Influence of the Gulf Stream on the Barents Sea ice retreat and Eurasian coldness during early winter

To understand interannual temperature variations in the Barents Sea sector, we focused on time series of surface air temperature anomalies during December at Bear Island (figure 1(a)). Referring to this time series, we selected typical warm and cold Decembers for which the temperature values exceed the 0.8 standard deviation (warm Decembers: 1984, 2004-2009, and 2011; cold Decembers: 1980–1981, 1983, 1987–89, 1995–1996, and 2003). Based on statistics (table 1), the number of the days exceeding 0${}^\circ$ C for daily mean air temperature with high cloud cover associated with the southerly warm advection (discussed later) was around three weeks in warm Decembers, whereas in cold Decembers this number was very limited. The difference of daily mean air temperature between warm and cold Decembers is about 10${}^\circ$ C, suggesting that southerly warm advection prevailed during warm Decembers.

Zoom In Zoom Out Reset image size

Table 1.  Monthly mean meteorological statistics at Bear Island for warm and cold Decembers. T: mean air temperature, N_day: number of days warmer than 0${}^\circ$C, CC: cloud cover.

	T (${}^\circ$C)	Nday	CC (%)
Warm Decembers	-1.3	20	80.8
Cold Decembers	-11.2	4	54.0

To compare broad atmospheric circulation patterns, we turn to the difference map of SLP and T950 by subtracting the composites of cold Decembers from those of warm Decembers (figure 2(a)). An anticyclonic anomaly is clearly seen over northern Eurasia. A strong pressure gradient straddles the western Barents Sea and Svalbard, indicating southerly wind anomalies. A positive anomaly of meridional sea-ice drift (southerly component) during warm Decembers likely caused the decrease in sea-ice cover (figure 1(b)). Warming over the Barents Sea sector, which has a pattern reported in previous studies (e.g., Honda et al 2009, Inoue et al 2012), would result from change in horizontal temperature advection in the lower troposphere, but not from change in surface heat flux (figure 3(a)). Although a limited positive LHF anomaly is found along the marginal ice zone of the Barents Sea, a broad negative LHF anomaly extending from that sea to the Norwegian Sea (figure 3(b)) supports the idea that the contribution of sea-ice retreat to the warm anomaly there is small (the horizontal distribution of sensible heat flux is similar to that of LHF). Therefore, it seems that the sea-ice retreat over the Barents Sea is under the control of atmospheric circulation over the Norwegian Sea and upstream region (e.g., North Atlantic sector). The cold anomaly over Eurasia (figure 2(a)) is similar to the pattern found in previous studies (Inoue et al 2012, Outten et al 2013).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Positive anomalies of precipitation from the southeast of Greenland to the Barents Sea and negative anomalies over northern Eurasia are consistent with the poleward shift of storm tracks represented by the eight-days high-pass filtered eddy heat flux ($\overline{v^{\prime} T^{\prime} }$) averaged between 1000 hPa and 850 hPa (contours in figure 3(c)). Because the cyclones tend to shift poleward (Inoue et al 2012) with less sea-ice coverage over the Barents Sea, it is natural that sea-ice motion contributes to its retreat (figure 1(b)).

According to 300 hPa level geopotential height (Z300) anomalies (figure 2(b)), a significant wavelike anomaly from the Barents Sea to Eurasia suggests that stationary Rossby waves are excited around that sea (Honda et al 2009). However, this wave train appears to originate from the east coast of North America, in particular the Gulf Stream and North Atlantic regions, in association with the propagation of wave-activity flux (Takaya and Nakamura 2001) which represents propagation of quasi-stationary Rossby waves (indicated by arrows in figure 2(b)). This pattern has some similarities to the Scandinavian pattern that consists of a primary center of action around the Scandinavian Peninsula, with two other centers of action with opposite sign over the northeastern Atlantic and central Siberia. The importance of SST anomalies over the North Atlantic has been discussed to explain subtle seasonality in the pattern (Bueh and Nakamura 2007).

It has been investigated whether the sharp SST gradient over the Gulf Stream has great impact on tropospheric circulation through diabatic heating over the surface wind convergence zone (Minobe et al 2008). In that study, annual heating over the Gulf Stream was shown to contribute significantly to the westerly jet from North America to Eurasia (figure S5 in Minobe et al (2008)). In winter, strong air-mass modification occurs over the Gulf Stream because of continental cold air (e.g., Bane and Osgood 1989), resulting in a shallow-heating mode there (Minobe et al 2010). Therefore, the possible influence of the SST anomaly over the Gulf Stream on atmospheric circulation over the downstream region would be worth investigating in detail. For example, there is a significant relationship between SST over the Gulf Stream and air temperature in the lower troposphere over the Barents and Norwegian Seas (figure 1(c)). Furthermore, there is a negative correlation over central Eurasia, suggesting that variability of the Gulf Stream influences both areas directly and/or indirectly.

We then focused on the difference of SST over the Gulf Stream between warm and cold Decembers at Bear Island. The composited SST in warm Decembers was clearly higher than that in cold Decembers, especially over the SST front (figure 4(a)). Interestingly, the surface heat flux anomaly (e.g., LHF) was not coherent with the horizontal SST difference pattern (figure 4(b)). A negative flux anomaly (−80 W m⁻²) was prominent along the eastern US coast, partly because the easterly wind anomaly prevented the cold air mass over the North America from blowing off (vectors in figure 4(b)). The meridional distribution of SST around 41${}^\circ$ N (figure 4(d)) agrees with the warming trend from 1950–2008 (Wu et al 2012). Both the poleward shift of SST front but also the increase of SST gradient from 2.9${}^\circ$ C 100 km⁻¹ in cold Decembers to 3.3${}^\circ$ C 100 km⁻¹ in warm Decembers would modify the storm track (contours in figure 3(c)). An upward motion anomaly was anchored over the SST anomaly and was strongest between 900 hPa and 700 hPa (figure 4(c)). This likely caused a difference in formation of clouds and precipitation over the downstream region (figure 3(c)) and consequent condensational heating in the lower troposphere. On the other hand, there was a remarkable downward motion anomaly at the same level over the southern region from 34${}^\circ$N–38${}^\circ$N. This anomaly was not statistically significant, partly because of large deviation of the center of action of downward motion among warm Decembers. This descending flow is consistent with reduced precipitation associated with the poleward shift of storm tracks (figure 3(c)). Therefore, a possible reason for the wave train originating from the Gulf Stream region (figure 2(b)) is the poleward shift of the SST front and change of clouds and precipitation.

Zoom In Zoom Out Reset image size

To elucidate the impact of the Gulf Stream heating anomaly on the atmospheric circulation from the North Atlantic to Eurasia, steady atmospheric response was calculated by the LBM. figure 5(a) shows the vertically averaged difference in $Q1$ from sigma levels 0.9–0.3 between warm and cold Decembers. Although $Q1$ includes sensible heat and radiative effects, condensational heating and evaporative cooling are usually dominant in the troposphere. In fact, the distribution over the Gulf Stream is similar to that of precipitation (figure 3(c)), suggesting that the difference in condensational heating was a major factor for the $Q1$ difference.

Zoom In Zoom Out Reset image size

When the $Q1$ anomaly enclosed by dashed lines in figure 5(a) was entered in the LBM, a strong anticyclonic anomaly developed over the Barents Sea, resulting in a pair of warm and cold anomalies over the Norwegian Sea and Russian coast (figure 5(b)). This pattern is partly similar to its observed counterpart (figure 2(a)), namely the warm anomaly over the eastern US coast and cold anomaly over Eurasia. The experiment forced by the negative $Q1$ anomaly only over the Norwegian and Barents Seas (figure 5(c)) suggests that the strong response shown in figure 5(b) came from the cooling anomaly over the Norwegian Sea (-3 K d⁻¹; figure 5(a)). This localized response, however, suggests that the steady response to the heating anomaly does not propagate to the central Eurasia. The other process related to nonlinearlity in the dynamical atmosphere (e.g., blocking), which is not treated in the LBM, would influence the downstream temperature anomaly over Eurasia. When the pair of positive and negative $Q1$ anomalies resulting from the poleward shift of storm track was only over the Gulf Steam (area enclosed by solid lines in figure 5(a)), positive SLP and T950 responses were excited over the Gulf Stream and eastern Europa regions (figure 5(d)). They are similar to figure 5(b), although the amplitudes of T950 and SLP are smaller than observations. If the background state is changed from the climatology to the mean state for warm years, as is perhaps more appropriate for the tests being undertaken here, the SLP response increased by over 0.1 hPa over the North Atlantic and Scandinavian regions. This anomalous atmospheric pattern induces southerly wind over the Norwegian Sea, enhancing poleward ice advection as seen in figure 1(b). Further, the cold anomaly over Eurasia was influenced by the Gulf Stream forcing. Therefore, the warm Arctic and cold Siberian pattern was triggered by the change in diabatic heating over the Gulf Stream. From the LBM response, the Barents Sea region is an amplifier of the Gulf Stream remote response.