Home › Forecasts › Trends › North Atlantic › The NAO and Barents Sea climate › The NAO and Barents Sea climate

The NAO and Barents Sea climate

A conventional view of climatic variations in the Barents Sea (as well as the Norwegian Sea and adjacent regions) is that they are driven largely by the NAO: When the NAO is in its positive (negative) phase, the Barents Sea is warmer (colder) than normal (Dickson et al., 2000). It is surprising, therefore, that the correlation coefficient between the winter (DJF) NAO index and winter (DJFM) temperature at the Kola section is only 0.21 for the entire period of observation from 1922-2008. For the time lags of 1 and 2 years (NAO leads), possibly required for Atlantic waters to reach the Barents Sea, the correlation coefficients are 0.24 and 0.14, respectively.

The cross-correlation function between the NAO index and Kola section temperature is presented in Fig. 7. It suggests that both time series have substantial energy of fluctuations at the period of about 80 years. The maximum of the cross-correlation function of 0.47 (significant at the 95% level) is reached when the NAO leads Kola temperature by 16-17 years. 


Fig. 7. Cross-correlation function between the NAO index and Kola Section temperature for individual winter values (blue line) and values smoothed by 5-yr running means (red line). Data for 1922-2008.

In Fig. 8 both time series are overlapped, with the NAO series shifted forward by 17 years. The match between the series is closer since the late 1970s, when the correlation coefficient reaches 0.71 (data for 1979-2008). Obviously, this correlation coefficient is somewhat inflated due to the strong upward trend during this period. When the linear trend is removed the correlation coefficient is reduced to 0.49, but still remains statistically significant at the 95% level.  

A significant portion of NAO variability occurs at the interannual time scale. At this time scale, an advection of warm Atlantic air and water into the Norwegian Sea during a positive NAO phase is not enough to increase SST in the region substantially enough to exert any feedback effect on the Icelandic low. If, for some reason, a positive phase of the NAO continues for a number of years, it may lead to an increasingly intense and widespread influence of Atlantic waters in the Nordic Seas (Dickson et al., 2000), although the exact mechanisms linking the anomalous wind field to the inflow are not clear (Furevik and Nilsen, 2005). The increased presence of warm Atlantic waters causes an enhanced upward ocean heat flux creating a situation conducive to a more eastward positioning of the Icelandic low (Bengtsson et al., 2004). A strong, single-centered and shifted eastward Icelandic low implies further warming in the Norwegian and Barents Seas. When the Icelandic low is shifted eastward far enough, it also intensifies an advection of polar water and sea ice transport through Fram Strait, and hence cooling in the waters off the east coast of Greenland (Hilmer and Jung, 2000). A sharpening contrast between the cold west an warm east in the Nordic Seas establishes favorable conditions for further diabatic contributions to cyclonic development and, hence, strengthening of the Icelandic low (Rogers et al., 2004). Thus, a positive feedback loop is created that leads to an amplified warming in the Norwegian and Barents Seas and the entire Arctic.


Fig. 8. Time series of winter (DJFM) Kola Section temperature (color) and winter (DJF) NAO index (black line). The latter is shifted 17 years forward.

Since the Icelandic low is shifted eastward, SLP near Iceland increases. The advection of cold arctic air west of Greenland is also reduced and the area warms up. As a result, the NAO index (computed using either SLP or temperature data) decreases. One may say that the decrease of the NAO index in this case is strictly technical, simply because of the way the index is calculated. Meantime, the Icelandic low remains strong, with its central pressure being below normal.

As the Arctic becomes anomalously warm, the equator-pole gradient diminishes and the zonal hemispheric circulation weakens. Being part of this circulation, the Icelandic low also starts weakening, and the advection of warm air along its eastern periphery northward diminishes. This starts the cooling trend in the Barents Sea. Another mechanism that leads to a weakening of the Icelandic low is associated with an intense freshwater flux through Fram Strait by a strong and eastward shifted Icelandic low. Due to anomalous freshwater inflow, water stratification increases, deep convection ceases, ice appears in the Greenland Gyre, the atmosphere starts cooling and SLP increases (Dukhovskoy et al., 2004).

After the Arctic cools down, the equator-pole gradient increases again and the hemispheric zonal circulation intensifies. The strong zonal flow acts as a barrier to warmer air reaching the high latitudes and the Arctic continues to cool, which creates a positive feedback. In the Southern hemisphere, which has significantly more ocean and less land than the Northern hemisphere, this effect is expressed even stronger (Thompson and Solomon, 2002). The break of the loop occurs when the westerlies accelerate to the point when it becomes unstable. 

As the westerlies slow down and the Icelandic low retreats to its more normal position between Iceland and the southern tip of Greenland, the fresh water inflow to the Nordic Sea though Fram Strait decreases. The reduction of fresh, buoyant surface water combined with overall colder temperature in the Arctic decreases water column stability and favors deep convection in the Greenland Sea. This thermohaline ventilation generates overflows to the North Atlantic, which also require a compensating inflow of Atlantic water to the Nordic Seas (Hansen and Osterhus, 2000). This starts a new cycle of warming, deepening of the Icelandic low and its shift eastward.

In light of this conceptual model, the sequence of climatic events in the past 100 years can be interpreted as follows:

1900s - mid-1920s: This period was characterized by a strong zonal circulation as evidenced by the frequency of Vangenheim-Girs’ zonal atmospheric circulation type W (Girs, 1971) or atmosphere’s angular momentum (Salstein and Rosen, 1986). The latter reached the record maximum (for the period 1860-1985) in the early 1910s. The general westerly type (W) of Lamb (1972), which corresponds mostly with SW'ly winds at the surface, especially in southern Britain, was most frequent in the early 1920s. Overall, this period was the coldest in the Arctic in the last 100+ years (Polyakov et al., 2003).
 Mid 1920s - 1940s:  A sharp increase of the Arctic temperature in the 1920s was one of the most fascinating climatic events of the 20th century. This Arctic warming was associated with an eastward shift of the Icelandic low, which can be seen in Fig. 6 in van Loon and Rogers, 1978. This is also consistent with (Bengtsson et al., 2004) who found a steadily increasing transport of warm water into the Barents Sea during this time driven by increasingly southwesterly to westerly winds between Spitsbergen and the northernmost Norwegian coast. The warming in the Barents Sea (Fig. 3) and the Arctic (Polyakov et al., 2003) peaked in the late 1930s. At that time, the NAO index (both SLP-based and SAT-based) was already declining.
 1950s - 1970s:  The NAO index continued to decline and reached extremely negative values in the 1960s (Fig. 4). The Icelandic low was weak (and often split into two centers) and unable to pump much warm air into the Nordic Seas. Temperature in the Barents Sea and the Arctic overall cooled down substantially. In the 1950s and 1960s, water stratification in the Greenland Sea was weak, suggesting an intensification of deep-water convection (Vinje et al., 2002). In the early 1970, the mid-latitude westerly winds over the northern hemisphere sharply increased, perhaps due to a strong temperature gradient between the pole and the equator. However, the warming in the Barents Sea was not enough to cause an eastward shift of the Icelandic low. Therefore, when the Icelandic low weakened again in the late 1970s (probably in association with the regime shift in the Pacific climate), the Arctic was still too cold and temperature in the Barents Sea plunged to the lowest values on record (Fig. 3).
 1980s-2008:  Strong and sustained zonal circulation has established over the Northern hemisphere during this period. Its characteristic feature is a northward shift of the jet streams, accompanied by a widening of the tropical belt (Seidel et al., 2008). Both the inflow of the Atlantic waters and the upward oceanic heat flux in the Barents Sea has increased substantially (Zhang et al., 2004). As a result, the Icelandic low was consistently shifted eastward since the 1990s (Jung et al., 2003; Hurrell and Dickson, 2004) and the Arctic warming accelerated. Years 2005-2008 were particularly warm in the Barents Sea rivaling those in the late 1930s (Fig. 3). The NAO index, however, after reaching the maximum in the early 1990s, has started to decline (Fig. 1). The discrepancy between the NAO/AO indices and Arctic temperature variability in recent years is termed the “Arctic paradox” (Overland and Wang, 2005).
 2009-2023:  If the 17-yr time lag between the NAO index and Kola Section temperature remains in the future, one can expect that the Barents Sea will remain anomalously warm until about 2012 (Fig. 8). After that, a long term cooling trend until about 2023 is likely. The cooling may start as a sharp decline of winter temperatures.