Recent Increased Warming of the Alaskan Marine Arctic Due to Midlatitude Linkages

This paper provides an update on the major climate changes in the marine Alaskan Arctic over the past decade (Wood et al., 2013; Ballinger and Sheridan, 2014; Overland et al., 2014a; Cassano et al., 2015). At the decadal/regional scale of climate change, Fig. 1 shows the 13-month running mean surface air temperature (SAT) anomalies for Barrow, Alaska (red), and areal averages for the Northern Hemisphere land area (blue), relative to a baseline period of 1981-2010. Temperature anomalies at Barrow are predominantly positive since 2003, and in winter 2015-16 are roughly four times the magnitude of the Northern Hemisphere mean temperature increase. Alaskan Arctic waters participate in the hemispheric-wide warming of the Arctic over the last two decades. Additional warming of Alaskan Arctic coastal waters from autumn 2014 through autumn 2016 can be attributed to the shift to warm ocean temperature anomalies in the North Pacific and associated shifts to southerly winds over Alaska. Such North Pacific contributions subside after autumn 2016, but greenhouse gas (GHG) and regional Arctic feedback contributions remain.

Figure 1.  13-month running mean SAT anomalies for Barrow and Northern Hemisphere lands relative to their respective 1981-2010 means (from CRUTEM4; available at www.cru.uea.ac.uk/cru/data/temperature/).

Arctic air and ocean surface temperature increases coincide with the expansion of sea-ice-free areas, increases in the mobility of sea ice, shifts in ocean currents, and biological impacts at all trophic levels from primary productivity increases through loss of walrus habitat (Wassmann, 2015). In the following sections, we track atmospheric changes in the Alaskan Arctic, compare them relative to the Bering Sea, assess future climate projections, and address Alaskan Arctic temperature increases since August 2014 due to warming in the North Pacific.

Figure 2.  Mean (1961-2010) near-surface temperature (units: ^°C) for the four seasons over the western Arctic. Data are from the NCEP-NCAR Reanalysis via NOAA/ESRL, generated online at http://www.esrl.noaa.gov/psd/cgi-bin/data/composites/printpage.pl. Figure is similar to Fig. 2.2 in Overland et al. (2014).

Figure 3.  Mean (1961-2010) SLP (units: hPa) for the four seasons over the western Arctic. Data are from the NCEP-NCAR Reanalysis via NOAA/ESRL, generated online at http://www.esrl.noaa.gov/psd/cgi-bin/data/composites/printpage.pl. Figure is similar to Fig.2.3 in Overland et al. (2014).

The Pacific Arctic discussed here is defined as regions north of 66^°N, which covers the area north of the Bering Strait and the southern Chukchi and Beaufort Seas. Climatologically, this region is on the northern side of the transition zone between the relatively warm and moist storm tracks of the Aleutian low weather system reaching into the Bering Sea, and the colder, drier, and higher-pressure Arctic air mass to the north. As summarized in (Overland et al., 2014a): "Located in the southern part of the Pacific Arctic is a region of large north-south gradients in atmospheric properties such as near-surface air temperature (Fig. 2) and atmospheric sea level pressure (SLP, Fig. 3). This region of strongest gradients moves north and south with the seasonal cycle. Maximum temperature gradients in winter are located over the central Bering Sea with sub-freezing temperatures and extensive sea-ice coverage. In summer, the greatest air temperature gradients are found across the southern Chukchi Sea and seaward of the coast of Alaska with SAT above-freezing. Large north-south gradients in SLP produce a vast east-west trending region of strong climatological winds from the east across a relatively narrow band of latitudes in all seasons but summer." The Aleutian low center to the south of the Alaskan Arctic is a dominant feature shown in the climatology of SLP plots (Fig. 3) in all seasons except summer.

From 2007 through mid-2014, the Pacific air mass to the south and the Arctic air mass to the north are on different trajectories. To the north, the Chukchi/Beaufort Sea region is part of the decadal change of Arctic warming where recent sea-ice and snow losses are allowing extensive areas to absorb more late-spring and summer solar radiation than in the past, and are changing the atmospheric climatology of the region with positive temperature anomalies extending throughout the year (i.e. Arctic amplification, AA). Figure 4 (top) shows monthly SAT at Barrow, in which it is notable that there is an observed shift to positive temperature anomalies beginning in 1995. To the south, the Bering Sea, represented by Saint Paul SAT (Fig. 4, bottom), turns colder with extensive seasonal sea-ice cover in 2007-13, which has not been observed since the mid-1970s. This period contrasts an earlier warmer-than-normal Bering Sea temperature anomaly period for the southern Bering Sea from 2000 through 2006. Beginning in 2014, the Bering Sea returned to consistent warm anomalies, tied to ocean temperature changes in the greater North Pacific. While this short-term warming event ends in autumn 2016, lower tropospheric air temperatures continue to remain above-normal in the Alaskan Arctic. We return to discussing this latest North Pacific Ocean impact on the Pacific Arctic in section 5.

Figure 4.  Monthly SAT anomalies for Barrow and Saint Paul, Alaska, compared to their respective 1981-2010 mean values. Anomalies are based on NWS weather station data.

Arctic-wide average surface temperatures have increased at double the rate of global mean temperatures——a well-documented phenomenon referred to as AA (Holland and Bitz, 2003; Serreze and Barry, 2011). Figure 5 shows the difference in mean annual Northern Hemisphere lower tropospheric air temperatures for 2010-14 relative to the end of the 20th century (1971-2000). Although the entire Northern Hemisphere polewards of 40^°N has witnessed positive changes in annual mean temperatures in recent years, much of central Arctic shows increases of at least +2^°C. Note that the southeastern Bering Sea does not show a change in recent temperatures relative to the late 20th century. The spatial pattern of AA (Fig. 5) does not resemble the temperature spatial pattern of major atmospheric circulation variability indices such as the Arctic Oscillation (AO), suggesting that radiative forcing is a primary forcing for AA. Mechanisms for AA include reduced summer albedo due to sea-ice and snow-cover loss, decreased total cloudiness in summer and increased cloud cover in winter, additional atmospheric heating generated by newly sea-ice-free ocean areas that are maintained later into the autumn, increased longwave radiation due to local and advected atmospheric moisture sources, and the decreased rate of heat loss to space in the Arctic relative to the subtropics due to lower mean temperatures (Makshtas et al., 2011; Pithan and Mauritsen, 2014).

Figure 5.  Annual Arctic 925-hPa air temperature increases for 2010-14 relative to the end of the 20th century (1971-2000). Figure created through the NOAA/ESRL website.

Upward trends in GHGs and resulting AA significantly influence multiple changes throughout the Arctic environment (Stroeve et al., 2012). This is evident in the Alaskan Arctic where robust lagged relationships are found between Northern Hemisphere SAT and September Beaufort Sea ice extent, particularly since the early 1990s (Ballinger and Rogers, 2014). This lag is explained by ongoing increases in GHGs causing global warming that contribute to AA temperature increases and resultant thinning of summer sea-ice cover in the Alaskan Arctic over the course of several years, resulting in a series of recent, anomalous September sea-ice losses.

Figure 6.  (a) Composite of June SLP (units: hPa) for 2007-15, illustrating the SLP distribution for the negative phase of the AD pattern. Data are from the NCEP-NCAR Reanalysis through the NOAA/ESRL. (b) The AO Index, an Arctic wide index low pressure in its positive phase, and the AD, during early summer months. The negative phase of the AD pattern is often associated with higher pressure in the Beaufort Sea. Note the recent presence of negative AD values in June and July 2015 (b and c) (updated from Overland and Wang 2005). Note also that some authors define the dipole with the opposite sign (e.g., Wu et al., 2006).

Warm temperatures in the Alaskan Arctic have persisted since 2007. Extended periods of sea-ice-free conditions have a role in the pronounced temperature departures from the norm. In particular, Beaufort Sea ice formation occurs progressively later during this era, especially during autumn 2012 when ice formed approximately five weeks later relative to the 1981-2010 climatology (Fig. 6). Since 2007, there has been an increase in easterly winds, which has helped set the stage for AA by advecting sea ice out of the Alaskan region and enhancing ocean surface stratification due to the offshore transport of fresh water from the large Mackenzie River discharge plume (Wood et al., 2013).

Figure 7.  Beaufort Sea ice freeze date anomalies, 1979-2016, compared to the 1981-2010 mean freeze date (data obtained from Jeffrey Miller, Cryospheric Sciences Laboratory, NASA Goddard Space Flight Center and KBRwyle).

Recent decades are also associated with the predominance of a large spatial-scale climate pattern referred to as the Arctic Dipole (AD; Figs. 7a, b and c), which is characterized by low SLP on the Siberian side of the Arctic and high SLP on the North American side in its negative phase (Fig. 7a; Overland and Wang, 2005; Wang et al., 2009, 2014). Note, however, some authors define the dipole with the opposite sign (e.g., Wu et al., 2006). One can also interpret this decadal change as the increased presence of a summer Beaufort high region located north of Alaska continuing from its springtime climatology (Ballinger and Sheridan, 2014; Ballinger et al., 2014). These anomalous Beaufort high patterns have occurred more often since 2007, aligned with an era of abrupt sea-ice decline, as compared to previous years dating to the late 1970s. This summer Beaufort high is a major change for the Alaskan Arctic, as the previous summer climatology often consisted of weak pressure gradients and monthly-averaged low pressure in the central Arctic basin. As summarized by (Overland et al., 2014a): "While the negative AD pattern was present in spring as early as 1997, its recent occurrence began in summer 2007 when it was present in all months and contributed to 2007 record minimum summer sea ice extent (Wang et al., 2009). Most years after 2007 have seen the AD pattern persist for at least part of the summer. For example, in 2010, the AD pattern was present in May and June, but then the Arctic reverted to the more traditional climatological summer SLP pattern involving a weak central-Arctic low-pressure center. But by August 2010 the AD pattern had returned." The AD pattern was absent in the summer 2013 and 2017, but reappeared in 2014 and 2015 with Beaufort/Chukchi sea-ice extents below normal.

The increased Beaufort high and AD patterns since 2007 are also connected with unprecedented higher pressure systems across Greenland and the North Atlantic Arctic sector in one large positive hemispheric SLP anomaly pattern (Overland et al., 2012; Belleflamme et al., 2015; Bezeau et al., 2015; Petrie et al., 2015). Whether this shift in atmospheric pressure and wind patterns is tied to AA is unknown, but its persistence is noted relative to more strictly interannual variability before 2007 (Fig. 7b and c). This provides further evidence that Alaskan Arctic changes are tied to large-scale Arctic-centric changes.

In summary, the Alaskan Arctic has participated in the Arctic-wide AA driven by increases in GHGs and amplified by regional, Arctic-specific feedback processes. Persistent higher than normal surface pressures in the Pacific Arctic influenced by the anomalous occurrence of the summer Beaufort high pressure system have increased easterly winds in the region and contributed to ocean circulation changes and sea-ice loss.

The AA of SAT is projected to continue through the 21st century (Fig. 8) according to CMIP5 climate models, which formed the basis for IPCC AR5 (Overland et al., 2014b). Because CO2 remains in the atmosphere for many decades, the GHG contribution from the previous decades and projected emissions for the next two decades lead to a model-estimated, Arctic-wide October-March SAT increase of +4^°C by 2040. Beyond 2050, the SAT depends on which GHG emissions scenario is chosen for the projection. The red curves for summer and winter periods give the high CO2 business-as-usual emissions scenario, referred to as RCP8.5, according to IPCC AR5 (IPCC, 2013). The blue curves provide the change in SAT based on the aggressive but not extreme GHG mitigation scenario, RCP4.5.

According to the CMIP5 models, sea-ice loss for the Alaskan Arctic is projected to continue over the next decades. The important change is the increase in the number of sea-ice-free months (Fig. 9). The duration of months with open-water conditions generally decreases with northerly latitude. A rough change estimate at 74^° N is from three months of open water in 2010 to five months by 2040 (Wang and Overland, 2015, updated). These average changes are based on GHG increases; actual sea-ice loss is projected to continue to have a large year-to-year component due to variations in weather patterns. It will be difficult for the open-water duration to extend much later than November due to seasonal darkness, or occur earlier in spring due to solar reflection off of snow and sea-ice cover; thus, the future duration of Alaskan sea-ice-free duration will be limited by the winter atmospheric climatology.

Figure 8.  Future Arctic-wide SAT increases for a business-as-usual increase in CO₂ emissions (RCP8.5, red) and for a modest (RCP4.5, blue) CO₂ mitigation scenario.

Figure 9.  Annual duration of sea-ice cover averaged over the period 1990-2014 (left) based on satellite data. Right: change (relative to 1990-2014) in annual sea-ice duration by the middle of the century (2030-44) based on seven CMIP5 model means under the RCP8.5 emissions scenario. Figures are adapted from Wang et al. (2017) with modification. The subset of CIMP5 models were selected by matching the monthly sea-ice extent and magnitude of the seasonal cycle. See Wang and Overland (2015) and Wang et al. (2017) for more information.

After more than a decade-and-a-half of both positive and negative SSTs and associated weather patterns in the central and eastern North Pacific Ocean, there is evidence of multi-month persistent positive ocean temperature anomaly patterns since 2013. These consist of near-record positive SST anomalies centered near (45^°N, 145^°W) during 2013, labeled the "blob" by (Bond et al., 2015), and the return of the positive Pacific Decadal Oscillation (PDO) climate pattern in 2015, with SST maxima near the northeastern North Pacific coast. Persistent, above-average geopotential heights in the mid-level atmosphere during 2012-15 associated with warmer air temperatures, which steer the prevailing wind direction more from the south and transport heat towards the north over the central and eastern North Pacific, have resulted in what has been referred to as the Ridiculously Resistant Ridge (RRR) of high pressure (e.g. Fig. 10b).

Figure 10.  September 2014 through July 2015 925-hPa air temperature anomalies over western North America (a), and corresponding anomalies in 700-hPa geopotential height (b); anomalous winds follow the contours with a southerly wind component over the Gulf of Alaska. Anomaly maps are presented with respect to the 1981-2010 climatological values. Data are from the NCEP-NCAR Reanalysis through NOAA/ESRL.

Figure 11.  The PDO index time series from 1900-2016. Positive values correlate with elevated SST in the Gulf of Alaska. The PDO index is obtained from http://research.jisao.washington.edu/pdo/PDO.latest.

Beginning in late autumn 2014, Alaska experienced record positive temperature anomalies associated with the RRR orientation of mid-tropospheric geopotential heights over the west coast of North America, and a positive PDO with above-average lower tropospheric air temperatures, situated polewards from the southern Alaskan coast (Fig. 10a and b). Winds flow clockwise around high geopotential height centers (parallel to contours), thus directing the air flow from the North Pacific northwards across Alaska to the Alaskan Arctic region. The PDO index supports this wind pattern and is strongly positive (>+1.0) beginning September 2014 and decreased into 2017 (Fig. 11).

Winter 2015-16 continued the warm pattern, with widespread Alaskan temperature anomalies of +5^°C (Walsh et al., 2017). The 700-hPa geopotential height pattern is similar to 2015, as shown in Fig. 10, but the low geopotential height Aleutian low center is more dominant than the coastal ridge feature. Loss of snow cover and decreased land surface albedo in southern Alaska added to the persistence of positive near-surface air temperature anomalies (Walsh et al., 2017). El Niño conditions, as well as warm North Pacific SSTs, continue for winter 2015-16. Previous research suggests warm temperature anomalies in the Alaskan marine Arctic during El Niño often result in diminished Beaufort and Chukchi ice cover (Papineau, 2001; Liu et al., 2004; Bond and Harrison, 2006). (Walsh et al., 2017) also estimated that about 20% of the 2015-16 Alaska warm temperature anomalies (about +1^°C) was due to global warming, as projected by CMIP5 models.

Autumn 2016 marked the end of warm northeast Pacific SSTs, with a return to more zonal 700-hPa wind flow and with the Aleutian low feature moving northwest spanning northeastern Siberia and the Sea of Okhotsk with above normal temperatures confined to the Chukchi Sea and the Alaskan Arctic (Figs. 12a and b).

Figure 12.  925-hPa air temperature anomalies over western North America (a), and corresponding 700-hPa geopotential height (b) for autumn 2016. Anomaly maps are presented with respect to the 1981-2010 climatological values. Data are from the NCEP-NCAR Reanalysis through NOAA/ESRL.

(Newman et al., 2016) discusses causal contributions to North Pacific SSTs and the PDO and concludes there is a combination of tropical forcing, North Pacific Ocean memory, and interannual chaotic atmospheric variability. North Pacific atmospheric processes have a long-memory stochastic (random) character (Overland et al., 2006), rejecting purely cyclic predictions.

Despite a shift towards a weak La Niña, autumn 2016 showed some evidence that the PDO might continue to be neutral or weakly positive based on persistence, and there is some evidence for warm subsurface ocean temperature anomalies (Zhang and Delworth, 2015). Yet, strong zonal atmospheric flow (Fig. 12b) is the primary reason for the termination of the North Pacific contribution to Alaskan Arctic warming. (Baxter and Nigam, 2015) show that notable climate anomalies in the Pacific-North American sector can be caused by such internal variability of regional atmospheric patterns, and need not originate from the tropics or local surface forcing. The future for the Alaskan marine Arctic primarily involves continued warm temperatures based on AA with occasional midlatitude support.

One should note that future air temperature increases are likely to manifest as considerable year-to-year extremes based on internal random variability of the atmosphere added to long-term GHG-induced trends, rather than the smooth projections shown in Fig. 8. Extreme Arctic temperature events, as a combination of anthropogenically forced temperature increases combined with natural variability, will become common, exceeding previous thresholds. Such an event occurred with +4^°C temperature anomalies for Alaska in November-December 2014 and +5^°C January-April 2016, related to recent warm Pacific SSTs. Breaking the string of cold, southern Bering Sea temperature anomalies and mostly negative PDO years from 2006-13, recent years show interaction of the Beaufort and Chukchi Seas with the subarctic. Regional warm temperature anomalies associated with loss of sea ice and snow for the Alaskan Arctic have been supplemented by southerly air flow in addition to the monotonic AA signal. This North Pacific SST connection broke down in autumn 2016 due to internal atmospheric variability that manifested as strong zonal winds.

For the foreseeable future (out to 2040), continuing rapid environmental changes in Alaskan Arctic seas, land, atmosphere and sea ice are likely, and the appropriate response is to plan for adaptation to meet these mean and extreme-event changes. Arctic and global climate changes will continue to propagate throughout the biological ecosystem through shifts in winds and air temperatures, sea-ice loss, ocean circulation and stratification changes, and permafrost melt, with impacts on societal systems.