Persistent cold weather spells have considerable social and economic impacts. The focus of the present paper is on cold spells that last longer than 10 days and affect a large area covering the central–eastern part of North America (NA)'s highly populated midlatitudes. A persistent cold event can occur when the circulation pattern of the atmosphere near the affected region stays in a state that favors cold air flowing southward from the Arctic. Cold weather spells in central–eastern NA during winters are often associated with cold air outbreaks during a weak tropospheric polar vortex (negative Arctic Oscillation (AO)) or a strong quasi-stationary wave with a proper phase to perturb the polar vortex and to position the Arctic air over the midlatitudes of NA (Higgins et al., 2002;Jung et al., 2011;Smith and McDonald, 2014).

Owing to the existence of the climatological trough in the eastern United States and Canada, the temperatures are lower for the same latitudes in eastern NA than those in western NA. The eastern and central United States and Canada have many large metropolitan regions with large populations (Toronto, Montreal, New York, Washington D. C., etc.). Persistent cold spells also cause much higher energy consumption for heating than on normal winter days, as well as severe disruption to southern regions with usually mild winters as they are not as accustomed to the snow and freezing rain accompanying cold surges. With these considerations in mind, we choose a region east of the Rockies from near the Gulf of Mexico to Hudson Bay to investigate persistent cold spells in central–eastern NA.

The Arctic has warmed more than the midlatitudes during the past several decades. This Arctic amplification is especially prominent in the lower atmosphere near the surface, which decreases the meridional temperature gradient in the lower troposphere. Researchers (Francis and Vavrus, 2012;Screen and Simmonds, 2013) have proposed the possibility that global warming could weaken the tropospheric polar vortex in a manner that makes the jet stream flowing around it wavier, leading to an increased chance of both extreme warmth and extreme coldness in the Northern Hemisphere midlatitudes. According to their hypothesis, in a weakened meridional temperature gradient, Rossby or planetary waves tend to progress more slowly with larger north–south meandering of the jet stream in a reduced meridional gradient of potential vorticity and zonal winds (Francis and Vavrus, 2012), though both observational analyses and modeling studies are needed to support this hypothesis. Aside from the AO's important role in winter weather in NA, tremendous natural variability occurs in the large-scale atmospheric circulation. For instance, the tropical Pacific has been shown to affect polar and midlatitude weather in NA. Winter cold spells in central–eastern NA can simply be manifestations of strong quasi-stationary wave events over NA, with their sources in the tropical or extratropical Pacific. For example, in the winter of 2014–2015, a tropical cyclone migrated into the extratropical North Pacific and formed a strong negative anomaly of geopotential height (GPH) in the central Pacific near the dateline, which generated a strong wave train downstream that eventually brought a cold air outbreak in midlatitude NA.

The ocean's impacts on atmospheric circulation have been investigated extensively, because of their seasonal forecast potential for the climate system due to the ocean's long-term “memory” and long adjustment time relative to the atmosphere. Sea surface temperature (SST) anomalies in the tropical Pacific have a significant influence on climate over NA. The El Niño–Southern Oscillation (ENSO) is a coupled ocean–atmosphere phenomenon and is the largest mode of interannual variability in the global ocean, providing useful information for seasonal forecasts of weather and climate conditions. Unlike the tropical Pacific Ocean, where interannual variability dominates, most of the North Pacific SST variances occur in a mode on mostly decadal timescales, i.e., the Pacific Decadal Oscillation (PDO) (Bond and Harrison, 2000). The PDO is classically defined as the first empirical orthogonal function (EOF) of the anomalies of monthly mean SST north of 20°N in the Pacific basin. It is a decadal oscillation that features the cooling of the central part of extratropical Pacific SST and a warm SST anomaly in the surrounding “horseshoe” shape regions in its positive phase. This mode of SST variation is accompanied by atmospheric variability, particularly in the strength and position of the sea-level Aleutian low during winter and changes in the Pacific storm track (Frankignoul and Sennéchael, 2007).Bond et al. (2003) found that the variability of the North Pacific SST is not sufficiently represented by the PDO alone; an additional component, the North Pacific Mode (NPM), explains about the same percentage of variance (Hartmann, 2015) as the PDO. The winter of 2013–2014 was an extremely cold winter for the eastern part of the United States and Canada. There were three main episodes of extreme cold relative to the climatology. The large monthly mean cold anomalies for January 2014 were caused by two cold spells running into each other with a gap of about 10 days between, rather than an extremely persistent event lasting for an extended period. The polar vortex for this month was strong, as indicated by the AO index. The ENSO index was generally neutral during December 2013 and January 2014 and evolved into a weak negative ENSO index in February 2014.Hartmann (2015) found that the cold winter in central NA was likely associated with a strong positive phase of the NPM of SST variability, though it is difficult in this case to discern whether the ocean surface temperature anomalies affected the atmosphere, or the other way around.

The goal of the present study is to survey historical persistent cold spells in central–eastern NA, identify circulation patterns that are related to them, and explore the possible links between various atmospheric circulations and SST modes with these cold spells. Following this introduction, the data and methods used to identify persistent cold spells are described in Section 2. Then, the statistical results with respect to the cold spells and their weekly evolution under different circulation patterns are presented and discussed in Sections 3 and 4, respectively. Section 5 summarizes the key findings.

The daily GPH, wind, and surface air temperature data are from the NCEP reanalysis from 1948 to 2014 (Kalnay et al., 1996). The monthly SST data are from the extended reconstructed SST version 3b from the National Oceanic and Atmospheric Administration/National Climatic Data Center (NOAA/NCDC) (Smith et al., 2008). To identify extended cold spells, the following criteria are set: the regional average temperature anomaly over (32°–52°N, 95°–65°W) during winter months (December, January, February) must be below 10th percentile for 10 consecutive days or longer. By assuming the distribution of temperature anomalies are approximately Gaussian, the 10th percentile criteria are set to be 1.28 standard deviations below the mean value. We focus on the central and eastern part of midlatitude NA where is climatologically colder. For the most part, its temperature/GPH varies in the opposite direction to that of the western United States–Canada region. This region also encompasses many large metropolitan regions: New York, Chicago, Toronto, and Washington D. C., to name a few. Three-day averaging instead of daily temperature is adopted to smooth out the variation due to transient eddies, because we are interested in timescales longer than synoptic weather. The considerations for the chosen region, as mentioned before, include the following factors: the climatological trough over eastern NA makes that part of the midlatitudes colder than the same latitudes in the west of the continent; the largest variational pattern of NA is the Pacific NA (PNA) pattern, which has activity centers of opposite sign for east and west NA; and this highly populated area in NA is the region most economically affected by extended cold periods during winter. Climate indices such as multivariate ENSO index (MEI), AO index, PDO, and PNA index are retrieved from the NOAA/NCDC website ( http://www.ncdc. noaa.gov). The NPM is calculated as the third EOF of winter monthly SST in the Pacific basin from 30°S to 65°N.

Assessment of the winters from 1948 to 2014 shows that persistent cold anomalies in central–eastern NA are often associated with a weaker polar vortex or a strong quasi-stationary wave over NA. However, an exceptionally cold winter does not need the atmosphere to stay in the same status for the whole winter, either in terms of the polar vortex or quasi-stationary wave situation. In fact, the monthly mean GPH anomaly is often caused by a strong anomalous event that lasts on a timescale of around one week, or several such anomalous events running into each other (Lindzen, 1986). Often, the circulation pattern, such as the AO or PNA indices, can be in opposite phase during a month that also witnesses a strong cold weather event. Except on rare occasions, monthly mean anomalies show a damped or even distorted picture of an anomalous event depending on the variability within a month. By examining the daily evolution of cold spells, we can separate the spells that are caused by persistent large-scale circulation anomalies, and those occur only by chance. Thus, analyses based on daily/weekly composites are more meaningful than monthly means when we try to investigate the detailed processes involved.

A composite of all the cold spells is made to identify the common characteristics. As shown in Fig. 1, there is a strong anomalous trough over eastern NA at 200 hPa, which overlays on top of the climatological trough. To the west of this anomalous trough is an anomalous ridge over the western coast of midlatitude NA and Alaska. Upstream of this anomalous ridge is an anomalous trough in the central Pacific. The anomaly pattern in GPH resembles a quasi-stationary wave pattern that initially occurs in the central Pacific and then propagates eastward. The second common feature of these events is weaker than normal zonal flow in the midlatitudes from the central Pacific to North Atlantic.

Fig. 1 Mean geopotential height anomaly (gpm) of the persistent cold spells over central–eastern North America.

Figure options

Our chosen area lies within the strong negative GPH anomaly and a large positive area covers the western coast of Canada, Alaska and into the Arctic Ocean north of the Bering Strait. There is also a weaker positive anomaly stretching along the Pacific coast of the United States. A smaller strong positive anomaly area is centered on the southwest coast of Greenland. It turns out that the cold spells associated with a weaker tropospheric polar vortex or negative AO often have strong positive GPH anomalies near Greenland, in addition to an anomalous ridge over the midlatitude Pacific coast of NA. Whereas, persistent cold spells associated with a strong polar vortex usually only have a single strong positive GPH anomaly in the North Pacific sector near Alaska or the Bering Strait (figures omitted).

To investigate the favorable circulation conditions for persistent cold spells for our chosen region, the statistics for circulation indices during the cold spells are shown in Table 1. The persistent cold spells are categorized based on their mean circulation indices around the cold events. The period for the calculation of mean indices is 21 days starting from two weeks before a persistent cold event and one week during the cold event. The categorized cold spells based on circulation indices are independent of each other, i.e., the categorization is not exclusive for each index. The number of cases in each circulation condition and the total number of such conditions are listed in the table. This method of categorization of cold spells is also applied to the analyses and results later in the paper.

As expected, more persistent cold spells occur during negative AO phases. The AO and the North Atlantic Oscillation, which is the AO's manifestation in the North Atlantic region, are found to be strongly correlated with the surface air temperature in eastern NA and western Europe (Hurrell et al., 2003). The higher frequency of occurrence of persistent cold spells after and during a strongly negative AO phase is consistent with this positive correlation. Persistent cold spells during positive AO phases are associated with the positive phase of the PNA pattern, which has a strong anomalous upper-level ridge in the western United States–Canada region and an anomalous trough in the eastern part of the continent.

A positive-phase PNA pattern is more favorable for persistent cold spells than a negative-phase pattern, which is consistent with the fact that positive phases of the PNA pattern are associated with positive GPH anomalies in the western part of NA and negative GPH anomalies in the southeastern part of NA. This configuration of GPH anomalies in winter corresponds to a cold anomaly in surface temperature in the eastern United States and Canada.

Positive phases of both the PDO and the NPM favor the occurrence of persistent cold spells. These two modes of SST variation patterns are associated with a warm SST anomaly near the eastern extratropical Pacific near the coast of NA. The occurrence of the warm anomaly by the Pacific coast of NA is often concurrent with an upper-level anomalous ridge over western NA, which is favorable for northerly flow from the Arctic towards the midlatitudes of central–eastern NA.

Major oceanic and atmospheric patterns are known to affect how the atmosphere responds to forcing. For example, the response of the atmosphere to SSTs is strongly affected by the mean zonal wind distribution (Peng and Whitaker, 1999), and the mean wind distribution has a strong influence on quasi-stationary Rossby wave propagation and meridional dispersion. In this section, we categorize the persistent cold spells based on several climate indices, such as ENSO and AO, and investigate the differences of the persistent cold events in different phases of ENSO and AO.

The leading pattern of variability in the tropical Pacific is ENSO (Straus and Shukla, 2002), and this phenomenon is usually associated with suppressed convective rainfall and above-average air pressure across Indonesia, the western equatorial Pacific, and northern Australia during El Niño. The precipitation in the tropical central and eastern Pacific is enhanced during El Niño years.Figure 2 shows the regression of hemispheric 200-hPa GPH in winter months (December, January, and February), from NCEP reanalysis data during 1948–2015, on the MEI (Wolter and Timlin, 1993,1998). The regression pattern shows that El Niño is associated with higher than average GPHs in all sectors of the northern subtropics, especially over the central/eastern Pacific. Higher heights near the tropopause are typically associated with a warmer air column underneath because the atmosphere is in hydrostatic balance on a timescale longer than synoptic weather (DeWeaver and Nigam, 2004). The pattern is consistent with the warming of the tropics during El Niño in the Northern Hemisphere winter (Simmons et al., 1983). The subtropical ridge to the southeast of the Hawaiian Islands is connected to the development of an upper-level anticyclonic circulation in the subtropics that is associated with the equatorial divergent outflow during El Niño (DeWeaver and Nigam, 2004). Over NA, there is a ridge over the region between the Canadian Rockies and Hudson Bay and a trough south of Alaska. The pattern resembles the PNA pattern in the midlatitudes, which is often represented by the second leading principal component of the Northern Hemisphere's 500-hPa GPH. It explains the second largest percentage of variance after the AO by using rotated EOF analysis. A positive PNA phase is associated with a negative GPH anomaly in the southeastern United States and North Pacific, implying anomalous westerlies over the midlatitude northern Pacific by geostrophic balance. It is also associated with positive anomalies in the western United States–Canada region and the central Pacific near the Hawaiian Islands. This pattern amplifies the climatological trough in the eastern United States–Canada region. Hence, we would expect many cases of cold spells to be associated with positive PNA index values, as cold spells are coincident with a deepened trough in eastern NA. Though the PNA pattern is often associated with El Niño, it can also be generated by the internal variability of the atmosphere (Straus and Shukla, 2002).

Fig. 2 Regression of the winter monthly geopotential height anomaly (gpm) on the multivariate ENSO index. Shaded areas are statistically significant above the 90% confidence level according to the Student's t-test.

Figure options

In order to compare the development of the persistent cold spells during different phases of ENSO, we investigate the temporal evolution of these events during El Niño and La Niña. The El Niño phase of ENSO is defined as the period when the MEI is larger than 0.5, and the La Niña phase is when the MEI is less than –0.5. To compare the temporal evolution of GPH anomalies at 200 hPa for the two phases of ENSO, the composites of cold spells, from two weeks before to two weeks after the start of cold spells, are plotted in Figs. 3 and 5 for El Niño and La Niña, respectively.

Fig. 3 The evolution of geopotential height anomalies (shaded; gpm) at 200 hPa during El Niño for persistent cold spells over central–eastern North America.

Figure options

In Fig. 3, two weeks before the cold spells start, the composite GPH anomaly resembles the pattern shown in Fig. 2: a positive anomaly in the tropics, but concentrated in the central/eastern Pacific; a negative anomaly south of the Aleutian Islands; and a positive anomaly in northern Canada and near the pole. One week before a cold spell starts, the negative anomaly over the Aleutian Islands starts to intensify and the positive anomaly at the pole increases. When the cold spell begins, the anomalous trough in central and eastern NA becomes the largest negative anomaly. The configuration of a positive anomaly near the central/eastern Pacific and a negative anomaly south of the Aleutian Islands in the north seems to imply a Rossby wave source from the tropical heating. However, the quasi-geostrophic quasi-stationary wave propagation is originating from the strong negative anomaly south of the Aleutian Islands, north of Hawaii, as shown by the quasi-stationary wave flux vectors (Takaya and Nakamura, 2001) in Fig. 4. The stationary wave flux vectors are derived based on an approximate conservation of the wave-activity pseudomomentum for quasi-stationary eddies on a zonally varying basic flow, which is suitable for the diagnosis of circulation anomalies defined as departures from the climatological mean.

Fig. 4 Rossby wave fluxes at 250 hPa for one week before the cold spells under El Niño. Shaded regions are the composite geopotential height anomaly (gpm) scaled by the local Coriolis coefficient and are statistically significant above 90% confidence level according to the Student's t-test. The arrow to the lower right of the figure is the scale of the flux vector in units of m2 s–2.

Figure options

The La Niña phase composite of cold spells (Fig. 5) shows a drastically different scenario. Two weeks before the cold spells start, the anomaly patterns are generally of opposite sign to those of El Niño phase: negative in the east of Japan and central/eastern tropical Pacific; positive south of Alaska; negative in Canada and northern United States; and positive in southeastern United States. One week before a cold spell starts, the anomalous ridge along the west coast of NA strengthens. During the first week of the cold spell, the negative anomaly east of Japan moves to the south of the Aleutian Islands, as in the case of El Niño. At this stage, the difference between the two ENSO phases mainly resides in the tropics. Indeed, within the family of cold spells, 10-day intervals that are in the El Niño or La Niña phases could both have anomalies with the trough bringing frigid air over the studied region and a ridge over the western United States–Canada region. The quasi-geostrophic stationary wave propagation starts from the negative anomaly south of the Aleutian Islands one week before a cold spell, as shown by the top-right plot of Fig. 5 and the wave flux vectors in Fig. 6.

Fig. 5 As in Fig. 3, but for La Niña.

Figure options

Fig. 6 As in Fig. 4, but for La Niña.

Figure options

Both El Niño and La Niña phases show similar GPH anomaly patterns in the extratropics. The major difference lies in the tropical Pacific. The composite Rossby wave vectors in both cases show that the wave trains originate from the midlatitudes near the central Pacific, which implies that the cause of persistent cold spells in central–eastern NA is the internal variability of the atmosphere.

During the Northern Hemisphere winter, a large-scale teleconnection pattern, the AO, explains the largest fraction of temperature and GPH variance of any other known climate mode (Thompson et al., 2003). The daily AO index is constructed by projecting the daily 1000-hPa height anomalies polewards of 20°N onto the loading pattern (EOF) of the AO (Thompson et al., 2003). The Northern Annular Mode (NAM) is a similar index that represents a zonally symmetric mode of variability of GPH through the troposphere and the stratosphere (Thompson et al., 2003). The NAM for each isobaric level is constructed by finding the leading EOF of GPH during winter (Baldwin and Thompson, 2009). A positive NAM represents an above average height difference between the midlatitudes and polar region, and thus a strong polar vortex in the zonal mean sense.

We select cold spells during a strong polar vortex (AO index > 1) or weak polar vortex (AO < –1) to investigate the circulation anomalies associated with cold spells. The positive AO phase has a composite GPH anomaly pattern with a strong negative anomaly in the extratropical central Pacific and a positive anomaly over the west coast of NA one week before a cold spell starts, as shown in Fig. 7a. When this wave pattern extends another wave front with a strong negative anomaly into central and eastern NA, the cold spells start. In the case of the negative AO phase, as shown in Fig. 8a, two weeks before a cold spell starts, the positive anomaly resides in the polar region and weakens the polar vortex, with negative anomalies in the surrounding midlatitudes. As the negative anomaly in the northern central Pacific intensifies, the ridge over the west coast of NA and the trough in central and eastern NA are both enhanced, and the cold spell starts. The quasi-stationary wave propagation for the positive AO phase is more zonal between the 40° and 50° latitudinal circles than the negative phase of the AO. For the negative AO phase, the quasi-stationary wave first propagates from south of Alaska towards the pole, then propagates towards the midlatitudes. In Fig. 7a, the positive AO seems to hint at the influence of subtropical divergence associated with tropical heating in the central Pacific as a positive GPH anomaly locates near the dateline.

Fig. 7 (a) Evolution of geopotential height anomalies (shaded; gpm) at 200 hPa during AO > 1 for persistent cold spells over central–eastern North America. The shaded areas have a significance level less than 0.1 are statistically significant above the 90% confidence level according to the Student's t-test. (b) Composite of the Northern Annular Mode during AO > 1 for persistent cold spells, with the time axis (day) relative to the start date of the cold spells.

Figure options

Fig. 8 As in Fig. 7, but for AO < –1.

Figure options

Cold spells during a strong tropospheric polar vortex (AO > 1) usually have wave trains propagating from the subtropical Pacific to the west coast of NA at least by the time of one week before a cold spell starts. The maximum negative anomaly is in the south of Alaska, and the maximum positive anomaly is over the west coast of NA. By the time of the cold spell, the wave train has extended into the eastern United States–Canada region, with a node of the maximum negative anomaly in this region. During a strong polar vortex, the wave trains from the subtropical Pacific are important for initiating cold spells in the investigated area.

The strength of the stratospheric polar vortex, as depicted by the NAM, has been shown to affect lower tropospheric circulation after stratosphere sudden warming (SSW) (Baldwin and Dunkerton, 2001,Mitchell et al., 2013). Using the vortex breakdown dates as compiled by Charlton and Polvani (2007) and our own analysis for the years after 2007, we find that sudden stratospheric vortex breakdowns do not always precede persistent cold spells in central–eastern NA.Figure 8a shows that the cold spells in the troposphere occur after a positive AO and NAM in the stratosphere, which indicates that these cold spells have little connection with the stratospheric polar vortex condition.Figure 8b shows the composite NAM for negative AO type cold spells. There is a foreshadowing of a weak stratospheric polar vortex before the cold spells. However, the negative NAM also occurs in the lower troposphere. After all, the regional weather events are unlikely to be strongly associated with a zonally symmetric index such as the NAM. Further investigation with zonally asymmetric parameters is planned for future studies.

Persistent cold spells are often associated with strong quasi-stationary Rossby waves, as indicated by the composite upper-level GPH anomalies. Here, we examine the meridional dispersion condition of quasi-stationary Rossby waves in the PNA sector. Following the work by Held (1983), zonal wavenumbers of propagating quasi-stationary eddies from the central Pacific towards NA are calculated using NCEP reanalysis data (Kalnay et al., 1996).

As shown in Fig. 9, the focus is placed on the waveguide in the midlatitudes for zonal wavenumbers 3 and 4, which is the approximate scale of the quasi-stationary wave pattern associated with the persistent cold spells over central–eastern NA. The common feature of these cold events is that the waveguide is more confined to the PNA sector. For persistent cold spells under the AO > 1 condition, the favorable propagation channel is more to a 20° latitudinal band centered on 40°N. For the case of AO < –1, the waveguides are located in 40°–65°N, which is consistent with the fact that the GPH anomaly pattern for AO < –1 tends to have positive nodes in Alaska and Greenland.

Fig. 9 Zonal wavenumber of propagating quasi-stationary Rossby waves for persistent cold spells under (a) AO < –1 and (b) AO > 1 in the Pacific–North America sector (150°–30°W). Quasi-stationary waves are evanescent in the areas above the curves. The zonal wavenumber 4 and 3 lines are drawn to show the waveguide in the midlatitudes between these lines.

Figure options

Persistent cold spells in the central–eastern United States and Canada region from 1948 to 2014 are analyzed by using NCEP reanalysis data. The cold events are defined as a minimum of 10 days of continuous daily cold anomalies in the 10th percentile for the regional average over (32°–52°N, 95°–65°W). The SST anomaly and polar vortex conditions associated with these cold spells are examined. In particular, the circulation anomaly patterns of these events are categorized based on the ENSO, PDO, and AO (indicator of the tropospheric polar vortex) indices. The circulation patterns before and during cold events are identified through composite daily GPH anomaly maps for these cold spells. From these composites, a common feature is an anomalous ridge in western NA and a trough in eastern NA, which can be expected, with frigid air outpouring over our chosen area centered at (42°N, 80°W) in eastern Canada and the United States.

The mean climate indices during the period from two weeks before a cold spell starts (day 0) and one week after the cold spell starts are defined as the climate indices associated with this event. The consideration is to include the upstream information in our mean indices. For a certain persistent cold spell, the circulation condition can be vastly different. After all, many factors such as a weak tropospheric polar vortex, a strong quasi-stationary wave, or strong warm SST anomaly near the Pacific coast of NA, can lead to cold air outbreak/stagnation in eastern NA. Based on the catalogue of persistent cold spells for different climate indices (Table 1), the negative AO phase, positive PNA pattern, NPM, and PDO are more likely than their opposite phases to be associated with persistent cold spells. Phases of SST modes that are associated with warm SSTs (positive PDO and NPM) in the eastern extratropical Pacific favor persistent cold events in the investigated region. The circulation patterns and lower boundary conditions that precede persistent cold spells in central–eastern NA could be instrumental for subseasonal forecasting in further studies.

To examine the difference in the evolution of persistent cold spells in different ENSO and AO phases, we investigate the impact of these atmospheric circulation anomalies on the persistent cold weather events through the planetary wave propagation condition. ENSO, the most important interannual variability in the climate system, seems to only slightly favor a persistent cold spell during El Niño. During El Niño, the GPH anomaly pattern of a persistent cold spell resembles that of the PNA teleconnection pattern, implying a tropical Pacific connection. During La Niña, the composite wave pattern of cold spells indicates the internal variabilities in the extratropics. A weak tropospheric polar vortex, as indicated by an AO index less than –1, strongly favors persistent cold spells. When a cold spell happens, it occurs first in the upstream, in the central Pacific, for both AO phases. The main difference in the GPH anomaly pattern in the development stage between persistent cold spells under AO > 1 and AO < –1 is that the wave pattern for the latter travels a more polar route through Alaska than AO > 1.

A weak stratospheric vortex, as represented by a negative NAM, is not a good indicator of regional persistent cold spells in NA, because regional weather anomalies can happen under many circulation conditions that are not reflected by a zonally symmetric parameter. Thus, it is not a surprise to see SSW has little association with the persistent cold spells in central–eastern NA.

The AO phase affects the quasi-stationary wave propagation paths through modification of the background flow and meridional potential vorticity gradient. The quasi-stationary wave propagation condition in the PNA sector is investigated by using the meridional dispersion condition of quasi-stationary Rossby waves for AO < –1 and AO > 1. The waveguide for AO > 1 is in a narrow latitudinal band centered on 40°N, whereas for AO < –1 the waveguide is in a broader latitudinal band from 40° to 65°N.