winter atmospheric circulation over WA is driven by the thermal Siberian/Asian anticyclone, centered over north-west Mongolia and Siberia. Nevertheless, and as stated by Lambert et al. [2001], a local maximum of cyclogenesis associated with the Caspian Sea relative warm surface and with Caucasus lee waves, do modulate the pressure field over the WA southern part. The winter atmospheric circulation over NA develops in connection with baroclinic low-latitude penetrating perturbations as a function of the position and intensity of the Azores anticyclone. In addition, several eastern Atlantic winter cyclones penetrate over North Africa and cross the Maghreb from Morocco, towards Algeria and Tunisia [Martyn, 1992]. Subsequent decreases in the SLP pressure field there are then measured. Pressure differences between WA and NA measured by NAWA index, could thus vary contemporaneously and out-of-phase depending upon large scale hemispheric features and associated cyclones tracks during winter.

[10] The undetrended (original) winter NAWA index time-series (Figure 2, top) displays a significant change in sign, toward mostly positive values, after the late 70s. This is also the period when the winter AO index became mostly positive [Thompson and Wallace, 1998] indicating that the cold arctic air was more confined in high latitudes impeding frigid air from moving south. This could have resulted in a significant decrease of SLP values over WA with weakening of the Asian high [see also Paz, 2000].

[11] For more insight into the inter-annual variability of NAWA, the detrended time-series (Figure 2, bottom) is partitioned into 3 phases, namely:

[12] 1. Neutral phase when: −0.5 < NAWAIi < +0.5

[13] 2. Positive (P) phase when: NAWAIi > +0.5

[14] 3. Negative (N) phase when: NAWAIi < −0.5

[15] Composite winter SLP anomalous patterns for N and P phases are displayed in Figures 3a and 3b.

Figure 3.

Winter standardized SLP anomalies during P phases (1980/81, 83/84, 87/88, 88/89, 90/91, 91/92, 92/93, 96/97), top; Winter standardized SLP anomalies during N phases (1964/65, 66/67, 68/69, 70/71, 72/73, 76/77, 78/79, 84/85) Bottom.

[16] During the N (P) phases SLP anomalies over the east Mediterranean and north Africa are lower (higher), while higher (lower) over the Caspian area in western Asia. The maximum anomalous gradients are found in both cases over the Middle East (with reversed signs) indicative of increased southeasterlies (north-westerlies). During N (P) phases, moister (dryer) and warmer (colder) air is advected from the Indian Ocean over the EM and Turkey in particular.

[17] The above results are corroborated by comparing standardized rainfall amounts and averaged temperatures in several northeastern Mediterranean sub-regions during N (P) phases:

[18] - Negative (positive) correlations were found between N (P) phases and rainfall amounts.

[19] - Negative (positive) correlations were found between N (P) phases and temperature.

[20] Based upon recent results dealing with winter relationship between SLP pressure patterns, rainfall and temperatures in the EM [e.g., Kutiel and Paz, 1998; Maheras et al., 1999], one should expect that the significant positive long-term change should decrease the total precipitation amounts and be linked with colder temperatures, as a result of increased frequency of northerlies winds. Accordingly, it is suggested that a close monitoring of the NAWA index, in parallel with modeling studies, could help in contributing to new understanding about drought/flood processes in the EM and the Middle East.

4. Conclusion

[21] A new winter SLP index has been identified between North Africa and West Asia, or NAWA index. Winter rainfall regimes over the EM are dependent, upstream, upon NAO phases, and phase-space trajectories [Eshel et al., 2000]. It is shown here that winter EM rainfall is also linked to the SLP patterns (downstream), over north-west Mongolia/Siberia and Caspian Sea regions. Consequently the regional NAWA index is directly associated with climatic inter-annual variability and changes in the EM. It is believed that by incorporating that index and associated patterns into the forecasting schemes (patterns projection, hindcasting and cross validation techniques) developed by Eshel et al. [2000], could improve not only seasonal to inter-annual prediction but long-lead forecasting scheme over the EM as well. The paper does not address directly the possible linkages between winter EM rainfall and the robust quasi-decadal variability over the North Atlantic [see Tourre et al., 1998 for example]. Nevertheless the low frequency variability in NAWA, a valuable information for early warning systems and mitigation of impacts, could have a direct relationship with positive NAWA winter trend during the studied 40-year period. It could be associated with a similar winter hemispheric AO trend as proposed by Rimbu et al. [2001], when they analyzed Red Sea coral isotopes for the past 250 years and found composited SLP patterns remarkably similar to the ones presented in this paper (see their Figures 1 and 3, for comparison). Our record is somewhat too short to distinguish between quasi-decadal variability and longer-term trend. The links between the NAWA index phases and the EM climate are summarized in Table 1.

Table 1. NAWA Index Phases and its Linkages With EM Climate
ParameterP phaseN phase
Total RainfallDecreaseIncrease
Average temperatureCoolerWarmer
Dominant windsNortherliesSouth-easterlies

[22] The NAWA index could then not only be seen as modulated by the large scale AO, but also as the regional signature of larger winter SLP patterns, including the NAO, which modulate rainfall and temperature over the EM. The trend towards mostly P phases in the NAWA index after the late 70s could explain increased drought processes in that region. For example, the specific role of embedded regional factors such as cyclogenesis from the western part of the Mediterranean area, needs further investigation. While providing more insight into the physical mechanisms involved and developing complementary downscaled modeling studies, long-lead forecasting schemes could be immediately improved.


[23] The authors would like to thank MEDIAS-France, Meteo-France and University of Haifa for their support. Special thanks to Dr. Gerard Begni, Director of MEDIAS-France, who had the initial vision for this fruitful collaboration with the University of Haifa.