1. Introduction
The recent loss of Arctic sea ice (Stroeve et al. 2012a,b) has been one of the most notable aspects of late twentieth-century and early twenty-first century climate change. A robust human contribution to the observed (1979 onward) Arctic sea ice loss has been detected (Min et al. 2008; Kay et al. 2011), especially since the early 1990s when the rate of decline increased (Comiso 2012). It is predicted that Arctic sea ice will continue to decline, with a seasonally ice-free Arctic Ocean (defined as a sea ice extent less than 1 × 106 km2) before midcentury likely under representative concentration pathway 8.5 (Wang and Overland 2009, 2012; Collins et al. 2013). The phenomenon known as Arctic amplification (Serreze et al. 2009)—observed as the Arctic warming at a faster rate than the global average—is largely driven by positive ice–albedo feedback associated with diminishing sea ice (Screen and Simmonds 2010), although a number of processes contribute to Arctic amplification (Cohen et al. 2014). The strongest signature of Arctic amplification is seen in the lowermost part of the troposphere, notably in the late autumn and early winter when the ocean–atmosphere temperature gradient and, in turn, the surface turbulent heat fluxes are largest (Deser et al. 2010; Screen et al. 2013). Also associated with observed Arctic sea ice loss are a decrease in the strength of the surface temperature inversion, an increase in specific humidity, and an increase in lower-tropospheric thickness (Screen et al. 2013).
Local warming above regions of sea ice loss can be advected to the nearby land–atmosphere boundary layer by climatological submonthly transient eddies (Deser et al. 2010). The advection of warmed Arctic maritime air to adjacent continents can even act to cause mean warming and less severe cold air outbreaks in the absence of circulation changes (Ayarzagüena and Screen 2016). However, in remote regions, dynamically induced changes can dominate over thermodynamically induced changes. For example, sea ice loss—specifically in the Barents–Kara Sea—has been shown to force seasonal-mean atmospheric circulation anomalies (a strengthening of the Siberian high) that lead to strong cold air advection and cold Eurasian winters (Mori et al. 2014; Kug et al. 2015).
Weakened and equatorward-shifted midlatitude westerlies are another feature of Arctic amplification, through the associated decrease in the equator-to-pole temperature gradient (Deser et al. 2010). Francis and Vavrus (2012) suggested that the slowdown of the midlatitude upper-level winds can be linked to an increased meandering of the jet stream, with higher-amplitude planetary waves leading to more persistent weather patterns. However, it remains difficult to find robust evidence linking Arctic amplification and midlatitude extreme weather, with changes in planetary waves dependent on the metric being used. It should be noted that many of these dynamically induced changes are found in studies that explore the response to Arctic sea ice loss in isolation, often utilizing an uncoupled atmospheric general circulation model (AGCM). The net effect of greenhouse gas (GHG)-driven warming (that induces the Arctic sea ice loss) can dominate the mean climate response (Deser et al. 2015) and the number of daily cold extremes (Screen et al. 2015) on longer time scales. Further, it has been shown recently that the remote atmospheric response to Arctic sea ice loss can extend into the tropics and Southern Hemisphere in coupled model experiments (Deser et al. 2015).
Finding a robust response to observed and projected Arctic sea ice loss is typically regarded as a signal-to-noise problem. Random atmospheric internal variability is seen as an obstacle to detecting a significant atmospheric response to sea ice loss, especially since the mean state of the atmospheric circulation can condition the atmospheric response to a given sea ice anomaly (Balmaseda et al. 2010; Semenov and Latif 2015). But can this state dependence actually be a help rather than a hindrance? For example, we know that certain background ocean–atmosphere states, while not strictly predictable, vary on (multi-) decadal time scales. Two dominant patterns of ocean–atmosphere variability in the Northern Hemisphere midlatitudes are the Pacific decadal oscillation (PDO) and the Atlantic multidecadal oscillation (AMO). Recently, Screen and Francis (2016) showed that Arctic warming is enhanced for an identical loss of sea ice in the negative phase of the PDO relative to the positive phase. Given this and previous findings, we suggest that the local and/or remote atmospheric response to a prescribed reduction in Arctic sea ice may also be sensitive to the phase of the AMO. Because of the persistence of a particular phase of the AMO, such a finding could offer opportunities for decadal time-scale prediction.
Here, we investigate the sensitivity of the response to an idealized Arctic sea ice loss to sea surface temperature (SST) patterns representative of the opposite phases of the AMO. It has often proved difficult to robustly identify such state dependence in previous work because of the brevity of observations of sea ice extent. Our experimental design is such that we can systematically address this research question by utilizing a large number of ensemble members in an AGCM with large-magnitude (relative to the observed period) perturbations to sea ice concentration (SIC) and SST surface boundary conditions. This approach allows us to isolate the differing responses to sea ice loss under opposite phases of the AMO in a way not possible using observations.
\ 2 describes the model used and the experimental design as well as how the influence of the background ocean–atmosphere state on the response to Arctic sea ice loss is measured. Model results are presented in section 3, with an initial consideration of the response to our prescribed sea ice loss alone (section 3a), before a more in-depth presentation of the differing responses to sea ice loss under the negative and positive phases of the AMO (section 3b). Section 3c explores similarities in our model results and reanalysis data. In section 4, conclusions are presented.
2. Experimental design and methodology
A total of four model experiments are run, using the Met Office Hadley Centre Global Atmospheric Model, version 2 (HadGAM2) (Martin et al. 2011). This version has a horizontal resolution of 1.875° × 1.25° in longitude and latitude, respectively, with 38 vertical levels. This AGCM is forced with prescribed boundary conditions (SICs and SSTs), while GHGs and other radiatively active species are held constant.
The four experiments performed are prescribed with unique combinations of either low or high sea ice states (LICE and HICE, respectively) with negative or positive AMO-like SSTs (AMO− and AMO+, respectively), giving LICE/AMO−, HICE/AMO−, LICE/AMO+ and HICE/AMO+. In an effort to maximize the signal-to-noise ratio, each of the four experiments is 100-yr long. This is a sequential simulation with each year initialized from the atmospheric state of the end of the previous year. Each year is treated as independent of all other years.
Low and high sea ice states, and negative and positive AMO SSTs, are prescribed at the extremes of what has been observed in the late twentieth century and early twenty-first century. Again, this is an attempt to identify a robust signal of different responses to Arctic sea ice loss in the opposite AMO phases. SIC and SST data are taken from the Met Office Hadley Centre Sea Ice and Sea Surface Temperature dataset (HadISST) (Rayner et al. 2003). For each calendar month and in each grid box, we calculate the 1979–2013 climatological mean μSIC and standard deviation σSIC of SIC. Note that we restrict these calculations to the satellite era (1979 onward), since before this date, sea ice data are taken from less reliable sources, such as digitized sea ice charts. A SIC anomaly of μSIC − 2σSIC is prescribed in the LICE experiments, and a SIC anomaly of μSIC + 2σSIC is prescribed in the HICE experiments, ensuring that SICs do not fall outside the range of 0%–100%.
The absence of ocean–atmosphere coupling here means that local changes in SST in response to the prescribed changes in SIC are not directly accounted for. Local SST changes in response to sea ice loss were estimated to be as high as 5°C during the (then record) SIC minimum of 2007 (Steele et al. 2008). Screen et al. (2013) proposed a method to account for these local SST changes, which we use in our experimental design; for calendar months and grid boxes where we prescribe a SIC anomaly (i.e., σSIC > 0), we also prescribe an SST anomaly. Again, we calculate the 1979–2013 climatological mean μSST and standard deviation σSST|SIC (where the subscript SST|SIC refers to local SST changes in response to sea ice loss). We use SST anomalies of μSST + 2σSST|SIC and μSST − 2σSST|SIC for the LICE and HICE experiments, respectively. We restrict SSTs to no lower than −1.8°C, the freezing temperature of saltwater. In the Southern Hemisphere and where no SIC anomaly is prescribed (σSIC = 0; as is the case in always-ice-free or always-100%-SIC grid boxes), we use μSIC.
In grid boxes that are always ice free (Northern and Southern Hemispheres), we apply an AMO-like SST anomaly. We regress the detrended and normalized annual-mean AMO index against detrended annual-mean global SST to determine an SST anomaly per standard deviation departure in the AMO index βSST|AMO. The AMO index used is from the National Oceanic and Atmospheric Administration (NOAA)/Earth System Research Laboratory (ESRL) (Enfield et al. 2001). We use data from the earlier time of 1948 through 2013 because this AMO time series is calculated from SST data area weighted over the largely ice-free North Atlantic. Therefore, SST data are expected to be more reliable here than over regions with (at least seasonal) sea ice. We prescribe SST anomalies of −2βSST|AMO and +2βSST|AMO for the AMO− and AMO+ experiments, respectively. The same AMO-like SSTs are applied in each calendar month, again ensuring that SSTs are no lower than −1.8°C. AMO-like SSTs are deliberately not restricted to the North Atlantic basin since a number of studies suggest that the North Atlantic can play a key role in inducing North Pacific climate variability and tropic-wide SST anomalies (Zhang and Delworth 2007; Li et al. 2016).
Figure 1 shows the differences in surface boundary conditions between the LICE and HICE experiments for the individual winter months (December, January, and February) and winter (DJF) mean. The SST differences are those directly due to local SIC differences. The largest SIC differences (decreases) are in the marginal winter sea ice zones of the North Atlantic and the Bering Sea, as well as Hudson Bay in December and the Sea of Okhotsk in January–February. Sea ice–induced SST differences (increases) mirror SIC differences and can, in the DJF mean, exceed 4°C in the Greenland Sea and 2.5°C in the Sea of Okhotsk. The AMO-related SST anomalies are of the same sign over the North Atlantic, with larger-magnitude anomalies extending in a horseshoe shape from a maximum east of Newfoundland, Canada, to western Europe and southward to the subtropics west of North Africa (Fig. 2), consistent with previous studies (e.g., Alexander et al. 2014).
Prescribed (a)–(d) SIC and (e)–(h) SST differences between the LICE and HICE experiments for the individual winter months in (a)–(c) and (e)–(g) and DJF mean in (d) and (h). SST differences here are those resulting from local SIC differences (see section 2).
Citation: Journal of Climate 30, 5; 10.1175/JCLI-D-16-0531.1
Prescribed SST differences between the AMO− and AMO+ experiments. DJF mean is shown, although monthly AMO-related SST anomalies are invariant by design (see section 2).
Citation: Journal of Climate 30, 5; 10.1175/JCLI-D-16-0531.1
There is also a region of cool SST anomalies in the central–western North Pacific, between about 30° and 40°N. This cool SST pattern is consistent with the positive phase of the PDO. During the late twentieth century, there has been a roughly 10-yr lag between switches of the AMO phase and the phase of the PDO. This lagged response—with the AMO and PDO being out of phase during this period—is attributed to atmospheric teleconnections, which are reinforced by anomalous oceanic circulation over the North Pacific (Zhang and Delworth 2007). If this component of North Pacific climate variability is indeed forced by the AMO and not simply an artifact of the regression approach and the observed period used, then these SST anomalies should be prescribed for completeness. This approach also serves to test whether there is a different sensitivity of the atmospheric response to Arctic sea ice loss to AMO-like SST anomalies in the Atlantic and the Pacific. A more evident modulation of the atmospheric response to Arctic sea ice loss in the North Pacific, which is prescribed with lower-magnitude and less spatially extensive AMO-like SST anomalies than the North Atlantic (Fig. 2), would imply a far greater sensitivity to AMO-like SSTs in the North Pacific.
The response to Arctic sea ice loss during AMO− is calculated by subtracting the ensemble mean in the HICE/AMO− experiment from that in the LICE/AMO− experiment, with this response denoted as (LICE − HICE)AMO−. The response under AMO+ is (LICE − HICE)AMO+ and is the difference between the LICE/AMO+ and HICE/AMO+ experiments. The AMO modulation of the response to sea ice loss is then calculated by taking the difference between these two responses and is denoted as (LICE − HICE)AMO− − (LICE − HICE)AMO+. We also combine the two LICE experiments and two HICE experiments to produce two “super ensembles,” each with 200 ensemble members. We are then able to calculate an AMO-independent response to sea ice loss [(LICE − HICE)AMO−,AMO+] by subtracting the concatenated HICE/AMO− and HICE/AMO+ experiments from the concatenated LICE/AMO− and LICE/AMO+ experiments. Likewise, to calculate the response to the negative AMO phase that is independent of the sea ice state, we subtract the concatenated LICE/AMO+ and HICE/AMO+ experiments from the concatenated LICE/AMO− and HICE/AMO− experiments. This response is denoted as (AMO− − AMO+)LICE,HICE. The significance of the ensemble-mean difference is computed using a Student’s t test, with the null hypothesis of equal means rejected with 95% confidence when p ≤ 0.05.
3. Results
We focus on the wintertime response here. Winter atmospheric responses to Arctic sea ice loss are typically larger in magnitude than in summer (Petrie et al. 2015). This is found to be the case here (other seasons not shown). Further, the most obvious state dependence of the response to sea ice loss is seen in winter.
a. Sea ice loss response
The response to sea ice alone, independent of the AMO phase, displays the classic signature of Arctic amplification. The (zonally averaged) greatest warming is seen in the lowermost part of the troposphere over the Arctic (Fig. 3a). This tropospheric warming leads to increases in geopotential heights, with the greatest increases in the mid- and upper troposphere above the region of greatest tropospheric warming (Fig. 3b). Figure 3c shows a weakening of the mid-to-high-latitude westerlies (centered around 60°N), consistent with thermal wind balance in response to Arctic amplification. Stronger westerlies are seen on the equatorward side of the eddy-driven jet (about 30°–40°N), although these increases are smaller in magnitude than the decreases on the poleward side, as found in other studies (e.g., Deser et al. 2015). Considering the AMO-independent response to sea ice loss spatially, the greatest warming (at 850 hPa) is west of Greenland and over the Sea of Okhotsk (Fig. 3d). Interestingly, there is evidence of the “warm Arctic–cold Eurasia” response, with a significant cooling response over parts of China. Consistent with previous studies (e.g., Mori et al. 2014; Kug et al. 2015), this can be linked to an intensified Siberian high and cold air advection over East Asia, with positive sea level pressure (SLP) anomalies found south of the Kara Sea (see Figs. 4a,b for a comparison of SLP responses to sea ice loss under the opposite AMO phases). Sea ice loss causes pan-Arctic significant increases in geopotential height (at 500 hPa), with significant decreases in geopotential height over the western North Atlantic, western Europe, and the Sea of Japan (East Sea) (Fig. 3e). The greatest reduction in the mid-to-high-latitude westerlies (at 300 hPa) is found to the north of these regions (Fig. 3f).
(a)–(c) Zonal-mean and (d)–(f) pressure-level wintertime responses to Arctic sea ice loss, independent of the AMO phase [(LICE − HICE)AMO−,AMO+]. Hatching shows responses that are not statistically significant at the 95% (p = 0.05) confidence level. White shading in (d) indicates regions of high topography, where the surface pressure falls below 850 hPa.
Citation: Journal of Climate 30, 5; 10.1175/JCLI-D-16-0531.1
(a)–(c) Wintertime SLP and (d)–(f) near-surface (1.5 m) air temperature responses to Arctic sea ice loss during (left) AMO− [(LICE − HICE) AMO−], (center) AMO+ [(LICE − HICE)AMO+], and (right) their difference [(LICE − HICE)AMO− − (LICE − HICE)AMO+]. Additional contours in (a)–(c) show the climatological SLP (average across all four experiments). Contours are drawn at 10-hPa intervals. Hatching shows responses that are not statistically significant at the 95% (p = 0.05) confidence level. Note the different color scales in each panel.
Citation: Journal of Climate 30, 5; 10.1175/JCLI-D-16-0531.1
b. AMO modulation of the sea ice loss response
Figure 4 shows the response of atmospheric variables to our prescribed wintertime Arctic sea ice loss under the opposite phases of the AMO, as well as their difference. The intensified Siberian high in response to ice loss is apparent in both AMO− and AMO+ conditions (Figs. 4a,b). In both phases, there are cyclonic SLP anomalies over the Sea of Okhotsk. This dipole structure acts to increase the pressure gradient across eastern Russia, leading to the increased cold air advection over East Asia already discussed. The reduction in SLP over the Sea of Okhotsk can be interpreted as a shallow thermal low due to the large difference in SIC between the LICE and HICE experiments here (Screen et al. 2014). The increase in open ocean area causes a large air–sea temperature difference, with increases in upward turbulent heat fluxes (not shown) warming the near-surface atmosphere (Figs. 4d,e) and forcing ascent in the planetary boundary layer (PBL), which lowers SLP (Strey et al. 2010). A similar thermal low is apparent west of Greenland and over Hudson Bay. Indeed, in the zonal mean (Fig. 3b), the lowermost atmosphere shows a negative, albeit insignificant, geopotential height response to sea ice loss over the Arctic (north of 70°N; recall that this is the AMO-independent response to Arctic sea ice loss). The SLP response difference [Fig. 4c; (LICE − HICE)AMO− − (LICE − HICE)AMO+] shows a slightly southward-shifted and deepened Aleutian low, as well as significant negative anomalies over northeastern North America. This Aleutian low response difference is due to a significant deepening response to sea ice loss under AMO− that is not apparent in the response to sea ice loss under AMO+. Interestingly, there are no significant response differences in the Atlantic–European sector, where we prescribed the largest AMO-related SST anomalies (Fig. 2). The greatest near-surface air temperature increases are found in the regions of greatest sea ice loss (Figs. 4d,e). There is a muted warming response to sea ice loss during AMO− over the central–western Pacific and northeastern North America, with an enhanced warming response across northwestern North America (Fig. 4f). These response differences are linked to the SLP response differences in Fig. 4c.
We also find significant modulation of the atmospheric circulation response to Arctic sea ice loss by AMO phase at higher altitudes (Fig. 5). Perhaps the most striking feature of the upper-level response difference between AMO− and AMO+ phases is found in 500-hPa geopotential heights (Figs. 5a–c). There is a wave train pattern in the response difference across the Pacific–North American sector. This is due to an amplification of the wintertime climatological Aleutian trough–western North American ridge–eastern North American trough pattern in response to sea ice loss in the negative phase of the AMO, which is not apparent in the positive phase of the AMO. This trough–ridge–trough (TRT) pattern appears to emanate from the central–western North Pacific but, again, there are no obvious response differences in the Atlantic–European sector. The 500-hPa temperature responses and their difference reflect the changes in 500-hPa geopotential height, and vice versa (Figs. 5d,f).
As in Fig. 4, but for (a)–(c) 500-hPa geopotential height response, (d)–(f) 500-hPa temperature response, and (g)–(i) 300-hPa westerly wind response. Additional contours in (g)–(i) show the climatological 300-hPa westerly wind (average across all four experiments). Contours are drawn at 10 m s−1 intervals. Note the different color scales in each panel.
Citation: Journal of Climate 30, 5; 10.1175/JCLI-D-16-0531.1
The 300-hPa zonal wind direct response to sea ice loss (Fig. 3f; irrespective of AMO phase) masks important response differences due to the AMO phase (Figs. 5g–i). During AMO+, there are positive westerly wind anomalies to the south of the wintertime climatological jet stream maximum in the western North Pacific in response to sea ice loss. There are also decreases to the north of this region, over the Sea of Okhotsk. There is a similar pattern of westerly wind response in the North Pacific in AMO− but shifted downstream of the climatological maximum with larger-magnitude positive anomalies. This is suggestive of an eastward shift in the North Pacific jet stream maximum. Equatorward shifts are also apparent over the North Atlantic in response to sea ice loss, in both AMO phases (Figs. 5g,h), which project onto the negative phase of the North Atlantic Oscillation (Screen et al. 2013). During AMO−, the shift is seen over the western North Atlantic, but during AMO+, the shift is more evident over the eastern North Atlantic. The response differences over the North Atlantic are mostly nonsignificant (Fig. 5i). We use the methodology of Screen et al. (2014) to check that these response differences are robust (the ensemble size is large enough to separate the forced signal from internal variability). This methodology can be used to estimate the minimum ensemble size required to detect a statistically significant ensemble-mean difference. For the response differences shown in Figs. 4 and 5, the minimum ensemble size required to identify a statistically significant signal is typically 50–60 (not shown).
Why should the Pacific–North American circulation response to Arctic sea ice loss be conditional on the AMO phase? By design, there is no residual prescribed boundary condition forcing in the response difference [(LICE − HICE)AMO− − (LICE − HICE)AMO+]. Figures 6a and 6b show 850-hPa temperature (shading) and circulation (arrows) responses to sea ice loss independent of AMO phase [(LICE − HICE)AMO−,AMO+] and AMO− independent of sea ice state [(AMO− − AMO+)LICE,HICE], respectively. The prescribed sea ice loss forces a broadly anticyclonic circulation around the subarctic, south of the regions of ice loss, which can be understood as the weakening of the westerly wind at these latitudes centered around 60°N (as already discussed; see Fig. 3c). However, this spatial map reveals some latitudinal deviations from this subarctic anticyclonic anomaly, with a notable southward excursion in the northwestern North Pacific. There are clear wintertime-mean atmospheric circulation responses to our prescribed AMO-like SSTs also.
The (a) 850-hPa temperature (shading) and wind (arrows) responses to Arctic sea ice loss, independent of the AMO phase [(LICE − HICE)AMO−,AMO+], and (b) 850-hPa temperature and wind responses to AMO−, independent of sea ice state [(AMO− − AMO+)LICE,HICE]. Hatching shows responses that are not statistically significant at the 95% (p = 0.05) confidence level. White shading indicates regions of high topography (where the surface pressure falls below 850 hPa). Note that values below the color-scale legend are for the Arctic sea ice loss response in (a) and values above the color-scale legend are for the AMO− response in (b). The responses are shown from 20° to 90°N in (a) and from 0° to 90°N in (b). The green boxes denote the region in the subarctic North Pacific where anomalous northerly flow in the separate responses are coincident.
Citation: Journal of Climate 30, 5; 10.1175/JCLI-D-16-0531.1
The AMO phase influences circulation in the Atlantic, as might be expected, with a prominent anomalous cyclonic circulation centered over the United Kingdom. However, there are also marked wind anomalies in the North Pacific, with cyclonic anomalies in response to AMO− around the position of the climatological Aleutian low. This is consistent with a strengthened Aleutian low in response to an SST pattern representative of positive PDO (Rodionov et al. 2007); the correlation coefficient between the strength of the Aleutian low, measured by the North Pacific (NP) index (Trenberth and Hurrell 1994), and the PDO is −0.72 for winter 1940–2005 (when the NP index is negative, the Aleutian low is strong). Although there is evidence that atmospheric changes lead changes in SSTs in the North Pacific domain, there are physical mechanisms describing the response of the atmosphere to changes in SSTs (Trenberth and Hurrell 1994; Sung et al. 2014). The mechanism used to describe the North Pacific ocean–atmosphere state dependence of the atmospheric response to Arctic sea ice loss (below) also applies to the response to AMO− independent of sea ice state. Northerly wind anomalies on the western flank of this anomalous cyclonic circulation reach as far south as the tropical North Pacific. This cyclonic circulation is positioned to the southeast of the anticyclonic circulation over the northwestern North Pacific that is a response to sea ice loss alone. It is worth noting that the significant temperature responses to AMO−, independent of sea ice state, extend across most of the Northern Hemisphere, with a local peak in the North Pacific around 35°–40°N. In contrast, the significant temperature responses to sea ice loss independent of AMO phase are largely confined to the Arctic and subarctic regions.
Figure 7 shows the 850-hPa temperature (shading) and wind (arrows) responses to Arctic sea ice loss. If the atmospheric response to sea ice loss was not dependent on the background ocean–atmosphere state, then we would not see significantly different responses under negative AMO and positive AMO. Also, with no dependence, the response to sea ice loss under a particular AMO phase will not contain any pattern of the response to AMO phase alone (Fig. 6b), since there should be no residual AMO-like SST forcing, motivating this separation of Fig. 7 into the separate responses to sea ice loss under opposite AMO phases. This is the case over the Atlantic–European sector under both AMO phases (Figs. 7a,b). There is, however, a strengthened Aleutian low in the atmospheric response to sea ice loss under AMO− (Fig. 7a), which echoes the response to AMO− independent of sea ice state in Fig. 6b. This suggests that in one (or both) of the AMO− experiments (LICE/AMO− and/or HICE/AMO−), the circulation pattern is not simply a linear addition of the separate circulation anomalies of LICE or HICE and AMO− (relative to climatology). In other words, there is a mechanism at work that nudges the deepened Aleutian low in response to AMO−, independent of sea ice state, toward either a much stronger state in the LICE/AMO− or a much weaker state in the HICE/AMO− experiment. It is notable that in the subarctic North Pacific, the separate wind anomalies in response to 1) Arctic sea ice loss alone, independent of the AMO phase, and 2) AMO− alone, independent of the sea ice state, are aligned in the same direction (Fig. 6, green boxes). Therefore, in the LICE/AMO− experiment, the anomalous northerly flow associated with the lobe of anticyclonic circulation in LICE is reinforced by the anomalous northerly flow on the western flank of the strengthened Aleutian low in AMO−. As mentioned, these northerly wind anomalies can cause local warming to be advected to nearby regions (Deser et al. 2010). But, as found by Mori et al. (2014), significant dynamical changes can cause cold air advection to the midlatitudes despite significant Arctic and subarctic warming. Here, such an anomalous northerly flow is likely to advect cold polar air from the Arctic to the North Pacific midlatitudes. This can have implications for the meridional temperature gradient in this region and the associated storm track.
Wintertime 850-hPa temperature (shading) and wind (arrows) responses to Arctic sea ice loss during (a) AMO− [(LICE − HICE)AMO−], (b) AMO+ [(LICE − HICE)AMO+], and (c) their difference [(LICE − HICE)AMO− − (LICE − HICE)AMO+]. Hatching shows responses that are not statistically significant at the 95% (p = 0.05) confidence level. White shading indicates regions of high topography (where the surface pressure falls below 850 hPa). Note the different color scale in (c).
Citation: Journal of Climate 30, 5; 10.1175/JCLI-D-16-0531.1
The climatological meridional SST gradient over the North Pacific is greatest at about 40°N and in the west of the basin, associated with a midlatitude frontal zone [the Kuroshio and Oyashio Extensions (KOE)]. The differential heat supply across this zone maintains surface baroclinicity, which has been shown to be key in anchoring and sustaining a storm track (Frankignoul et al. 2011; Sung et al. 2014). Indeed, in the North Pacific, it has been shown that the SST anomaly pattern, corresponding to the phase of the PDO, changes the meridional SST gradient and meridional PBL temperature gradient, which, in turn, has an effect on the location and strength of the jet stream and storm track (Bond and Harrison 2000; Sung et al. 2014). During the positive PDO phase—recall that the region of cool SST anomalies in the central–western North Pacific between about 30° and 40°N in AMO− is consistent with the positive PDO phase—the jet stream and storm track tend to move southward, and during the negative PDO phase, they tend to more northward (toward the anomalous meridional SST gradient) (Sung et al. 2014). There is strong evidence for such meridional shifts in the upper-level westerly wind here. A clear southward shift over the North Pacific in response to AMO−, independent of sea ice state, can be seen (Fig. 8), supporting previous literature. Sea ice anomalies in the Sea of Okhotsk can also affect the position and intensity of the Pacific storm track (Honda et al. 1999). The results suggest that the background ocean–atmosphere state in the North Pacific modulates the circulation response to sea ice loss. This leads to differing patterns of lower-tropospheric temperature (in response to sea ice loss) under AMO− and AMO+ and, in turn, different lower-tropospheric meridional temperature gradient responses. Different response patterns around the midlatitude frontal zone are expected to be particularly effective in further influencing the atmospheric response in the Pacific–North American sector.
The 300-hPa zonal wind response to AMO−, independent of sea ice state [(AMO− − AMO+)LICE,HICE]. Hatching shows responses that are not statistically significant at the 95% (p = 0.05) confidence level. The response is shown from 20° to 90°N.
Citation: Journal of Climate 30, 5; 10.1175/JCLI-D-16-0531.1
Sea ice loss acts to reduce the 850-hPa meridional temperature gradient across much of the central–western North Pacific, north of about 35°N, during both phases of the AMO (Figs. 9a,b; note that gradient is defined in the equatorward direction so that the climatological gradient is positive). The reduction is particularly strong just south of the Sea of Okhotsk, where there is a large decrease in SIC between the LICE and HICE experiments. Changes in the meridional temperature gradient have a greater impact when they are located close to the climatological maximum meridional temperature gradient (Kidston et al. 2011). The jet stream and storm track have been shown to move equatorward when warming is confined poleward of the climatological maximum meridional temperature gradient, as is the case in response to sea ice loss here, because of decreasing baroclinicity associated with these midlatitude frontal zones (Chen et al. 2010). We expect that small perturbations to the meridional temperature gradient close to the climatological maximum could lead to different responses to sea ice loss under AMO− and AMO+.
Wintertime 850-hPa meridional temperature gradient response to Arctic sea ice loss across the central–western North Pacific (20°–55°N, 120°E–160°W) during (a) AMO− [(LICE − HICE)AMO−], (b) AMO+ [(LICE − HICE)AMO+], and (c) their difference [(LICE − HICE)AMO− − (LICE − HICE)AMO+]. Contours show the climatological meridional temperature gradient (average across all four experiments) and are drawn at intervals of 2°C (1000 km)−1. The gradient is defined in the equatorward direction, so a positive response indicates an increased (more negative) equator-to-pole meridional temperature gradient. The green box in (c) denotes the region used for an index of wintertime central North Pacific 850-hPa meridional temperature gradient.
Citation: Journal of Climate 30, 5; 10.1175/JCLI-D-16-0531.1
The anomalous northerly flow extending from the central Arctic in response to sea ice loss (independent of AMO phase) (Fig. 6a) and the anomalously strong Aleutian low in response to AMO− (independent of sea ice state) (Fig. 6b) cause an anomalous draw of cold Arctic air to the central midlatitude North Pacific in the LICE/AMO− experiment, around 30°N, 180°. Anomalous cold air advection to this region causes an increase in the 850-hPa meridional temperature gradient to the east and south of the climatological maximum, opposing the pan–North Pacific decreases caused by sea ice loss (Fig. 9a). There is a significantly different gradient response to sea ice loss here under opposite AMO phases (Fig. 9c). This favors a southward-shifted storm track in the central midlatitude North Pacific. This southward shift in the storm track can cause further strengthening of the Aleutian low (Trenberth and Hurrell 1994; Zhang and Delworth 2007), which in turn supports the advection of cold Arctic air on its western flank. Such a feedback mechanism could explain the anomalously strong Aleutian low in response to Arctic sea ice loss under AMO−. In contrast, the meridional temperature gradient in the western North Pacific increases in the East China Sea and south of Japan in both AMO phases, because of the warm Arctic–cold Eurasia response. Because this pattern is stronger in AMO+, the increases in the meridional temperature gradient are greater and significant here in this phase. This is also reflected in the meridional temperature gradient response difference (Fig. 9c). This favors increases in the upper-level westerly wind in response to sea ice loss in AMO+ to the south of the wintertime climatological maximum (Fig. 5h) and would seem to deepen the East Asian trough (Fig. 5b), but, unlike in AMO−, this does not force a downstream response. Near-identical results are found when considering 925-hPa meridional temperature gradient responses (not shown).
c. Observed meridional temperature gradient
We use global air temperature from the National Centers for Environmental Prediction (NCEP)–National Center for Atmospheric Research (NCAR) reanalysis (Kalnay et al. 1996) to create an index of “observed” wintertime central North Pacific (Fig. 9c, green box; averaged over 20°–35°N, 160°E–160°W) 850-hPa meridional temperature gradient (1948–2013). Figure 10b shows this index regressed against observed wintertime 500-hPa geopotential height, also from the reanalysis. Data are detrended to remove any global warming signal. There is evidence of a TRT pattern that looks very similar to that seen in the difference in the response to sea ice loss under the opposite AMO phases (cf. Figs. 5c and 10b). Although this does not show that changes in the meridional temperature gradient here lead changes in the Aleutian low, it does show the close association of changes in the Aleutian low with circulation anomalies over North America. A near-identical pattern is found using our model simulations (Fig. 10a).
Wintertime 500-hPa geopotential height regressed on an index of wintertime central North Pacific (Fig. 9c, green box; averaged over 20°–35°N, 160°E–160°W) 850-hPa meridional temperature gradient for (a) the model and (b) the observations. All four experiments (400 ensemble members) are used in the model regression with the climatology removed from each individual experiment. For the observed regression, the 1948–2013 period is used, with the climatology removed and data detrended to remove any global warming signal.
Citation: Journal of Climate 30, 5; 10.1175/JCLI-D-16-0531.1
4. Discussion and conclusions
The aim of this study was to determine whether the response to Arctic sea ice loss is conditioned by the background ocean–atmosphere state. We have devised a unique experimental design, forcing four atmosphere-only model simulations with low or high sea ice states in combination with negative or positive AMO-related SST anomalies. This allows us to isolate components of the atmospheric response that are dependent on the AMO phase.
Although the AMO-related SST anomalies are most prominent in the North Atlantic basin, as would be expected, the most evident modulation of the wintertime atmospheric response to sea ice loss by the AMO appears over the Pacific–North American sector. In the negative phase of the AMO, there is a trough–ridge–trough (TRT) pattern here that is not apparent in the positive phase of the AMO. This pattern is evident throughout the troposphere, with implications for the wintertime climatology from the near-surface to jet stream altitudes. We attribute these differences to an anomalous northerly flow, extending from the central Arctic to the central midlatitude North Pacific in the LICE/AMO− experiment. This anomalous northerly flow is seen in the response to Arctic sea ice loss alone, independent of AMO phase, as a southward departure of the pan-Arctic anticyclonic anomaly in the northwestern North Pacific. This anomalous northerly flow is reinforced by an anomalously strong Aleutian low in the negative phase of the AMO. Cold air advection to the central midlatitude North Pacific, close to the climatological maximum meridional temperature gradient, is likely associated with a southward shift in the baroclinic zone and associated storm track. This can further strengthen the Aleutian low, generating the TRT pattern over the Pacific–North American sector. Figure 11 shows the hypothesized mechanism behind the modulation of the atmospheric response to sea ice loss by AMO phase. In the experiment forced with the opposite combination of sea ice/SST anomalies (HICE/AMO+), an anomalously weak Aleutian low in positive AMO and an anomalous cyclonic anomaly in the northwestern Pacific in a high ice state might be expected to force the opposite response. We note, however, that in this setup any warm air advection on the western flank of the Aleutian low combines with cold continental air from northeastern Asia, which is transported over the Sea of Okhotsk in a high ice state.
Schematic of the hypothesized mechanism behind the differing responses to sea ice loss under AMO− and AMO+.
Citation: Journal of Climate 30, 5; 10.1175/JCLI-D-16-0531.1
It is perhaps counterintuitive that the most prominent influence of (multi-) decadal Atlantic variability on the response to Arctic sea ice loss is found in the Pacific region. This is clearly a consequence of our prescribed AMO-like SST anomalies (Fig. 2). Recall that we regress the detrended and normalized annual-mean AMO index against detrended annual-mean global SST to determine an SST anomaly per standard deviation departure in the AMO index. These anomalies are not limited to the Atlantic basin. Given the short observed record used in the regression (66 yr), if the AMO and PDO show similarly timed phase changes by chance, then North Pacific SST anomalies will be evident. Indeed, Zhang and Delworth (2007) showed that phase changes in the PDO lag phase changes in the AMO by about a decade. The authors proposed a mechanism in which the North Pacific responds to the North Atlantic through atmospheric teleconnections, with local dynamics and feedbacks playing a role. Li et al. (2016) propose another mechanism for Atlantic-induced tropical Pacific climate change. Further, Wang et al. (2014) showed that cold SST biases in the North Atlantic, which project onto AMO−, correspond with cold SST biases in the North Pacific, which project onto PDO+ and a strengthened Aleutian low, across the latest coupled climate models. Decadal prediction systems are also capable of showing a lagged North Pacific SST response (a shift to PDO−) to rapid warming in the North Atlantic (a shift to AMO+) (Robson et al. 2013).
The AMO and PDO have been (broadly) out of phase between the years considered here (1948–2013). Negative anomalies in the central–western North Pacific, which are prescribed in the negative phase of the AMO, are a common signature of the positive phase of the PDO. It is impossible to say if the North Pacific SST anomalies are forced by the AMO or are simply due to sampling. Many of the studies that suggest an influence of North Atlantic SSTs on North Pacific SSTs are keen to stress that a robust physical mechanism behind this apparent linkage does not yet exist. In the meantime, we have no reason to exclude North Pacific SST anomalies in our experimental design, with both statistical and modeling evidence suggesting that the true pattern of AMO extends beyond the Atlantic basin. In making this choice, we find unexpected but interesting results, suggesting a greater sensitivity of the atmospheric response to Arctic sea ice loss to the SST pattern in the North Pacific than in the North Atlantic. Foremost, it also shows that the atmospheric response to Arctic sea ice loss is dependent on the background ocean–atmosphere state. Both of these findings may have been missed in an experimental design restricting prescribed SST anomalies to just the North Atlantic basin. Future work should explore this, with further experiments limiting AMO-like SSTs to just the North Atlantic basin and just the North Pacific basin in turn.
By using an AGCM, we do not include a representation of coupled atmosphere–ocean interactions, which could have an influence on results. For example, oceanic feedbacks have been shown to amplify the midlatitude westerly wind response to projected Arctic sea ice loss by around 50% (Deser et al. 2016). The spatial pattern of the response to Arctic sea ice loss in a fully coupled ocean–atmosphere–ice experiment has been shown to be noticeably different from that in atmosphere-only experiments (Petrie et al. 2015). Here, one might expect an intensified Aleutian low to cause an anomalous North Pacific Subpolar Gyre circulation (Nakamura et al. 1997), with cold water transfer to the midlatitudes increasing the meridional SST gradient around the region of the KOE (Schneider and Cornuelle 2005; Zhang and Delworth 2007). This may act as a positive feedback, reinforcing lower-tropospheric increases in the meridional temperature gradient. Also, (for example) advecting cold air southward over warm ocean water causes increases in surface sensible heat flux, which would lead to cooler SSTs. Therefore, we expect these results could hold in a coupled atmosphere–ocean model but with the potential for even greater amplification of the TRT pattern in the Pacific–North American sector.
The PDO shifted from a negative phase to a positive phase during 2014. As a result, we might propose that the TRT pattern will be more prevalent in the coming years as sea ice diminishes. The perturbations from SIC and SST in our experiments, however, were at the extremes of what has been observed. Also, in a given winter season, SIC anomalies may display considerable spatial variability. For example, the sea ice states of the Bering Sea and Sea of Okhotsk, which are probably key in forcing the southward departure of the pan-Arctic anticyclonic anomaly in the northwestern North Pacific in LICE, may actually oppose one another. Therefore, it is difficult to relate these findings to observations. Here, we show that changes in the central North Pacific meridional temperature gradient are tightly coupled to the TRT pattern in a reanalysis and our model (Fig. 10). This does not, however, lend full support to the positive feedback that we suggest is behind the nonlinearity in the North Pacific. This is because cause and effect are unclear in observations. Nevertheless, we have shown that the background ocean–atmosphere state in the North Pacific, which displays persistence on multiyear time scales, should be considered when predicting the atmospheric response to sea ice loss. This could have implications for seasonal-to-decadal climate predictability, helping to separate forced responses from internal variability.
Acknowledgments
Three anonymous reviewers are thanked for reviews that helped improve the manuscript. This work was supported by the Natural Environment Research Council Grants NE/M006123/1 and NE/J019585/1. The HadGAM2 simulations were performed on the ARCHER U.K. National Supercomputing Service. The Met Office Hadley Centre and NOAA/ESRL are thanked for providing observed and reanalysis data. Model data are available from the authors upon request.
REFERENCES
Alexander, M. A., K. H. Kilbourne, and J. A. Nye, 2014: Climate variability during warm and cold phases of the Atlantic multidecadal oscillation (AMO) 1871–2008. J. Mar. Syst., 133, 14–26, doi:10.1016/j.jmarsys.2013.07.017.
Ayarzagüena, B., and J. A. Screen, 2016: Future Arctic sea ice loss reduces severity of cold air outbreaks in midlatitudes. Geophys. Res. Lett., 43, 2801–2809, doi:10.1002/2016GL068092.
Balmaseda, M. A., L. Ferranti, F. Molteni, and T. N. Palmer, 2010: Impact of 2007 and 2008 Arctic ice anomalies on the atmospheric circulation: Implications for long-range predictions. Quart. J. Roy. Meteor. Soc., 136, 1655–1664, doi:10.1002/qj.661.
Bond, N. A., and D. E. Harrison, 2000: The Pacific decadal oscillation, air–sea interaction and central North Pacific winter atmospheric regimes. Geophys. Res. Lett., 27, 731–734, doi:10.1029/1999GL010847.
Chen, G., R. A. Plumb, and J. Lu, 2010: Sensitivities of zonal mean atmospheric circulation to SST warming in an aqua-planet model. Geophys. Res. Lett., 37, L12701, doi:10.1029/2010GL043473.
Cohen, J., and Coauthors, 2014: Recent Arctic amplification and extreme mid-latitude weather. Nat. Geosci., 7, 627–637, doi:10.1038/ngeo2234.
Collins, M., and Coauthors, 2013: Long-term climate change: Projections, commitments and irreversibility. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 1029–1136.
Comiso, J. C., 2012: Large decadal decline of the Arctic multiyear ice cover. J. Climate, 25, 1176–1193, doi:10.1175/JCLI-D-11-00113.1.
Deser, C., R. Tomas, M. Alexander, and D. Lawrence, 2010: The seasonal atmospheric response to projected Arctic sea ice loss in the late twenty-first century. J. Climate, 23, 333–351, doi:10.1175/2009JCLI3053.1.
Deser, C., R. Tomas, and L. Sun, 2015: The role of ocean–atmosphere coupling in the zonal-mean atmospheric response to Arctic sea ice loss. J. Climate, 28, 2168–2186, doi:10.1175/JCLI-D-14-00325.1.
Deser, C., L. Sun, R. A. Tomas, and J. Screen, 2016: Does ocean coupling matter for the northern extratropical response to projected Arctic sea ice loss? Geophys. Res. Lett., 43, 2149–2157, doi:10.1002/2016GL067792.
Enfield, D. B., A. M. Mestas-Nuñez, and P. J. Trimble, 2001: The Atlantic multidecadal oscillation and its relation to rainfall and river flows in the continental U.S. Geophys. Res. Lett., 28, 2077–2080, doi:10.1029/2000GL012745.
Francis, J. A., and S. J. Vavrus, 2012: Evidence linking Arctic amplification to extreme weather in mid-latitudes. Geophys. Res. Lett., 39, L06801, doi:10.1029/2012GL051000.
Frankignoul, C., N. Sennéchael, Y.-O. Kwon, and M. A. Alexander, 2011: Influence of the meridional shifts of the Kuroshio and the Oyashio Extensions on the atmospheric circulation. J. Climate, 24, 762–777, doi:10.1175/2010JCLI3731.1.
Honda, M., K. Yamazaki, H. Nakamura, and K. Takeuchi, 1999: Dynamic and thermodynamic characteristics of atmospheric response to anomalous sea-ice extent in the Sea of Okhotsk. J. Climate, 12, 3347–3358, doi:10.1175/1520-0442(1999)012<3347:DATCOA>2.0.CO;2.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471, doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.
Kay, J. E., M. M. Holland, and A. Jahn, 2011: Inter-annual to multi-decadal Arctic sea ice extent trends in a warming world. Geophys. Res. Lett., 38, L15708, doi:10.1029/2011GL048008.
Kidston, J., A. S. Taschetto, D. W. J. Thompson, and M. H. England, 2011: The influence of Southern Hemisphere sea-ice extent on the latitude of the mid-latitude jet stream. Geophys. Res. Lett., 38, L15804, doi:10.1029/2011GL048056.
Kug, J.-S., J.-H. Jeong, Y.-S. Jang, B.-M. Kim, C. K. Folland, S.-K. Min, and S.-W. Son, 2015: Two distinct influences of Arctic warming on cold winters over North America and East Asia. Nat. Geosci., 8, 759–762, doi:10.1038/ngeo2517.
Li, X., S.-P. Xie, S. T. Gille, and C. Yoo, 2016: Atlantic-induced pan-tropical climate change over the past three decades. Nat. Climate Change, 6, 275–279, doi:10.1038/nclimate2840.
Martin, G. M., and Coauthors, 2011: The HadGEM2 family of Met Office Unified Model climate configurations. Geosci. Model Dev., 4, 723–757, doi:10.5194/gmd-4-723-2011.
Min, S.-K., X. Zhang, F. W. Zwiers, and T. Agnew, 2008: Human influence on Arctic sea ice detectable from early 1990s onwards. Geophys. Res. Lett., 35, L21701, doi:10.1029/2008GL035725.
Mori, M., M. Watanabe, H. Shiogama, J. Inoue, and M. Kimoto, 2014: Robust Arctic sea-ice influence on the frequent Eurasian cold winters in past decades. Nat. Geosci., 7, 869–873, doi:10.1038/ngeo2277.
Nakamura, H., G. Lin, and T. Yamagata, 1997: Decadal climate variability in the North Pacific during the recent decades. Bull. Amer. Meteor. Soc., 78, 2215–2225, doi:10.1175/1520-0477(1997)078<2215:DCVITN>2.0.CO;2.
Petrie, R. E., L. C. Shaffrey, and R. T. Sutton, 2015: Atmospheric impact of Arctic sea ice loss in a coupled ocean–atmosphere simulation. J. Climate, 28, 9606–9622, doi:10.1175/JCLI-D-15-0316.1.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.
Robson, J., R. Sutton, and D. Smith, 2013: Predictable climate impacts of the decadal changes in the ocean in the 1990s. J. Climate, 26, 6329–6339, doi:10.1175/JCLI-D-12-00827.1.
Rodionov, S. N., N. A. Bond, and J. E. Overland, 2007: The Aleutian low, storm tracks, and winter climate variability in the Bering Sea. Deep-Sea Res., 54, 2560–2577, doi:10.1016/j.dsr2.2007.08.002.
Schneider, N., and B. D. Cornuelle, 2005: The forcing of the Pacific decadal oscillation. J. Climate, 18, 4355–4373, doi:10.1175/JCLI3527.1.
Screen, J. A., and I. Simmonds, 2010: The central role of diminishing sea ice in recent Arctic temperature amplification. Nature, 464, 1334–1337, doi:10.1038/nature09051.
Screen, J. A., and J. A. Francis, 2016: Contribution of sea-ice loss to Arctic amplification is regulated by Pacific Ocean decadal variability. Nat. Climate Change, 6, 856–860, doi:10.1038/nclimate3011.
Screen, J. A., I. Simmonds, C. Deser, and R. Tomas, 2013: The atmospheric response to three decades of observed Arctic sea ice loss. J. Climate, 26, 1230–1248, doi:10.1175/JCLI-D-12-00063.1.
Screen, J. A., C. Deser, I. Simmonds, and R. Tomas, 2014: Atmospheric impacts of Arctic sea-ice loss, 1979–2009: Separating forced change from atmospheric internal variability. Climate Dyn., 43, 333–344, doi:10.1007/s00382-013-1830-9.
Screen, J. A., C. Deser, and L. Sun, 2015: Reduced risk of North American cold extremes due to continued Arctic sea ice loss. Bull. Amer. Meteor. Soc., 96, 1489–1503, doi:10.1175/BAMS-D-14-00185.1.
Semenov, V. A., and M. Latif, 2015: Nonlinear winter atmospheric circulation response to Arctic sea ice concentration anomalies for different periods during 1966–2012. Environ. Res. Lett., 10, 054020, doi:10.1088/1748-9326/10/5/054020.
Serreze, M. C., A. P. Barrett, J. C. Stroeve, D. N. Kindig, and M. M. Holland, 2009: The emergence of surface-based Arctic amplification. Cryosphere, 3, 11–19, doi:10.5194/tc-3-11-2009.
Steele, M., W. Ermold, and J. Zhang, 2008: Arctic Ocean surface warming trends over the past 100 years. Geophys. Res. Lett., 35, L02614, doi:10.1029/2007GL031651.
Strey, S. T., W. L. Chapman, and J. E. Walsh, 2010: The 2007 sea ice minimum: Impacts on the Northern Hemisphere atmosphere in late autumn and early winter. J. Geophys. Res., 115, D23103, doi:10.1029/2009JD013294.
Stroeve, J. C., V. Kattsov, A. Barrett, M. Serreze, T. Pavlova, M. Holland, and W. N. Meier, 2012a: Trends in Arctic sea ice extent from CMIP5, CMIP3 and observations. Geophys. Res. Lett., 39, L16502, doi:10.1029/2012GL052676.
Stroeve, J. C., M. C. Serreze, M. M. Holland, J. E. Kay, J. Malanik, and A. P. Barrett, 2012b: The Arctic’s rapidly shrinking sea ice cover: A research synthesis. Climatic Change, 110, 1005–1027, doi:10.1007/s10584-011-0101-1.
Sung, M.-K., S.-I. An, B.-M. Kim, and S.-H. Woo, 2014: A physical mechanism of the precipitation dipole in the western United States based on PDO-storm track relationship. Geophys. Res. Lett., 41, 4719–4726, doi:10.1002/2014GL060711.
Trenberth, K. E., and J. W. Hurrell, 1994: Decadal atmosphere–ocean variations in the Pacific. Climate Dyn., 9, 303–319, doi:10.1007/BF00204745.
Wang, C., L. Zhang, S.-K. Lee, L. Wu, and C. R. Mechoso, 2014: A global perspective on CMIP5 climate model biases. Nat. Climate Change, 4, 201–205, 10.1038/nclimate2118.
Wang, M., and J. E. Overland, 2009: A sea ice free summer Arctic within 30 years? Geophys. Res. Lett., 36, L07502, doi:10.1029/2009GL037820.
Wang, M., and J. E. Overland, 2012: A sea ice free summer Arctic within 30 years: An update from CMIP5 models. Geophys. Res. Lett., 39, L18501, doi:10.1029/2012GL052868.
Zhang, R., and T. L. Delworth, 2007: Impact of the Atlantic multidecadal oscillation on North Pacific climate variability. Geophys. Res. Lett., 34, L23708, doi:10.1029/2007GL031601.
