Characteristics of the Arctic Ocean’s Beaufort Sea high are examined using fields from the NCEP–NCAR reanalysis. At a 2-hPa contour interval, the Beaufort Sea high appears as a closed anticyclone in the long-term annual mean sea level pressure field and in spring. In winter, the Beaufort Sea region is influenced by a pressure ridge at sea level extending from the Siberian high to the Yukon high over northwestern Canada. As assessed from 6-hourly surface winds, the mean frequency of anticyclonic surface winds over the Beaufort Sea region is fairly constant through the year. While for all seasons a strong closed high can be interpreted as the surface expression of an amplified western North American ridge at 500 hPa, there is some suggestion of a split flow, where the ridge linked to the surface high is separated from the ridge to the south that lies within the main belt of westerlies. The Aleutian low in the North Pacific tends to be deeper than normal when there is a strong Beaufort Sea high. In all seasons but autumn, a strong Beaufort Sea high is associated with positive lower-tropospheric temperature anomalies covering much of the Arctic Ocean; positive anomalies are especially pronounced in spring. Seasons with a weak anticyclone show broadly opposing anomalies. A strong high is found to be a feature of the negative phase of the summer northern annular mode, the positive phase of the Pacific–North American wave train, and, to a weaker extent, the positive phase of the summer Arctic dipole anomaly and Pacific decadal oscillation. The unifying theme is that, to varying degrees, the high-latitude 500-hPa ridge associated with the Beaufort Sea high represents a center of action in each teleconnection pattern.
A prominent feature of the annual mean sea level pressure (SLP) field for the Arctic Ocean is an anticyclone centered north of Alaska, often referred to as the Beaufort Sea high (BSH). In the annual mean, the BSH appears as a closed surface high embedded within a pressure ridge extending from northeastern Eurasia into northwest Canada. The surface wind field associated with the BSH, in conjunction with winds associated with the trough of low pressure that extends from the Icelandic low into the eastern Arctic, largely controls the mean circulation of the Arctic sea ice cover (Thorndike and Colony 1982). This circulation is characterized by the anticyclonic Beaufort gyre and a transport of ice from the Siberian coast, across the pole and into the North Atlantic, known as the Transpolar Drift Stream (Fig. 1) (Thorndike and Colony 1982; Colony and Thorndike 1984).
As is widely known, end-of-summer (September) Arctic sea ice extent has declined over the past few decades. September 2007 saw the lowest ice extent of the modern satellite era (Stroeve et al. 2008). The ice cover is also thinning (Nghiem et al. 2006; Maslanik et al. 2007b; Kwok and Rothrock 2009). Simulations from coupled global climate models used in the Intergovernmental Panel on Climate Change Fourth Assessment Report that include observed increases in atmospheric greenhouse gas concentrations consistently show declining September ice extent over the period of observations (Stroeve et al. 2007; Zhang and Walsh 2006). However, viewed as a group, simulated trends are conservative compared to observations (Stroeve et al. 2007). Many factors may be contributing to rapid ice loss. These include Arctic warming linked to increased concentrations of black carbon aerosols (Shindell and Faluvegi 2009); increased spring cloud cover, enhancing the downward longwave radiation flux at the surface (Francis and Hunter 2006); and altered ocean heat transport (Polyakov et al. 2005; Shimada et al. 2006). However, as shown in numerous studies (e.g., Rigor and Wallace 2004; Ogi and Wallace 2007; Wang et al. 2009; see section 2) variability in the atmospheric circulation, including the strength and location of the BSH, has played an especially prominent role.
The present paper examines characteristics and variability of the BSH, using data from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) Reanalysis I (Kalnay et al. 1996). It is motivated by recognition that while differing points of view have developed regarding the role of atmospheric circulation anomalies on sea ice conditions (see section 2), they can find some common ground through recognition that the BSH projects to varying degrees onto several atmospheric modes. We stress that we are not conducting a study of atmosphere–sea ice interactions. Our focus is rather on variability of an atmospheric feature recognized as highly relevant to the sea ice cover.
Section 2 provides an overview of known links between sea ice and the BSH. After evaluating mean seasonal expressions of the BSH in section 3, attention turns to characteristics of the large-scale midtropospheric circulation that favor a strong or weak surface high (section 4). These analyses provide context for assessing how the strength and location of the BSH and associated temperature anomalies are expressed with respect to the phase of atmospheric teleconnection patterns highlighted by other authors in the context of declining Arctic sea ice extent. Our analysis focuses on the period 1979–2008. While the NCEP–NCAR fields are available back to 1948, those from 1979 onward correspond to the modern satellite era and are therefore of higher quality. Use is made of surface winds, sea level pressure, 500-hPa height, and 925-hPa temperature. As shown in past studies (e.g., Serreze et al. 2009), basic circulation and tropospheric temperature features from NCEP–NCAR are very similar to those from other reanalyses. The 925-hPa temperatures are preferred over 2-m temperatures, which are strongly influenced by the modeled surface energy budget.
Interest in links between declining September sea ice extent and atmospheric circulation started in the mid- to late 1990s (e.g., Serreze et al. 1995; Maslanik et al. 1996) and thereafter grew quickly. Rigor et al. (2002) and Rigor and Wallace (2004) provided important insight on influences of the northern annular mode (NAM) in winter. The NAM, also known as the Arctic Oscillation, can be viewed as an oscillation of atmospheric mass between the Arctic and middle latitudes. It is in its positive phase when the zonally averaged surface pressure is high in midlatitude pressures and low in Arctic latitudes (Thompson and Wallace 1998, 2000). The North Atlantic Oscillation (NAO), which relates to covariability in the strengths of the Icelandic low and Azores high, is often viewed as the North Atlantic component of the NAM. From about 1970 through the mid-1990s, winter indices of the NAM shifted from negative to strongly positive. Rigor et al. (2002) show that as the winter NAM shifted toward the positive state, there was a retreat of the BSH to the southern Beaufort Sea, a more cyclonic motion of ice, and an enhanced transport of ice away from the Siberian and Alaskan coasts, fostering openings in the ice cover. While open water in coastal areas quickly refroze in response to low-surface air temperatures, these regions were nevertheless left with an anomalous coverage of young, thin ice in spring, especially prone to melting out in summer. The strongly positive NAM phase characterizing the period 1989–95 saw strong transport of thick, multiyear ice out of the Arctic and into the North Atlantic through the Fram Strait. While the NAM subsequently regressed to a more neutral state, the sea ice system may still have memory of these thinning processes (Rigor and Wallace 2004).
Other studies have focused on aspects of the summer circulation, and it is in this season that impacts of variability in the BSH are especially prominent. Ogi and Wallace (2007) find that years with a low September sea ice extent tend to be preceded by anticyclonic summer (July–September) anomalies in the SLP field over the Arctic Ocean, with the core of the anomaly centered at about 85°N, 210°E, hence somewhat north of the location of the BSH in the mean annual field (Fig. 1). Years with high September ice extent have the opposing anomaly pattern. They view the atmospheric link largely in terms of Ekman drift in the marginal seas, with cyclonic winds leading to sea ice divergence, spreading ice over a larger area, and anticyclonic winds promoting ice convergence, compacting the ice into a smaller area. The anticyclonic anomaly pattern is also linked to positive anomalies in surface air temperature over the regions of reduced sea ice. The anticyclonic–cyclonic SLP anomaly pattern is, in turn, similar to the pattern of the summer NAM as defined by Ogi et al. (2004). In their analysis, the NAM is defined separately for each calendar month through an empirical orthogonal function (EOF) analysis of NCEP geopotential height fields from 1000 to 200 hPa, poleward of 40°N. This contrasts with the approach of Thompson and Wallace (2000), in which the NAM is based on a single EOF analysis of geopotential height fields for all calendar months. While by their approach, the NAM is most strongly expressed in winter, the NAM as defined by Ogi et al. (2004) is strong throughout the year, albeit with seasonal differences in structure, and with the summer center of action lying over the central Arctic Ocean.
The record September sea ice minimum of 2007 fostered a series of papers focusing on the summer atmospheric circulation. The key feature of summer 2007 was an unusually strong Beaufort Sea high, paired with unusually low pressure over northern Siberia (Figs. 2a,b). Resulting southerly winds between the pressure anomaly centers promoted transport of ice away from the coasts of Siberia and Alaska toward the North Pole (Ogi et al. 2008) as well as strong melt in the East Siberian and Chukchi Seas associated with the strongly positive air temperature anomalies in this region (Fig. 2c) (Stroeve et al. 2008). Unusually clear skies in the vicinity of the strong Beaufort Sea high may have enhanced surface and basal melt (Kay et al. 2008), although the importance of cloud forcing has been questioned (Schweiger et al. 2008). The pattern also led to a strong surface pressure gradient across Fram Strait (between Greenland and Svalbard) enhancing wind-driven transport of sea ice out of the Arctic Ocean and into the North Atlantic (Wang et al. 2009). The general view is that while this atmospheric pattern was key in driving rapid summer ice loss, its effectiveness was enhanced by the extensive coverage of thin, first-year ice in spring 2007—a reflection of the ongoing thinning of the ice cover noted earlier (e.g., Ogi et al. 2008; Stroeve et al. 2008). A similar, albeit less well-developed, atmospheric pattern dominated during the summer of 2008, as well as for most of the summer of 2009.
Wang et al. (2009) performed an EOF analysis of seasonal mean SLP for the region north of 70°N. They define the first EOF mode as the NAM, which has a seasonal structure very similar to that shown by Ogi et al. (2004). The second mode, termed the Arctic dipole anomaly (DA), has two centers of action of opposing sign—a pattern that has been remarked upon in various contexts in past studies (e.g., Overland and Wang 2005; Maslanik et al. 2007a; Stroeve et al. 2008). Focusing on the summer season, they conclude that while the NAM largely affects ice extent through Ekman drift processes, as discussed by Ogi and Wallace (2007), the DA, with centers of action centered over the northern Beaufort Sea and the Kara Sea, is associated with an anomalous meridional wind pattern. The positive DA phase, characterized by a strong, northward-shifted BSH and negative anomalies centered over the Kara Sea, leads to an anomalous wind component blowing across the Arctic Ocean that favors flushing of sea ice out of the Arctic Ocean through Fram Strait and, while not discussed in that paper, promotes warm southerly wind anomalies over the East Siberian Sea. In this framework, the record seasonal sea ice minimum of September 2007, the second-lowest minimum of September 2008, and the extreme minimum of September 2005 are consistent with the combination of a positive summer mode of the DA and a negative mode of the summer NAM. This combination was especially prominent in summer of 2007, with its effects enhanced by the anomalous spring coverage of thin first-year ice. Wang et al. (2009) also present evidence that the positive DA with its strong BSH tends to promote the import of warm ocean waters from the Pacific through the Bering Strait favoring melt, and that this effect can also help to explain the record 2007 sea ice minimum and possibly that of other recent summers.
L’Heureux et al. (2008) provide yet another perspective on summer 2007 and the role of the BSH. This is best discussed with the aid of Fig. 3, which provides the summer 2007 mean 500-hPa height and anomaly fields down to 40°N. The anomalous mean BSH of summer 2007 (Fig. 2a), which in the framework of Wang et al. (2009) is part of the summer dipole anomaly pattern, is seen to be located just downstream of an anomalously strong 500-hPa ridge favoring anticyclonic vorticity advection increasing with height and downward motion. This 500-hPa pattern is viewed by L’Heureux et al. (2008) as an expression of an extreme positive phase of the Pacific–North American (PNA) wave train—three standard deviations above the 1950–2007 summer mean (July to September in their analysis). Recall that the winter PNA index describes the amplitude of the wave train spanning the North Pacific Ocean (mean trough) through northwestern North America (mean ridge) to northeastern North America (mean trough) (Wallace and Gutzler 1981; Barnston and Livezey 1987). As discussed by L’Heureux et al. (2008), the summertime PNA wave train tends to be shifted north of its winter position, and with a ridge located in the Beaufort Sea region.
To summarize: 1) Variability in the high-latitude atmospheric circulation is identified as a contributor to the observed downward trend in September sea ice extent and recent extreme seasonal minima; 2) while this variability has been examined in several frameworks leading to differing interpretations, a common thread between different studies is that an important role is played by variability in the strength and location of the BSH. The following sections assess variability in the BSH and relationships with variability in the Arctic-wide and larger Northern Hemisphere circulation.
3. Mean seasonal cycle
We start by examining mean seasonal cycles of anticyclonic and cyclonic surface winds across the Arctic, how conditions across the Arctic compare to those in the Beaufort Sea region, and how spatial patterns in frequencies of cyclonic and anticyclonic flow are reflected in the distributions of SLP. We use the frequency of negative relative vorticity as an index of the occurrence and strength of the BSH. Relative vorticity, in representing a measure of rotation of the wind field, is a better index of anticyclonic conditions than sea level pressure.
Surface u and v winds from the NCEP–NCAR reanalysis were used to compute relative vorticity every 6 hours at all grid locations poleward of 60°N for the time period 1979–2008. The 6-hourly NCEP–NCAR data are provided on a 2.5° × 2.5° latitude–longitude grid. The panels constituting Fig. 4 depict the winter season (December through February) percent frequency of negative (anticyclonic) (Fig. 4a) and positive (cyclonic) relative vorticity events at the surface by grid cell (Fig. 4b), and the mean SLP (Fig 4c). For a given grid location, anticyclonic (negative vorticity) frequency is defined as the percentage of all 6-hour time periods (4 analyses per day times the 90 days constituting the winter season times the 30 years analyzed) that the relative vorticity was more negative than the lowest 25th percentile of all negative vorticity events for winter, as assessed for the region poleward of 60°N. Cyclonic vorticity frequency was calculated in the same way using the number of positive vorticity events with magnitude larger than the 25th percentile value of all positive vorticity events. The screening eliminates the weak vorticity cases. All results that follow are based on these thresholds. Maps based on different thresholds (e.g., the 50th percentile) have very similar spatial patterns.
While the frequency of anticyclonic events in the Beaufort Sea region stands out as a relative peak (50%–60%), frequencies are higher over northeastern Siberia. Fairly high frequencies are also found over parts of northern Canada. Peak frequencies over the central Greenland ice sheet (>90%) should be viewed with extreme caution because of problems in reducing surface pressures over the cold, high-elevation ice sheet to sea level (Serreze et al. 2001). The spatial pattern of anticyclonic wind frequency is very similar to the winter frequency pattern of closed high pressure cells as shown by Serreze et al. (1993), based on an automated detection algorithm applied to daily SLP fields for 1952–89. It follows that regions of anticyclonic wind frequency maxima stand out as a relative minima in the frequency of cyclonic winds. The cyclonic wind frequency pattern in Fig. 4b is similar to the frequency pattern of closed lows at the surface shown by Serreze et al. (1993, 1997), Zhang et al. (2004), and others. High frequencies in the North Atlantic are the expression of cyclone activity along the primary North Atlantic storm track. Baffin Bay, by contrast, is known as a region where cyclones tend to stall and dissipate, hence the particularly high winter frequencies in this region (locally >80%). The spatial patterns of anticyclonic and cyclonic flow are consistent with the mean seasonal distribution of sea level pressure. At a 2-hPa contour interval adopted for Fig. 4c, there is no closed Beaufort Sea high in the winter mean field. Rather the East Siberian, Chukchi, and Beaufort Seas are overlain by a ridge of high pressure. The highest pressures over Eurasia are part of the well-known Siberian high. A closed BSH does appear when the data are mapped using a finer 1-hPa contour interval.
Corresponding maps for spring (March–May; Fig. 5) document a reduction in the frequency of anticyclonic winds over Siberia compared to winter, with the Beaufort Sea region still appearing as a relative peak (50%–60%). Cyclonic wind frequency is reduced compared to winter in the Atlantic sector. In accord with weakening of the Siberian high and its poleward extension, the mean SLP field for spring shows a pronounced closed Beaufort Sea high with a peak central pressure of about 1022 hPa.
Patterns for summer are quite different from winter and spring (Fig. 6). The local maximum in anticyclonic winds in the Beaufort Sea seen in winter and spring is shifted to the south. The distribution of anticyclonic wind maxima is consistent with the frequency maxima of closed summer highs depicted in the earlier analysis of Serreze et al. (1993). The area around the pole shows up as a weak relative maximum in cyclonic winds. The positive and negative vorticity frequency patterns are expressed in the mean SLP field as a weak (1015 hPa) Beaufort Sea high (closed only for a 1-hPa contour interval) with a more limited spatial extent compared to its spring counterpart, paired with an area of weak mean low pressure centered just off the North Pole. This mean low has long been recognized (e.g., Reed and Kunkel 1960) and reflects the summer maximum in cyclone activity in this region. These systems variously form within the central Arctic Ocean or migrate into the region from Eurasia (Serreze and Barrett 2008). Manifesting an equivalent barotropic structure, the weak mean low lies almost directly underneath the center of the summer 500-hPa polar vortex (not shown).
The maps for autumn (Fig. 7) document the transition back toward winter conditions. Peak frequencies of anticyclonic winds in the Beaufort Sea (again 50%–60%) have shifted north of their summer location, and the frequency of cyclonic winds has increased in the Atlantic sector. The Beaufort Sea high region is again part of a ridge (a closed high only apparent at a 1-hPa contour interval) and the Siberian high has started to rebuild.
Fig. 8 provides monthly means and standard deviations of negative and positive vorticity frequency, along with monthly mean SLP based on aggregating data over the region bounded by latitudes 72.5°–80.0°N and longitudes 180.0°–225.0°E (see Fig. 7c). This region encompasses the center of the BSH region as it appears in the vorticity frequency maps. Figure 9 shows, for the same region, monthly means, 5th, 25th, 75th, and 95th percentile values of vorticity magnitude. While there is only a weak seasonal cycle in anticyclonic frequency in the BSH region, there is a 10-hPa range in mean SLP, from a maximum of 1021 hPa in March to a minimum of 1011 hPa in August. The August pressure minimum manifests the August maximum in the frequency of positive vorticity events (which tend to correspond to low-pressure systems), the southward shift in summer of maximum vorticity frequency (cf. Figs. 4a –6a), and a more general mass transfer out of the Arctic Ocean in summer. The latter issue was examined by Cullather and Lynch (2003). They show that the increase in surface pressure over the central Arctic (including the BSH region) from winter into spring is linked to a poleward mass transport from Eurasia. An increase in equatorial transport over the Canadian Arctic Archipelago in May and June is manifested as a decrease in surface pressure over the Arctic Ocean into summer. This pattern then reverses in autumn, with transport from the Canadian Arctic Archipelago into Eurasia.
It is useful to contrast March (pressure maximum) and August (pressure minimum) with respect to the distribution of omega (vertical motion). The plots in Fig. 10 show omega (Pa s−1, positive values meaning downward motion) along latitude 75°N from longitudes 90°E eastward to 90°W and vertically from 1000 to 100 hPa. The March SLP maximum in the BSH region is supported by a prominent region of downward (positive) omega centered along approximately 150°W, peaking at about 600 hPa (defining the approximate level of nondivergence), consistent with the location of the BSH and its associated divergent surface wind field ahead of a mean ridge at 500 hPa. The BSH pressure minimum in August is associated with a much weaker region of downward omega, with the maximum shifted east to about 120°W, and peaking closer to the surface. Links between interannual variability in the BSH and the location and strength of the midtropospheric flow are discussed in the next section.
4. Interannual variability and composite analyses
Standardized anomalies in the frequency of anticyclonic vorticity events over the Beaufort Sea subregion by year and season are plotted in Fig. 11. Anomalies are computed with respect to frequencies for the period 1979–2008. There are no statistically significant (at the 10% level) trends in any season. Anomalies have been positive for the last five years of the analyzed record (2004–08) in autumn and for four of the past five years in summer. The most positive frequency anomalies for both summer and autumn occurred in 2007.
Fig. 12 shows maps of correlation coefficients between the standardized anomalies in the frequency of anticyclonic vorticity events in the BSH region as we have defined it and fields of seasonal mean sea level pressure for the period 1979–2008. Coefficients are only shown where they are significant at the 5% level. Reinforcing the validity of the frequency of anticyclonic relative vorticity as an index of the strength of the BSH, strong positive correlations (>0.6) are located over the Beaufort Sea. In winter, spring, and autumn, vorticity in the BSH region is also negatively correlated with SLP in the North Pacific corresponding broadly to the Aleutian low region. In spring, a strong BSH also relates to lower pressures over western Eurasia. In summer and autumn, pressures in the BSH region are furthermore inversely correlated with those over Eurasia and (in summer) Europe and the Sea of Okhotsk.
Variations in the strength of the BSH are associated with distinctive signatures in large-scale atmospheric circulation and temperature fields. The composite fields constituting Fig. 13 of SLP anomalies, 500-hPa height, anomalies of 500-hPa height, and anomalies of 925-hPa temperature summarize the situation for winter months with a strong BSH. Strong BSH winter months are defined as those for which the monthly negative vorticity frequency for the BSH region falls into the highest 20% of values for the 90 winter months included in the 30-yr (1979–2008) record. The composites therefore represent the mean of 18 months; anomalies are with respect to winter means over the period 1979–2008.
Positive SLP anomalies in the Beaufort Sea (Fig. 13a), peaking at about 5 hPa, are associated with a closed anticyclone with a central pressure of 1022 hPa (not shown). The strong BSH lies east of a pronounced ridge in the 500-hPa flow (Figs. 13b,c). While viewed most simply as a high-latitude expression of an amplified western North American ridge, there is also suggestion of a regional split flow—note how the ridge with fairly tight height gradients linked to the anomalously strong BSH is separated by a region of slack height gradients from the ridge to the south associated with the main belt of westerlies. Separation is further evident in the much more pronounced 500-hPa height anomalies linked to the ridge in the Beaufort Sea region compared to the ridge at lower latitudes. Consistent with the correlation analyses just presented, the strong BSH at sea level and its associated 500-hPa ridge occur in conjunction with negative 500-hPa height and SLP anomalies over the North Pacific indicative of an anomalously deep Aleutian low. Linear correlations between the monthly BSH vorticity index and the minima of monthly mean pressures in the region 40°–60°N and 160°E–160°W; an index of the Aleutian low strength used by Overland et al. (1999) further supports this inverse relationship.
Positive pressure anomalies are found in the Icelandic low region (again, see Fig. 12), and negative pressure anomalies extend across the northern North Atlantic. Consistent with the 500-hPa anomaly structure, 925-hPa temperatures are higher than normal over northwestern North America and the western Arctic Ocean. Anomaly patterns for the case of a weak BSH (not shown), based on the winter months for which the negative vorticity frequency fell into the lowest 20% of the distribution, largely mirror those seen in the positive composites; positive pressure anomalies extend across the northern North Atlantic and 925-hPa temperatures are cooler over western Arctic Ocean and northwestern North America. There is no evidence of a regional split flow at 500 hPa and height gradients over the Arctic Ocean are quite slack. This mirroring of anomaly patterns in composites for a strong versus weak BSH is evident in all seasons.
Patterns associated with a strong BSH in spring (Fig. 14) share some similar features with the winter expressions. As with winter, positive SLP departures over the Beaufort Sea (reflecting a closed anticyclone with a central pressure of 1024 hPa) are paired with an anomalously deep Aleutian low. Positive pressure anomalies are found in the Icelandic low region. Negative SLP anomalies also characterize much of northern Eurasia (see Fig. 12). The amplified 500-hPa ridge associated with the strong BSH, with its axis roughly along the date line, is again somewhat separated from the amplified western North American ridge. While like winter when positive 925-hPa temperature anomalies are found over northwestern North America and the western Arctic Ocean, positive anomalies now also cover almost all the plotted region north of 40°N.
Plots for summer months with a strong BSH follow in Fig. 15. The dominant feature of the SLP anomaly field is the pairing of positive departures over the Beaufort Sea (linked to a closed anticyclone with a central pressure of 1016 hPa) with a weaker-than-normal Icelandic low. Weak negative anomalies characterize lower latitudes and are somewhat stronger over eastern Eurasia and part of the northern North Pacific. Similar to winter and spring, however, the anomalously strong BSH can be associated with a 500-hPa ridge that is separated from the western North American ridge to the south. The notable feature of the 925-hPa temperature anomaly field is the positive departures covering much of the western Arctic Ocean.
Composite fields for autumn months with a strong BSH are provided in Fig. 16. The dominant feature of the SLP field is the pairing of positive departures over the Beaufort Sea with a stronger-than-normal Aleutian low in the North Pacific. Note, however, that the SLP composite does not show prominent negative anomalies over eastern Eurasia, as would be expected from the correlation analysis presented in Fig. 12. The regional split flow at 500 hPa seen in winter, spring, and summer is again present. Interestingly, 925-hPa temperature anomalies over the domain from 40°N to the pole, including the Arctic Ocean, are mostly negative.
5. Links to atmospheric teleconnections
Summarizing the preceding discussion, a strong BSH in all seasons is linked to a pronounced ridge at 500 hPa. Except for autumn, a strong BSH is associated with positive lower-tropospheric temperature anomalies covering much or all of the Arctic Ocean. Seasons with a weak anticyclone show broadly opposing anomalies.
To build on these findings, we consider expression of the BSH with respect to the phase of atmospheric teleconnection patterns cited as having links with the observed decline in Arctic sea ice extent. These include the NAM, the Arctic DA, and the PNA wave train. For completeness, we also examine links with the Pacific decadal oscillation (PDO), which is known to have prominent climate signals in the North Pacific sector. We employ the same approach as above, except that composites are based on the top and bottom 20% of the normalized monthly index values for each pattern. Discussion focuses on the summer and winter seasons.
The NAM analysis uses monthly index values based on the work of Ogi et al. (2004) (http://wwwoa.ees.hokudai.ac.jp/people/yamazaki/SV-NAM/index.html).
Recall that they define the NAM separately for each calendar month. This contrasts with the approach of Thompson and Wallace (2000) who used a single EOF analysis for all calendar months. We calculate a monthly DA index following the approach of Wang et al. (2009) for their seasonal DA index, but using monthly rather than seasonal fields. This ensures that composites comprise the same number of cases for all indices. Recall that Wang et al. (2009) defined the DA as the second principal component of the seasonal mean SLP field north of 70°N. The first principal component is taken to be the NAM and has a seasonal structure very similar to that shown by Ogi et al. (2004). For the monthly DA index, following L’Heureux et al. (2008) (see below), the month of interest as well as the previous and following months are included in the EOF analysis. The first and second EOFs and time series of principal components for the monthly analysis compare well with the seasonal DA index time series of Wang et al. (2009).
However, the DA index used here needs to be treated with caution. The EOF analysis is performed over a limited area (north of 70°N), which can produce EOFs with patterns that are not related to underlying patterns in the original data (Buell 1979; Wilks 2006). The first four EOFs for each of the monthly analyses display patterns similar to those first described by Buell (1979); often referred to as one fried egg, two fried eggs, etc. Furthermore, inspection of the eigenvalue spectra for each of the months and variances accounted for by the second and third EOFs show that there is little separation between the second and third eigenvalues. In the monthly analyses, the leading EOFs account for between 37% and 65% of the variance. The second EOFs interpreted here as the DA account for between 10% and 22%. The third EOFs account for between 7% and 16%. Limited separation between eigenvalues indicates that the associated EOFs may represent an arbitrary mixture of the variance of the original data (Wilks 2006). Similar “fried egg” patterns and eigenvalue spectra are found for seasonal analyses as well.
The PDO (http://jisao.washington.edu/pdo/PDO.latest) is defined as the leading principal component of monthly anomalies of sea surface temperature for the northern Pacific Ocean (north of 20°N) (Mantua et al. 1997). The PNA time series, obtained from the NCEP Climate Prediction Center (http://www.cpc.ncep.noaa.gov) is the same as that used by L’Heureux et al. (2008) in their analysis of atmospheric circulation for the summer of 2007. A detailed description of the methods used to calculate the PNA is given at http://www.cpc.ncep.noaa.gov/data/teledoc/telepatcalc.shtml. Briefly, the 10 leading unrotated EOFs are calculated for each calendar month using standardized fields of monthly mean 500-hPa height anomalies for the region poleward of 20°N for the period January 1950 to December 2000. Instead of using data for just the month of interest, data for the two neighboring months are also used. For example, the EOFs for July are calculated using fields for June, July, and August. The 10 leading EOFs for each month are then subjected to a varimax rotation to produce 10 rotated EOFs and indices, one of which is the PNA. The NAO is one of the other indices. The PNA calculated using this method is viewed as an improvement over the three-point-based index (Wallace and Gutzler 1981) because it has better continuity between months and uses information from the whole flow field (L’Heureux et al. 2008).
Composite SLP anomaly fields for summer months with a strong BSH from the vorticity index appear in Fig. 17 along with corresponding fields for the negative phase of the NAM, the positive phase of the DA, the positive phase of the PNA, and the positive phase of the PDO. The composite anomaly field for the strong BSH and the negative phase of the summer NAM are highly similar—the primary difference is that the positive SLP anomalies over the Arctic Ocean are stronger and cover a larger area in the NAM composite. Of the 18 summer months represented in the strong BSH composite, 9 were associated with a negative summer NAM index value. The linear correlation between the two monthly index time series for summer months is statistically significant at the 5% level in June (−0.58) and July (−0.59) but not for August (−0.24). Time series of monthly indices for the NAM and BSH for June and July follow in Fig. 18.
There is little similarity between the strong BSH composite and the positive DA composite, for which the Arctic region is dominated by negative SLP anomalies along the Eurasian coast and weak positive anomalies over the western side of the Arctic; a pattern that, as discussed by Wang et al. (2009), promotes transport of ice out of the Arctic Ocean via the Fram Strait. The PNA composite anomaly field is similar to the summer NAM pattern in showing negative anomalies over the North Pacific and positive anomalies in the BSH region, but, in the PNA, the Beaufort Sea anomaly is much weaker. The positive PDO composite, by contrast, is associated with predominantly positive SLP anomalies over Arctic and subarctic latitudes, with a weak positive anomaly in the eastern BSH region extending into the Canadian Arctic Archipelago.
In summary, while a strong BSH based on the vorticity index is most closely allied with the negative phase of the NAM, there is some tendency for a stronger BSH during the positive phase of the DA, PDO, and particularly the PNA. Anomaly structures for the composites for the positive-phase summer NAM, and the negative phase of the DA, PNA, and PDO largely mirror those shown in Fig. 17. This mirroring largely holds with respect to remaining discussion in this section.
Turning to 500-hPa height fields (Fig. 19), the salient feature of the NAM composite is the more pronounced curvature in the streamlines in the Beaufort Sea sector as compared to the composites for DA, PNA, and PDO. This is seen as a strong positive anomaly in 500-hPa heights centered at about 85°N and along the date line, just west of the positive SLP anomaly (not shown). A weaker, albeit still prominent, positive height anomaly is associated with the positive PNA composite at about the same latitude, but slightly farther east. The negative NAM phase is associated with positive lower-tropospheric temperature anomalies over the western Arctic Ocean, but of smaller spatial extent than for the strong BSH composite or the DA (Fig. 20). During positive extremes of the PNA, essentially all of the Arctic Ocean is covered by positive temperature anomalies.
SLP anomaly patterns for the winter composites of the NAM, DA, PNA, and PDO bear little resemblance to their summer counterparts (Fig. 21) and none depict a strong BSH. While the BSH is clearly allied with the NAM in summer and, to a lesser degree, the PNA, the negative NAM composite instead depicts a large-scale pattern of compensating positive anomalies over all of the Arctic and negative anomalies in middle latitudes. This structure has, of course, been discussed in numerous studies (e.g., Thompson and Wallace 1998, 2000). In further sharp contrast to summer, the negative NAM phase is linked to positive 925-hPa temperature anomalies over much of the Arctic Ocean. The poor relationship between the NAM and BSH in winter is reflected in the linear correlation between the monthly indices for these patterns. Only January has a weak (−0.37) but statistically significant correlation at the 5% level. While the positive PNA phase in winter is dominated by positive high-latitude temperature anomalies, the spatial structures in winter and summer are quite different (Fig. 22).
6. Summary and discussion
As assessed using 6-hourly fields from the NCEP–NCAR reanalysis, the frequency of anticyclonic surface winds in the Beaufort Sea is fairly constant through the year. As a closed anticyclone (2-hPa contour interval) in the climatological mean SLP fields, a BSH is present only in the annual mean and in spring. In winter, the region is instead influenced by a pressure ridge extending from the Siberian high to the Yukon high over northwestern Canada. The mean summer SLP field in the BSH region is rather flat. For all seasons, a pronounced closed high at the surface is linked to a pronounced ridge at 500 hPa. While viewed most simply as a high-latitude expression of an amplified western North American ridge, there is also the suggestion of a regional split flow, in that the ridge with fairly tight height gradients linked to the anomalously strong BSH is separated by a region of slack height gradients from the ridge to the south associated with the main belt of westerly winds. Separation is further evident in the much more pronounced 500-hPa height anomalies linked to the ridge in the Beaufort Sea region compared to the ridge at lower latitudes. In all seasons but autumn, a strong BSH is associated with positive lower-tropospheric temperature anomalies covering much of the Arctic Ocean; positive anomalies are especially pronounced in spring. There are no obvious temporal trends in the strength of the BSH as measured by the frequency of anticyclonic winds in the region.
Variations in the strength of the summer BSH are clearly allied with the phase of the summer NAM. This is consistent with demonstrated links between the summer NAM and high-latitude extratropical cyclone activity. In a climatological sense, the frequency of extratropical cyclone centers over the central Arctic Ocean has a distinct summer peak. Summer activity is greatest at about 85°N along the date line (Serreze and Barrett 2008). This basic pattern has been recognized for many years (e.g., Dzerdzeevskii 1945; Reed and Kunkel 1960). Seasonal onset of the cyclone maximum is allied with a tendency for the summer 500-hPa circumpolar vortex to contract and become broadly symmetric about the pole, and with development of a region of high-latitude baroclinicity. Development of this high-latitude baroclinicity is driven at least in part by differential heating between the land and Arctic Ocean. The cold Arctic Ocean may also help to focus the center of the 500-hPa polar vortex over the pole. In turn, there is an eastward shift of the Urals trough, and the flow ahead of it becomes more zonal than in winter. Systems entering the central Arctic Ocean from the outside, or formed within the Arctic Ocean, migrate around the 500-hPa vortex and decay within the cyclone maximum region, or in close proximity (Serreze et al. 2001).
However, the strength of the cyclone pattern is highly variable. When well developed, the 500-hPa circumpolar vortex is particularly strong and symmetric about the pole, with negative SLP anomalies centered over the pole and positive anomalies over middle latitudes. The BSH is weak. This tends to occur in conjunction with the positive NAM phase. For summer months, when the cyclone pattern is weakly developed, the 500-hPa circumpolar vortex is weak and the flow is much more meridional, with a pronounced 500-hPa ridge over the northern Beaufort Sea and a strong BSH. This tends to occur in conjunction with the negative phase of the NAM.
Composite analyses presented here, as well as results from other studies, indicate that a stronger (weaker) than average BSH in summer months may also attend the positive (negative) phases of the Arctic dipole anomaly, the Pacific–North American teleconnection, and the Pacific decadal oscillation. At least in a composite mean sense, after the NAM, the strength of the BSH is most strongly linked to the phase of the PNA. Recall from earlier discussion that L’Heureux et al. (2008) linked the unusual circulation in summer 2007 that drove the record September sea ice minimum to an extreme positive (three standard deviation) phase of the PNA. While the NAM was also negative for June and July of 2007 (index values of −0.81 and −0.740), it was slightly positive in August (+0.29). The DA was, in turn, a positive mode.
The unifying theme, which provides some common ground to understanding results from past studies that have employed a variety of frameworks (NAM, PNA, DA) to diagnose links between atmospheric circulation and sea ice conditions, is that to varying degrees, the high-latitude 500-hPa ridge associated with the surface BSH represents a center of action in each teleconnection pattern: it is the dominant center of action for the NAM, weaker and one of several centers for the PNA, and weaker still for the DA and PDO. These differences in the strength and location of the 500-hPa center of action are of course reflected in the composites in the location and amplitude of the SLP anomalies. This contrasts with winter, when there are no clear relationships between the strength of the BSH and any of the teleconnection patterns examined. This follows in that as viewed individually, none have a 500-hPa center of action in the BSH region.
This study was supported by NSF Grants ARC-090162 and ARC 0805821.
Corresponding author address: Mark C. Serreze, National Snow and Ice Data Center, Cooperative Institute for Research in Environmental Sciences, Campus Box 449, University of Colorado Boulder, Boulder, CO 80309-0449. Email: email@example.com