1. Introduction
Exceptionally cold temperatures accompanied by strong winds swept southeastward from Canada into the U.S. Midwest, Great Lakes, and Northeast during 29 January to 1 February 2019. Daily cold temperature records were broken across the Midwest and into the Northeast with temperatures that competed with all-time record lows on a regional scale (Fig. 1). Mount Carroll, Illinois, for example, recorded the coldest ever state-wide temperature of −39°C (−38°F) on 30 January. At Chicago O’Hare International Airport (ORD) the temperature dropped to −31°C (−23°F) on 30 January, which tied for the fifth all-time coldest temperature in Chicago since 1871 (NWS 2019). The high temperature that same day was −23°C (−10°F), recorded at midnight, tying for the third coldest high on record. At 22°C (40°F) below the average temperature for the heart of the winter, 30 January was the second coldest day (average of the high and low) in recorded history for ORD. In addition, winds gusting to over 13 m s−1 (30 mph) made the temperatures feel even colder, producing a wind chill of −47°C (−52°F), which was the fifth lowest since 1929. By the afternoon of 31 January, ORD had spent 52 continuous hours below 0°F, ranking as the fourth longest duration. In summary, compared to the Arctic outbreak that impacted Chicago and the region in January 2014 (e.g., Screen et al. 2015), the 2019 event was superlative in many categories.
Coldest low temperatures (°C) observed between 29 Jan and 1 Feb 2019, from the Global Historical Climate Network (GHCN). The small-circle stippling signifies temperatures that are on average the coldest per year (0.27th percentile of a station’s history) and the large-diamond stippling signifies temperatures that are the coldest per 10 years (0.027th percentile).
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
Cold air outbreaks (CAOs) pose significant hazards in the United States. Much attention has been given to CAOs impacting the Southeast United States due to damage to agriculture, such as Florida citrus crops (Rogers and Rohli 1991; Miller and Downton 1993; Downton and Miller 1993). Extreme cold with CAOs in the northern United States has direct impact on human well-being, threatening life, and property. For instance, the wind chills observed in January 2019 can cause frostbite in under five minutes. The extreme conditions were implicated in 21 fatalities across the Midwest and Northeast (Gajanan 2019). These conditions also resulted in closure of schools and business and disruptions to transportation. Both Wisconsin and Michigan declared a state of emergency. Thus, this recent event and its impacts provide the motivation for further investigation of the meteorological factors responsible.
There have been a variety of definitions for CAOs in past studies. Konrad and Colucci (1989) defined CAOs as minima in a time series of 850 hPa temperatures averaged over the eastern United States, while Konrad (1996) used surface temperature over the Southeast. Walsh et al. (2001), Vavrus et al. (2006), and Wheeler et al. (2011) all use similar definitions based on normalized temperature anomalies for a given space or time threshold. See specifically Vavrus et al. (2006) for a discussion on the factors considered. Wheeler et al. (2011) applied their definition to the ERA-40 dataset, and found a maximum frequency in CAOs from British Columbia and Alberta southeastward into the plains of the United States. Wheeler et al. (2011) also found that the average intensity of these CAOs was maximized in the same region of Canada and into the northern plains and midwestern United States, with a local minimum in intensity over the Great Lakes region. The common theme of the various definitions used in these past studies is that anomalously cold temperatures occur on synoptic space and time scales. As such, CAOs were generally assumed to be connected to synoptic-scale atmospheric features (e.g., cyclones, anticyclones, fronts).
However, as part of a general desire to attribute extreme cold to some larger-scale atmospheric entity beyond low and high pressure systems, the 2014 Arctic outbreak into central and eastern North America popularized the term “polar vortex” into public vernacular through national and global media. This new catch-all for cold winter temperatures returned in January 2019. The public use of this term is often a misconception, since “polar vortex” is the name given to the radiatively driven climatological cyclonic circulation that resides in the high latitudes of the winter stratosphere (Palmer 1959). Meanwhile, the tropospheric polar vortex may refer to the baroclinically driven polar jet stream (PJS) that roams around the middle latitudes and exhibits much greater variability and asymmetry than the stratospheric polar vortex. A comprehensive summary of the distinctions between the stratospheric polar vortex and tropospheric polar vortex is presented in Waugh et al. (2017), including the role each may play in extreme weather. As discussed in Waugh et al. (2017), the amplitude of waviness and “distortions” in the PJS correspond to anomalous temperatures and high-impact events on synoptic time scales.
Past CAO studies have targeted this waviness in the PJS and the attendant surface pressure features including a focus on the surface anticyclone–cyclone couplet (e.g., Boyle and Bosart 1983) that classically advects cold air southward in these events. Konrad and Colucci (1989) provided the first systematic study of CAOs in North America determining that the most intense CAOs were associated with later timing of downstream cyclogenesis. Subsequently, Konrad (1996) stated that CAO intensity was more strongly connected to the amplitude of the upstream anticyclone than the downstream cyclone. Relationships between surface cyclones and anticyclogenesis, and cold surface temperatures transported equatorward from northwestern North America, have also been demonstrated by Colucci and Davenport (1987), Curry (1987), Colle and Mass (1995), and Walsh et al. (2001).
On the planetary scale, Konrad (1996) found a significant signal in the pattern up to 2 weeks prior to a CAO, including persistent positive 500-hPa height anomalies over Alaska and negative anomalies over the Great Lakes. This upper air configuration is consistent with the large positive sea level pressure anomalies over Alaska into northwest Canada found by Walsh et al. (2001). Konrad (1998) expanded on the investigation of antecedent large-scale circulations with similar conclusions, and referenced the Pacific–North American (PNA; Wallace and Gutzler 1981) pattern as one of possible correlation to CAO intensity and frequency. Meanwhile, Cellitti et al. (2006) found no preference in the PNA prior to or during CAOs, but did find a preference for a negative North Atlantic Oscillation (NAO). All of these teleconnection results have the one consistent message that CAOs are connected to the planetary-scale “waviness” and displacements in the PJS.
This study aims to demonstrate that a displacement of the PJS cannot alone achieve the extreme anomalies that were recorded in January 2019. In an investigation of tropopause folding in the Arctic, Shapiro et al. (1987) identified structures similar to folds associated with the PJS. The Arctic folds were associated with jet cores positioned lower in altitude than the PJS, as well as a thermal boundary extending to the surface. On the cold side of the boundary, Arctic air resided beneath a depressed tropopause. These observations form Shapiro’s “threefold” structure of the tropopause, and the first definition of the Arctic jet stream (AJS). The AJS encircled what Shapiro referred to as a “polar vortex,” which at this point may trigger a visceral reaction from the reader and confusion with the public misuse of the term. Instead of a planetary-scale feature, the maps and cross sections in Shapiro et al. (1987) reveal a structure similar to the subsynoptic-scale disturbance that has since been referred to as a tropopause polar vortex (TPV; Cavallo and Hakim 2009, 2010a).
TPVs are one type of feature of initially called mesoscale tropopause depressions and defined as a subsynoptic maximum in the potential vorticity (PV) field with a significant lowering of the tropopause (Browning et al. 2000). Subsequently these features became known as coherent tropopause disturbances (CTD; Pyle et al. 2004), identifiable by a local extremum in the height of the tropopause, with the additional distinguishing factor for a TPV being the residence time in the polar latitudes. TPVs were found by Hakim (2000) to be the physical feature characterized by the average three-dimensional structure associated with vorticity maxima on the 500-hPa surface, in which the coinciding flow at the tropopause resembles a vortex rather than a wave. In accordance with the PV framework established by Hoskins et al. (1985), TPVs are characterized by closed material contours such as isentropes on the dynamic tropopause, defined as a surface of constant PV. As such, fluid is trapped within the vortex, isolated from the surroundings, and conserved following adiabatic and frictionless flow.
Given this conservation property, TPVs can remain coherent vortices in high latitudes, reinforced by diabatic longwave radiative cooling at the tops of low-level clouds common in the Arctic (Curry et al. 1996; Cavallo and Hakim 2012, 2013). With local tropopause-level PV tendencies generally dominated by quasi-horizontal advection, TPV lifespans can be on the order of weeks to months in some cases (Hakim and Canavan 2005; Cavallo and Hakim 2009, 2012), making TPVs the longest-lived subsynoptic-scale features in the atmosphere. The lifespan, conservation properties, and closed vortex structure may have important implications in predictability as well (e.g., Provenzale 1999). Hakim and Canavan (2005) also found that TPVs preferentially moved equatorward over the course of their lives. The ejection of TPVs from the Arctic into the midlatitudes has been shown to play a key role in cases of significant cyclogenesis (e.g., Bosart et al. 1996). Browning et al. (2000) also noted that errors in the positioning in the location of these tropopause disturbances in the initial conditions of a simulation likely resulted in errors in the prediction of the extratropical transition of a tropical cyclone.
In a composite of TPVs, Cavallo and Hakim (2010a) show that the average structure includes anomalously cold air below a lowered tropopause in the center of the vortex. Given the conservation properties of the vortex, and favored equatorward paths out of the Arctic, it follows that TPVs could contribute to CAOs. Biernat et al. (2021) used a tracking algorithm by Szapiro and Cavallo (2018) to develop a climatology of TPVs that ejected out of the Arctic, and matched it against a climatology of synoptic-scale cold pools. Biernat et al. (2021) found that cold pools associated with TPVs accounted for 32.1% to 35.7% of the CAOs over northern regions of the United States, but only in 4.4% to 12.5% of the events within the southern regions. The study by Papritz et al. (2019) of CAOs in the Fram Strait between Svalbard and Greenland also found a linkage between CAOs and TPVs with 40% of the 40 most intense CAOs and 29% of the top 100 CAOs had a TPV in the vicinity of the Fram Strait.
These recent studies suggest that TPVs are likely one mechanism causing CAOs. Our investigation of the potential role of TPVs in CAOs used observational and modeling approaches to further our understanding of the dynamic causes of extreme CAOs. The motivation for this study includes the significant societal impacts of extreme CAOs and also the relevance of this dynamic understanding to efforts aimed at advancing our knowledge of how the frequency and intensity of CAOs will vary with the warming occurring within the changing Arctic (e.g., Screen et al. 2015). This study is also part of a larger research theme on the role of TPVs in midlatitude weather extremes. While high-amplitude waviness in the PJS is an important condition in the development of high-impact weather events, it is the introduction of the Arctic jet circulating the TPV and Arctic tropopause fold that increases the potential for achieving extremes. As such, this study seeks to establish the role that a TPV played in the January 2019 extreme CAO and then use this knowledge to undertake a climatological investigation of the linkages between CAOs and TPVs over the continental United States.
The following two sections provides a summary of the PV diagnostic framework used in this study followed by a discussion of the meteorological observations and analysis for the CAO. Section 4 details the methods by which the sensitivity of the CAO to a TPV is investigated utilizing perturbed numerical simulations. Sections 5 and 6 present the results and conclusions from this experiment.
2. Potential vorticity diagnostics
Various formulations of PV have been utilized to understand the structure and dynamics of upper-level cyclones beginning with Hoskins et al. (1985) and Thorpe (1985, 1986). As illustrated by Thorpe (1986), PV inversion reveals that a cold-core upper-cyclone disturbance centered on the tropopause would be associated with a descent of the tropopause, a positive potential temperature perturbation in the stratosphere and a negative potential temperature perturbation within the troposphere. Hoskins et al. (1985) showed that with this upper-level cyclone, the static stability within the troposphere would be decreased under the depressed tropopause as magnitude of the cold thermal perturbation decreased near the surface. Thus, in addition to the previously mentioned role that upper-level cyclones, such as TPVs, play in cyclogenesis, these disturbances will also create an environment more favorable for deep convection (e.g., Nielsen-Gammon and Gold 2008). In this formulation, with the cold thermal perturbation decreasing within the troposphere as one moves toward the surface, an intense cyclonic disturbance on the tropopause would be needed to cause an extreme CAO due to strong cooling descending relatively close to the surface. Thorpe (1986) also showed that a “classic cold core disturbance” where the sign of the relative vorticity changes with height requires a thermal perturbation on both the tropopause and at the surface. For example, a cold anticyclone at the surface would be paired with a cyclonic circulation associated with a descending tropopause.
3. Methods
This study utilizes the atmospheric component of the Model for Prediction Across Scales (MPAS; Skamarock et al. 2012). The model’s dynamical core is built on an unstructured Voronoi mesh with an Arakawa C grid discretization. Voronoi partitioning of cells allows for smooth transitions to regional refinement at higher resolution, without needing nested grids. Our approach employs a variable mesh with an elliptic refinement region centered on 60°N, 100°W and the major axis along the 100°W meridian. The result is a horizontal resolution of ~15 km in the Arctic and North America, that transitions to a quasi-uniform 60-km resolution around the rest of the globe (a total of 535 554 cells; Fig. 2). We utilized a hybrid terrain-following vertical coordinate system (Klemp 2011) with 55 levels up to a top of 30 km.
(a) The approximate horizontal resolution (km) of the mesh used in the MPAS simulations and (b) schematic illustrating the method of perturbing the TPV through artificial heating tendencies during the first 48 h of the simulation.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
MPAS employs a physics suite subset of the Advanced Research Weather Research and Forecasting (WRF) Model. This suite includes the Monin–Obukhov surface layer scheme (Grell et al. 1994), the Yonsei University (YSU) planetary boundary layer (PBL) scheme (Hong and Pan 1996), YSU gravity wave drag, Thompson double-moment microphysics (Thompson et al. 2004), RRTMG shortwave and longwave radiation (Iacono et al. 2008) called every 30 min, and the new Tiedke convection scheme (Tiedtke 1989) modified for WRF (Zhang et al. 2011). All simulations were integrated forward 7 days with a time step of 90 s.
Four simulations were initialized at 0000 UTC 25 January 2019 with the final operational analysis from the National Centers for Environmental Prediction’s Final Global Data Assimilation System (NCEP’s FNL-GDAS) (Myrick 2017). One simulation was considered a control run. The strength of the TPV varied over the first 48 h in the other three simulations through artificial heating tendencies imposed above and below the tropopause within the TPV “basin.” Based upon inspection of the control run, the TPV basin in the modified runs was defined as all adjacent cells with a potential temperature less than 285 K on the 2-PVU surface. Following Eq. (3), the TPV was weakened (strengthened) through cooling (heating) the stratosphere and heating (cooling) the troposphere.
Specifically, two simulations were designed to weaken the TPV by imposing a heating rate of 10 K day−1 (hereafter called TPVweak10) and 5 K day−1 (hereafter TPVweak5) in the troposphere, and a cooling rate of the same magnitude in the stratosphere. The other modified simulation strengthened the TPV (hereafter TPVstrong5) with a cooling rate of 5 K day−1 imposed in the troposphere, and a heating rate of the same magnitude imposed in the stratosphere. After 48 h of applying these heating and cooling rates, significant differences in TPV intensities were noted. Thus, at hour 48, the modified simulations are restarted with artificial tendencies removed, and run forward with the same model configuration as in the control simulation. These simulations were conducted for another 5 days, approximately the time for the TPVs to move through the area of interest. The effectiveness of the tendency modifications was assessed by examining the subsequent intensity of the TPV at hour 48 defined by the minimum potential temperature on the 2-PVU surface in the TPV basin. The center of the TPV at a given time was defined by the location of that minimum potential temperature. TPV tracks were generated simply by using a regional minimum filter applied to the 2-PVU potential temperature field, evaluated at the location from the previous time step, and then identifying the new location of the regional minimum.
Coarser simulations with quasi-uniform resolutions of 60, 120, and 240 km were also utilized to test the modified tendencies and evaluate the sensitivity of the TPV and CAO to the gird resolution. The primary differences with the coarser-resolution simulations are a systematic bias toward a weaker TPV, as well as a reduction in the equatorward reach of the TPV track. While we will not explore these differences in detail in this paper, one is cautioned about utilizing such coarse-grid models to examine the intensity and equatorward extent of CAOs that are associated with TPVs moving from the Arctic into the middle latitudes. Our findings and the subsequent results of Biernat et al. (2021) and Papritz et al. (2019) that show TPVs are often related to CAOs implies a more general caution should be considered when drawing inferences about CAOs from coarse-grid simulations.
In addition to these simulations, a climatology of CAOs were determined from the North American Regional Reanalysis (NARR; Mesinger et al. 2006) with a 32-km grid spacing for 1979 through January 2019. Daily mean 2-m temperatures (T2m) were standardized against the climatological seasonal mean and standard deviation (σ) for December through February (DJF). Only grid points south of 50°N and with a DJF T2m σ greater than 5°C were considered viable to be a part of a CAO. At these viable grid points, streaks of at least two consecutive days with T2m anomalies less than −2σ were identified, consistent with Walsh et al. (2001). Finally, CAO days were defined as any day with at least 75% of a 500 km × 500 km region composed of these cold streak grid points. This method thus includes both temporal and spatial criteria with subsynoptic-scale thresholds. Additionally, by using standardized anomalies, CAOs are dependent on the second- and higher-order moments. For example, high CAO frequency is associated with large negative skewness in DJF temperatures. The centroid of the CAO was defined as the center point of the region with the most cold-streak grid points, and the magnitude was determined by the median T2m standardized anomaly within the given region.
4. Results
a. Observations
At 1200 UTC 30 January 2019, an extratropical cyclone was located to the northeast of the Great Lakes (Fig. 3). An anticyclone and trailing cold front located over the central United States was also present with the front extending relatively far southward reaching the border of Oklahoma and Texas, southern Arkansas, and northern Mississippi. Comparison of the extreme cold in Fig. 1 and the position of the cold front (Fig. 3) suggests that the low temperatures (i.e., coldest per year) and the extremes (i.e., coldest in 10 years) are limited to the northern portions of the cold air mass with the southern portions behind the front relatively unscathed.
The NOAA WPC operational surface analysis for 1200 UTC 30 Jan 2019.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
The FNL-GDAS analysis at 1200 UTC 30 January is shown in Fig. 4. The 500-hPa winds and height field (Fig. 4a) reveals a large ridge over the eastern Pacific and western North America with a trough containing both an extended area of enhanced vertical vorticity and a strong cyclonic circulation over the Great Lakes that is consistent with a TPV. A vertical cross section through the region of strongest vorticity (Fig. 4b) reveals a structure characteristic of an tropopause fold with a dramatic descent of the 2-PVU surface from above 300 hPa to near the surface with a well-defined cyclonic circulation surrounding the tropopause fold. The dramatic descent of the 2-PVU contour to the surface is also visualized in Figs. 4c and 4d with a clear vortex signal characteristic of an intense TPV. The basic interpretation of Figs. 4b–4d is that the stratospheric air within the TPV has descended to below 900 hPa, just above Earth’s surface. From Figs. 4b and 4d, it is evident that the descending core of the TPV slopes to the southwest, to the right of the TPV core aloft. Earlier work has shown that vertical motion near the tropopause result in vertically tilting the tropopause and once this initial vertical tilt is established, the vertical shear associated with the cold air advection in the northwesterly flow acts to further tilt the tropopause in the vertical (Wandishin et al. 2000).
FNL-GDAS analysis for 120 UTC 30 Jan 2019. The (a) 500-hPa height (dm) contoured every 6 dm, wind barbs (full barb = 10 kt, flag = 50 kt; 1 kt ≈ 0.51 m s−1), and relative vertical vorticity (s−1) in color fill, (b) cross section along the green line plotted in (a) from southwest to northeast centered on Detroit. Potential temperature is contoured every 5 K, the 2-PVU EPV contour is drawn in green, wind barbs are plotted with the same convention as (a), and the wind component normal to the cross section is given by the color fill (positive = out of the page). (c),(d) The 2-PVU isosurface shaded by potential temperature (K) in a three-dimensional plot centered on the 500-hPa vorticity maximum. The axes are x: longitude; y: latitude; and z: height (m). The perspectives are (c) looking directly down and (d) looking up to the east.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
Backward trajectories were initiated at 1200 UTC 30 January 2019 utilizing NOAA ARL’s HYSPLIT program (Stein et al. 2015) based upon the FNL-GDAS analysis (Fig. 5). The initial points for these trajectories were taken from a 1° × 1° latitude–longitude grid within the tropopause fold from heights of 3500 m above sea level. These trajectories (Fig. 5) clearly show that the colder air (i.e., less than 270 K) within the tropopause fold can be traced to north of Canada in an area just west of Ellesmere Island and Greenland. Trajectories for the warmer air mass also suggests Arctic origins. The looping nature of the trajectory is consistent with the appearance of a prolate cycloid as air well away from the centroid circulates around the cyclone as the TPV moves equatorward. The geographical spread in the trajectories appears quite narrow suggesting that the air parcels remain within the cyclonic circulation as the TPV moves across Canada and into the areas west and south of the Great Lakes. This capturing of the TPV circulation is consistent with our earlier discussion of the PV framework as the closed material surfaces associated with a TPV will result in the air being trapped within the TPV circulation. Comparison of the minimum temperatures (Fig. 3) and the trajectories (Fig. 5) show that the coldest temperatures are located in the wake of the TPV and evolving tropopause fold. The trajectories (Fig. 5) also suggest a rapid movement of the cold lower tropospheric air mass into the middle latitudes as, for example, the trajectories to the west of the vortex center move from the northern Canadian Arctic coast to the U.S. border in the less than 24 h.
Backward trajectories initiated at 1200 UTC 30 Jan 2019 from a 1° × 1° latitude–longitude grid 3500 m above sea level within the tropopause fold shown in Fig. 4. Trajectories are run backward for 132 h using NOAA ARL’s HYSPLIT program (Stein et al. 2015) and FNL-GDAS analysis. Each point is at 1-h intervals along the trajectories, and their colors refer to the potential temperature of the parcel.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
Radiosonde measurements support the concept that stratospheric air has descended substantially as shown from soundings taken at Chanhassen, Minnesota, and from Green Bay, Wisconsin, at 0000 UTC 30 January 2019 (Figs. 6a,b, respectively). These profiles taken near and to the west of the vortex center in Fig. 4 reveal stable, dry air extending to 600 hPa. In particular, the stability profile within the Green Bay sounding shows the signature of a tropopause fold near 800 hPa. However, the air near 800 hPa in the sounding has high relative humidity suggesting air that does not originate within the stratosphere. Backward trajectories (not shown) reveal that this moist air mass originates at lower levels within the Arctic and is lifted into the fold. This lifting of lower-tropospheric flow is most likely due to the connection between the surface and upper-level fronts associated with the extreme lowering of the tropopause. Both of these soundings taken in the evening hours also have a mixed boundary layer extending to near 850 hPa.
Operational National Weather Service radiosonde measurements from (a) Chanhassen (Twin Cities) at 0000 UTC 30 Jan 2019, (b) Green Bay at 0000 UTC 30 Jan 2019, (c) Lincoln at 1200 UTC 30 Jan 2019, and (d) Davenport at 0000 UTC 31 Jan 2019. These soundings were obtained from the University of Wyoming’s Department of Atmospheric Sciences’ sounding archive.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
The soundings taken later and to the west from Lincoln, Nebraska, at 1200 UTC 30 January 2019 and Davenport, Iowa, at 0000 UTC 31 January 2019 show the tropopause (defined as the base of the stable layer associated with the stratospheric air mass) extending nearly to below 900 hPa (Figs. 6c,d, respectively). The lowering of the tropopause nearly to the surface is consistent with vertical cross section and the 2-PVU surface from the FNL-GDAS analysis shown earlier in Figs. 4b and 4d. The winds in these soundings also reveal a general increase with height with little change in direction in this layer. This lack of any significant directional vertical shear implies that little geostrophic cold air advection is occurring near Earth’s surface at this time.
The structure of the CAO and TPV is quite different for measurements taken near and “downstream” of the Great Lakes. For example, the soundings at Detroit, Michigan, and Buffalo, New York (Figs. 7c,d), have a much deeper layer of reduced stability relative to the soundings taken to the west at Chanhassen and Green Bay (Figs. 7a,b). This layer of reduced stability extends to 700 hPa at Detroit and 660 hPa at Buffalo. At both of these sites, significant moistening and warming (i.e., 10°–20°C) has taken place within this layer. While these changes could be due to differential advection, the impact of the open waters of the Great Lakes is explored in the subsequent simulations. This modification is also consistent with the lake-effect snows discussed in the introduction.
Operational National Weather Service radiosonde measurements taken from (a) White Lake (Detroit), Michigan, at 1200 UTC 30 Jan 2019 and (b) Buffalo at 1200 UTC 30 Jan 2019. These soundings were obtained from the University of Wyoming’s Department of Atmospheric Sciences’ sounding archive.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
b. Control simulation
The control run is examined first to establish the evolution of the atmosphere over the week leading up to and including the CAO. In the first 48 h, through 0000 UTC 27 January, a closed area of low potential temperatures colder than 270 K is evident on the dynamic tropopause over the northern Canadian archipelago, identifiable as a TPV (Fig. 8). The location and initial meandering path of the TPV agrees with the backward trajectories in Fig. 5 supporting the observational analysis that suggests a linkage between the CAO and the TPV. Another well-defined TPV is noted to the south-southeast, entering Ontario on 26 January. This first TPV skirts just north of the Great Lakes on 27 January and is associated with the initial surge of cold air entering Wisconsin, Michigan, and subsequently into the Northeast through 28 January (Fig. 9). The lead TPV plays an important secondary role in the CAO. While the NAO during this time was positive, as for example in the persistent ridge of warm tropopause temperatures over the North Atlantic (Fig. 8), the lead TPV temporarily pushes this ridge east along with a transient surface low near 50°N, 50°W (not shown) that mimics the effect of a negative NAO, helping upstream cold air to pour southeastward. From 29 to 30 January, the lead TPV slides eastward from Quebec out to the northern Atlantic.
Potential temperature (K) on the 2-PVU surface in the control run: (top right) 0000 UTC 26 Jan 2019, 24 h into the simulation. Each subsequent panel advances 24 h.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
Control-run forecast 2-m temperature (°C). The white line denotes the track of the TPV, and the white star denotes the locations of the TPV at the respective times: (top right) 0000 UTC 26 Jan 2019, 24 h into the simulation. Each subsequent panel advances 24 h.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
A large ridge characterized by tropopause potential temperatures over 330 K was a feature over the eastern Pacific and western North America through the control simulation (Fig. 8) consistent with the 500-hPa FNL-GDAS analysis shown earlier (Fig. 4). A key shortwave trough can be seen cresting the ridge over the Yukon territory at 0000 UTC 27 January (Fig. 8). As the shortwave moved equatorward over Alberta, the destructive interference is removed from the ridge which reamplifies toward Alaska and the Yukon on 28 January. As this ridge amplifies and the shortwave trough extends southward into the midwestern United States on 29 January, the TPV moves southward out of Canada. This TPV enters into the Midwest on 29 January subsequently crossing the Great Lakes on 30 January and moving into the northeastern United States on 31 January. At the surface, an Arctic air mass characterized by temperatures below −30°C is dragged southward at the leading edge of the TPV from 27 through 29 January (Fig. 9) leaving a swath of temperatures below −30°C in its wake.
The evolution of vertical structures was analyzed by averaging variables in the horizontal within a 400-km radius of the TPV center and plotting the averages in a time–height section (Fig. 10a). The radius was selected to focus on the core of the TPV as shown in Fig. 4. Increased stability above the cold surface layer contributes to increased PV below 850 hPa through 28 January, which can be seen in the time–height section of area-averaged potential temperature and PV in Fig. 10a. Over time, the cold air deepens as seen by the evolution of the 240-, 245-, and 250-K isentropes, pushing the high-PV air upward to between 700 and 800 hPa through 29 January. At its most pronounced stage, the TPV and cold pool are collocated over central Manitoba on 29 January.
Vertical cross section of Ertel potential vorticity (EPV) and potential temperature averaged over an area defined by a 400-km radius from the center of the TPV. The (a) potential temperature and (b) potential vorticity tendencies from longwave radiation (LW), (c) shortwave radiation (SW), (d) cumulus (CU), (e) boundary layer (BL), and (f) mixing (MIX) over this area.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
Insight into the physical causes of the evolution of the TPV in the observed soundings (Figs. 6 and 7) and the PV and potential temperature fields in the control simulation (Fig. 10a) can be explored using the time rate of change of PV, Eq. (3), within the context of the diabatic tendency fields from the parameterization schemes utilized in MPAS, Eq. (4). Within the middle and upper troposphere, the contributions are generally small with positive contributions from the longwave radiation (Fig. 10b) and negative contributions from shortwave radiation (Fig. 10c). The decrease in PV is associated with increasing shortwave radiation as the TPV moves equatorward (Fig. 11a). Positive contributions to TPV intensity from diabatic longwave cooling has been previously noted (e.g., Cavallo and Hakim 2012, 2013). During the southward movement of the TPV, the area-averaged dynamic tropopause lowers over time, by about 10 K and 50 hPa, assisted by longwave radiative cooling near the tropopause.
Track of the TPV in each simulation: (a) Control, (b) TPVweak10, (c) TPVweak5, and (d) TPVstrong5. The cyan line denotes the track of the TPV in ERA5. Squares denote the 24-h TPV positions at 0000 UTC for the four simulations. The color of the track corresponds to the depth of the TPV given by the minimum potential temperature on the 2-PVU surface.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
Below 700 hPa, the impacts of the diabatic terms are much larger (Fig. 10) as the TPV moves southward across Canada and into the United States (see Fig. 8). The diabatic impact of the boundary layer parameterization (Fig. 10e) is to create a shallow layer of PV enhancement near the surface beginning on 25 and 26 January, while the TPV is still within the Arctic (Fig. 8). As the TPV moves southward, the impact of the boundary layer parameterization deepens and includes times when the PV tendency changes sign. While a near constant temperature might be expected in the Arctic, the southward moving TPV is encountering a surface affected by the diurnal cycle in shortwave radiation and the presence of open water over the Great Lakes (Fig. 8). The effects of the Great Lakes on the TPV are evident due to the destruction of PV through the cumulus convection scheme (Fig. 10d). The reduction in stability combined with the moisture source from the lakes promotes shallow cumulus convection (e.g., Agee and Gilbert 1989). The low-level lapse rates also rapidly steepen as seen by the increased spacing between isentropes below 700 hPa in Fig. 10. As discussed earlier, the soundings at Detroit and Buffalo (Figs. 7a,b) near the Great Lakes revealed a steepening of the lapse rate and moistening of the lower atmosphere. This environment contributing to the production of cumulus convection result in the long wave radiation cooling at cloud tops and warming underneath decreasing PV (e.g., Fig. 3 in Cavallo 2009) from near the surface to below 850 hPa (Fig. 10). In contrast, mixing has a generally smaller impact (Fig. 10f). The end result of the passage of the TPV over the Great Lakes is a clear destruction of PV below approximately 600 hPa (Fig. 10) with deepest impacts due to the cumulus convection. Consistent with this destruction of PV, the tropopause height lifts upward on 31 January following the passage of the TPV over the Great Lakes (Fig. 10). The interaction between the TPV and the Great Lakes are consistent with the multiday lake-effect snowfall events, with, for example, NOAA reports of snowfall totals of up to 37.6 in. (95.5 cm) south of Watertown, New York, and up to 21 in. (53.3 cm) at Buffalo.
c. Modified simulations
All four simulations have similar tracks and timing of the TPV as it ejects out of the Arctic (Fig. 11) with TPVweak10 (Fig. 11b) and TPVweak5 (Fig. 11c) especially similar to the control track (Fig. 11a). These three simulations all have a track into southern Canada and subsequently into the midwestern United States, but east of the observed event. In contrast, the TPV in TPVstrong5 begins diverging farther east of the control during the first 4 days of the simulation (Fig. 11d), but after day 4 abruptly moves to the southwest and then follows a track similar to the observed event into the CONUS. The TPV rolls through the base of the longwave trough like the other simulations, but up to 400 km farther south. The center passes over Washington, D.C., around day 6, compared to Buffalo in the control.
Despite these differences, ultimately any perturbations in the intensity, size, and structure of the TPV induced by the artificial heating tendencies did not impact the synoptic-scale pattern of the track of the TPV. All simulations, including low-resolution tests, transported the TPV southward into the trough over Canada and the northern CONUS. This result is a testament to the intrinsic predictability of the large-scale flow, in which the sensitivity to this particular mesoscale perturbation introduced in the Arctic was small. These results do, however, show that the stronger TPVs (e.g., control and TPVstrong5) take paths farther south.
To compare the evolution of the vertical structure around the TPV in the modified simulations to the control, the area-averaged time–height analysis applied to the control simulation is again utilized. Systematic differences between the three modified simulations and the control are apparent in the dynamic tropopause over the first 3 days (Fig. 12). In accordance with the methods of modifying the TPV strength, TPVweak10 and TPVweak5 feature a reduction in PV near 500 hPa with a tropopause that is shifted upward from the control (Figs. 12a,b). Likewise, TPVstrong5 features an increase in PV near 600 hPa, with a tropopause that is shifted downward from the control (Fig. 12c).
Differences in EPV over the 400-km radius area from the TV center from the control and (a) TPVweak10, (b) TPVweak5, and (c) TPVstrong5. The color fill is the EPV difference between the modified run and control run, scaled by the average EPV of the two runs. In each panel, 2-PVU EPV is contoured for the modified run (solid line) and the control run (dashed line).
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
There are notable negative PV differences from 1800 UTC 28 January to 0000 UTC 29 January in both of the TPVweak simulations (Figs. 12a,b). These differences correspond with the negative PV tendencies due to the diabatic boundary layer tendency in the control (Fig. 10b), and are likely a result of stronger radiational surface warming in the modified simulations with less cloud cover. In all three of the modified simulations, there is generally higher PV in the boundary layer during 29 January, and higher PV in the middle troposphere on 30 January (Fig. 12). Particularly on 30 January, this coincides with the passage of the TPV over the Great Lakes. The differences in TPVweak10 and TPVweak5 may be attributed to the weaker TPV introducing a smaller temperature difference between the water and air, and thus developing less cumulus convection than in the control. The difference in TPVstrong5 is expected both due to the deviation in the TPV track to the south of the Great Lakes (Fig. 11d) and to antecedent positive PV perturbations from the prior intensification. The linkages between TPV strength and the severity of lake-effect snows is an area for future research.
d. Sensitivity of extreme cold to the TPV
The depth of the TPV, given by the minimum potential temperature in the TPV basin, diverges quickly among the four simulations in the first 24 h (Fig. 13a). The control and TPVstrong5 simulations deepen the TPV by 20–30 K, respectively, while TPVweak5 and TPVweak10 weaken the TPV by 10–20 K, respectively as expected given the experimental design. The intensities generally remain steady over the next 24 h. On 27 January, the control and TPVstrong5 begin to gradually weaken. By 29 January, the control TPV has weakened back to the intensity at the beginning of the simulation, while both TPVweak5 and TPVweak10 show little subsequent change in intensity. By 30 January, the control begins converging toward TPVweak5, and TPVstrong5 undergoes rapid weakening to join this cluster as well. On 31 January, the control, TPVweak5, and TPVstrong5 are tightly clustered while TPVweak10 is approximately 10 K weaker than the rest. This evolution likely shows the impact of the TPV moving into this specific middle latitude flow and the relatively large diabatic influences on the system that occur with cyclogenesis that takes place on 30 and 31 January.
Time plots from the 7-day forecast runs for the control (green), TPVweak10 (red), TPVweak5 (orange), and TPVstrong5 (blue). (a) Minimum potential temperature on the 2-PVU surface in the TPV and (b) SLP maximum (solid line) and minimum (dashed line) within a 1600-km radius of the TPV center.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
Following the past studies discussed in the introduction that attribute CAO magnitude to the strength of the surface anticyclone, there is a desire to test their null hypothesis in the case of the January 2019 CAO. The strength of surface pressure centers were assessed within a 1600-km radius of the TPV center; a horizontal length scale that exceeds one synoptic wavelength while much less than two wavelengths to avoid capturing unrelated features. The surface high within this radius of the TPV begins at 1023 hPa over the Arctic, and gradually strengthens over the next 24 h (Fig. 13b). From 26 to 27 January, all simulations bring the high pressure up to around 1040 hPa. TPVstrong5 intensifies the surface high by 12 hPa over a 24-h period up to 1044 hPa at 0000 UTC 28 January. The other simulations also exhibit strengthening rates of 9–11 hPa in 24 h, all exceeding the threshold for rapid anticyclogenesis used by Colucci and Davenport (1987).
While the spread is still small, by this point late on 27–28 January, a systematic separation of the simulations develops (i.e., a correspondence between the modification of the TPV and the strength of the high). By 0000 UTC 30 January, when the surface anticyclone is crossing into the northern plains from Canada, the maximum SLP in TPVstrong5 is 1039 hPa, and in TPVweak5 is 1034 hPa (Fig. 13b). The greatest separation in the simulations occurs on 30 January in which an inverse relationship exists between the minimum 2-PVU potential temperature in the TPV and the maximum SLP of the surface high. At 1200 UTC 30 January, the surface high pressure area was observed to stretch from Manitoba into the Midwest, with a maximum SLP of 1033 hPa. The simulations range from 1031 hPa in TPVweak10 to 1041 hPa in TPVstrong5. Thus, the strength of the anticyclones appear to be dependent on the intensity of the TPV.
The differences in surface temperatures between the modified simulations and the control are consistent with the surface higher pressure and colder air at the surface being associated with the stronger TPV. The surface temperatures were evaluated using the minimum of a 24-h running-mean convolution of 2-m temperatures over the 7-day simulation period (Fig. 14). The timing of the coldest 24-h period by location (not shown) is consistent among the four simulations, particularly within the swath of Arctic air carved by the TPV. There is remarkable linearity in the relationship between the modification of the initial strength of the TPV over the Arctic and the resulting 2-m temperatures in the eastern half of the CONUS (Fig. 14). Note that 2-m temperatures over northern Canada and the Arctic were generally not affected by the modifications, with systematic differences along the track of the TPV only beginning to appear in southern Canada.
Difference of each modified simulation from the control of the coldest 24-h running-mean 2-m temperature during the 7-day forecast period, for the (a) TPVweak10 run, (b) TPVweak5 run, and (c) TPVstrong5 run. Solid green line marks the track of the TPV in the modified simulations, and the dashed green line marks the track in the control.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
The magnitude of the temperature differences over the CONUS are substantial. Between TPVweak10 (Fig. 14a) and TPVstrong5 (Fig. 14c), the mid–Mississippi River valley in particular ranges more than 11°C (20°F). Rather than being maximized directly along the track of the TPV, the location of these temperature differences are displaced to the right of the track, and mostly to the south of the record cold shown in Fig. 1. In TPVstrong5, this placement could to an extent be explained by the southward deviation in the track itself (Fig. 11); however, the TPVweak5 and TPVweak10 runs have the same southward-shifted temperature differences but with TPV tracks similar to the control. These results suggest an asymmetric spatial relationship between the TPV intensity and 2-m temperatures and a strong impact of TPV intensity on surface temperature associated with the extreme CAO. These results together with the studies by Biernat et al. (2021) and Papritz et al. (2019) show the importance of considering the intensity and location of TPVs in CAOs. The tilt of the TPV with height, as shown earlier, also impacts this relationship between the location of the TPV center and the location of the coldest air.
The strength of surface low pressure was also evaluated within the same radius as the high (Fig. 13b). The minimum SLP was remarkably consistent among the TPVweak10, TPVweak5, and control simulations. Most notably, these three simulations all intensify a low over Quebec from around 998 to 974 hPa in the 24 h between 0000 UTC 30 January and 0000 UTC 31 January, qualifying it as a “bomb” cyclone (Sanders and Gyakum 1980). Only TPVstrong5 deviated from this tight clustering, with a substantially weaker low on 30 January that is much closer to observations. These differences raise the issue of whether there is a strong dependence of the intensity of cyclogenesis on the magnitude and location of TPVs. This issue is an area for future investigations and suggests the possible importance of an accurate representation of TPVs in midlatitude cyclogenesis.
5. Climatological relationship between TPVs and CAOs
Applying the approach outlined in the section 3 to the period from 1979 through 2019, 256 CAO days were identified, with the majority found along the full meridional extent of the plains (Fig. 15). Lower spatial density is found east of the Mississippi River. A second cluster is found in the northeastern United States and southern Quebec (limited to the north by the 50°N boundary criterion). The combination of smaller temperature variance and smaller skewness are likely due to modification of cold air masses by the large bodies of water resulting in a dearth of CAO events centered within the Great Lakes region. The position of the CAO for the 30 January event studied in this case (Fig. 15) suggests that many CAOs extend far farther south than this extreme event sometimes with a similar magnitude. The relationship between CAOs at the surface and dynamics aloft was explored by compositing the potential temperature, winds, and tropopause potential temperature anomaly on a 2-PVU surface (Fig. 16). In this composite, the CAO lies poleward of a jet streak on the 2-PVU surface (Fig. 16) and the CAO centroid lies to the southwest of a strong cold anomaly on the tropopause. This finding is consistent with our simulations and observations that revealed, on average, the coldest air is just to the southwest of the TPV.
Points of CAO centroids and corresponding median standardized temperature anomaly given by the color for period from 1979 through 2019.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
The potential temperature anomaly on the tropopause (shaded), potential temperature (black lines), and 250-hPa wind (green lines) on the 2-PVU level plotted relative to the CAO centroid. The data were obtained from ERA5
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
To explore if the relationship between CAOs and cold anomalies on the tropopause are linked to TPVs, the locations and tracks of all TPVs during DJF are obtained using the tracking algorithm from Szapiro and Cavallo (2018) applied to ERA-Interim (Dee et al. 2011) reanalysis for the period from 1979 through 2018. Two definitions were analyzed for tracking these vortices: one requiring genesis north of 60°N (60Ngen), and the other requiring 60% of the feature’s lifetime be spent north of 65°N (60N65). It is important for the reader to note that the later criteria is consistent with the widely utilized TPV definition put forth by Cavallo and Hakim (2005), while our first TPV definition is not. A relaxed criteria was also implemented by Biernat et al. (2021). The relationship between these two categories of Arctic vortices and CAOs utilizing the definition discussed is shown in Fig. 17. Specifically, the distance between the CAO centroid and a TPV was calculated for all CAOs and then the percentage of CAOs with a nearby TPV was determined as a function of distance between the two features (Fig. 17). For TPVs generated north of 60°N latitude, 85% of the CAOs had a vortex located with 1000 km with over 95% of the CAOs having a TPV within 2000 km.
The percentage of time there is a TPV within a specified distance threshold (x axis) given a cold air outbreak (CAO) location at the time of the CAO. The blue solid line (labeled 60Ngen) represents all tropopause PV anomalies originating poleward of 60°N, while the red solid line (labeled 60N65) utilizes the more common TPV definition that the TPV spends 60% of its lifetimes poleward of 65°N. The blue dashed line and shading show the average percentage of time that there is a 60Ngen TPV within a specified distance threshold (x axis) for a random December–February day between 1979 and 2019. The random sample was created by using all CAO locations but with random dates and repeating 1000 times. The shading is the 95% range of this sample. The red dashed line and shading represent the same approach but for the 60N65 TPVs.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
Since the close proximity of these features could be due to random chance rather than a dynamical relationship, we also took the location of the CAO and then calculated the distance to a TPV on days selected at random (1000 times) from the December through February 1979–2019 period. The results of this random selection process for the mean and 95% range for this sample contrast sharply with the CAO–TPV proximity as TPVs on random days are only near that location a few percent of the time (Fig. 17). This result supports the findings of our case study and also the recent work of Biernat et al. (2021) and Papritz et al. (2019) that have shown connections between TPVs and CAOs. While the results of Biernat et al. (2021) argue that a 400-km threshold should be utilized when searching for a linkage between TPVs, cold pools associated with TPVs and CAOs, Papritz et al. (2019) noted that the cyclonic circulation of TPVs can aid in the creation of CAOs through modifying the longer range transport of Arctic air.
The relationship is less clear for those TPVs that have spent over 60% of their lifetime poleward of 65°N (Fig. 17), as for example, the average distance is closer than 1000 km in these cases in less than 25% of the time. This weaker relationship for TPVs that have spent greater time in the Arctic related to those TPVs simply generated north of 60°N could imply differences in the large-scale flow such as a more positive annular mode that traps the 65°N TPVs in the Arctic with fewer escaping to the midlatitudes. The tracks for TPVs passing within 1000 km of a CAO are plotted in standardized form against the full winter climatology (Fig. 18). Over North America, there is a clear tendency for TPVs to move from the northern Canadian Arctic southward into the Great Lakes/Upper Midwest and then northeastward over Labrador. This result is quite similar to the event investigated in this study. Given our previous finding that TPVs that spent over 60% of the time poleward of 65°N and this TPV track density, TPVs associated with CAOs are likely to form in the Arctic and then be ejected relatively rapidly into lower latitudes. Thus, relaxing the requirement that the vortices must remain the Arctic for a majority of their lifetime to be called a TPV provides a method to diagnose and include those vortices that originate in the Arctic, are advected more rapidly equatorward and are more likely to include an extreme CAO in midlatitudes. Figure 18 also shows that the TPVs associated with CAOs over the CONUS are less likely to spend any extracurricular time between Greenland and Siberia.
Standardized track density for TPVs passing within 1000 km of CAO centroids. The track densities are plotted relative to the full-winter TPV climatology as determined from ERA-Interim.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
A comparison of the minimum anomaly, minimum latitude and lifetime for TPVs located 1000 km or less of the centroid of a CAO and all TPVs shows that CAO TPVs are substantially more intense, are located significantly equatorward, and have significantly longer lifetimes than typical TPVs (Fig. 19). Statistical significance is established using the two-sided Kolmogorov–Smirnov (K-S) test to determine whether the distribution of TPVs associated with CAOs are statistically different from the full sample of TPVs (Massey 1951). In terms of tropopause potential temperature anomaly in the vortex core, anomalies for TPVs located 1000 km or less of CAOs exhibit a peak in the distribution around −40 K with anomalies as low as −76 K with respect to the long-term climatology (Fig. 19a). TPVs associated with CAOs tend to have a strong maximum in equatorward movement with a well-defined peak in their minimum latitude near 45°N latitude (Fig. 19b). Furthermore, lifetimes are significantly longer for TPVs associated with CAOs, with higher probabilities of TPVs with lifetimes of over 1 week with probabilities of some exceeding one month (Fig. 19c). These results again suggest a linkage between CAOs and strong, long-lived TPVs that move equatorward from the Arctic into the midlatitudes.
Cyclonic tropopause polar vortex structural properties of (a) minimum core anomaly, (b) minimum latitude, and (c) lifetime. The bin interval is 1 K in (a), 1° latitude in (b), and 1 day in (c). The black contours correspond to the full climatological record, while the gray contours correspond to the vortices that are located 1000 km or less from a cold air outbreak. The P values from the K-S test are 3.62 × 10−101 in (a), 8.30 × 10−128 in (b), and 3.02 × 10−56 in (c). Anomalies in (a) are computed with respect to the long-term daily climatological mean (1979–2010) of tropopause potential temperature at the corresponding grid point of the vortex core (vortex core minus climatological value).
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
6. Discussion and conclusions
This study investigated an extreme CAO that began in late January 2019 and was responsible for breaking numerous climatological records for the lowest minimum and maximum temperatures across the U.S. Great Lakes, Midwest, and Northeast. The wind chills and extreme cold were associated with significant detrimental societal impacts including 21 fatalities, closures of schools and businesses, and transportation disruptions. Our observational and modeling investigations revealed that this extreme CAO was associated with an intense TPV.
A schematic of the TPV and CAO relationship is shown in Fig. 20. Within the Arctic, the TPV is associated with a lowering of the tropopause and a cyclonic circulation (Fig. 20a). Previous studies focused on the vertical structure of TPVs have revealed that a dome of anomalous cold temperatures occurs within the troposphere below the lowering of the tropopause (Cavallo 2009; Cavallo and Hakim 2010b; Papritz et al. 2019). The vertical gradient in longwave radiative heating often associated with TPVs in the Arctic will cause the TPV to strengthen with a further lowering of the tropopause and an enhancement of this thermal structure. In the event studied, the tropopause descended nearly to Earth’s surface.
Schematic of the interaction between the structure of the TPV and the CAO. (a) The structure of the TPV within the Arctic. There is a lowering of the tropopause with a cyclonic circulation around the lowered as indicated by the conventional symbols for flow into and out of the page. The vertical shear in the background westerly (U) winds are relatively weak. Longwave cooling increases the intensity of the TPV with a lowering of the tropopause and (b) the structure of an intense TPV moving into middle latitudes. As the TPV begins to interact with the polar jet, the vertical shear (U) increases and the system begins to tilt with height. The cyclonic flow around the center can also increase. PV is destroyed, and warming and moistening of the lower troposphere occurs as the TPV moves over warm and moist surfaces. The example is consistent with the observed lake-effect snows.
Citation: Journal of the Atmospheric Sciences 78, 9; 10.1175/JAS-D-20-0285.1
Our findings suggest that the dome of the anomalous cold temperatures remained associated with the TPV as these feature were ejected out of the Arctic. This behavior is expected since, as described in the introduction, the potential vorticity structure of the TPV means that the air remains within the system for adiabatic and frictionless flow. The evidence for the association between this intense TPV and the cold dome as the TPV entered the continental United States includes (i) the wind and thermal structure in the vertical cross section and the depiction of the 2-PVU surfaces (Fig. 4), (ii) the close proximity of the ensemble of lower tropospheric backward trajectories (Fig. 5), (iii) the relatively slow initial modification of the TPV structure within the simulation (Fig. 10), and (iv) soundings revealing extremely cold surface temperature near the lowest descent of the tropopause (Fig. 6).
A schematic of the subsequent evolution of the intense TPV moving into middle latitudes is shown in Fig. 20b. Within middle latitudes cold air was found at the surface and within the lower troposphere beneath the tropopause depression associated with the intense TPV. A tilt of the TPV with height is also evident with the extreme temperatures found in the quadrant where the Arctic tropopause descended nearly to the surface (Fig. 20b). The vertical tilt of the TPV is likely connected to the vertical shear associated with the cold air advection from the northwest flow (Wandishin et al. 2000). Recent studies investigating jet superposition events found that vertical circulations associated with the polar jet had a strong ability to vertically restructure the tropopause (Winters and Martin 2017; Winters et al. 2020). The combination of the vertical circulation driven by the polar jet and the vertical shear could be driving the tropopause vertical extent and tilting. A subsequent impact of middle latitude processes is that as the TPV moved through the region PV was destroyed due to longwave radiation, boundary layer processes and cumulus convection (Fig. 10). This modification was especially large when the TPV moved over the ice-free Great Lakes creating lake-effect snows.
The linkage between CAOs and TPVs was also found in our climatological investigation that showed TPVs often (85% of the time) exist within 1000 km of a CAO and that these TPVs tend to be higher amplitude, longer lived, and move farther equatorward than the climatological distribution of all TPVs. Our finding is consistent with recent studies of CAOs by Biernat et al. (2021) and by Papritz et al. (2019). For example, Papritz et al. (2019) found that 40% ± 5% of the 40 most intense CAO events over the Fram Strait were directly associated with a TPV. Biernat et al. (2021) concluded that cold pools associated with TPVs accounted for a significant minority (32.1%–35.7%) of the CAOs over northern regions of the United States, but far less (4.4%–12.5%) of the events within the southern regions. Biernat et al. (2021) used a threshold distance of 400 km to define a CAO that was directly linked to the cold pool of a TPV, while Papritz et al. (2019) noted that TPVs may also indirectly contribute to a CAO through enhancing long-range transport of Arctic air. Our studies show that 85% of CAOs had a TPV within 1000 km allowing the possibility of direct and/or indirect impacts of TPVs to be present. These results show that intense TPVs moving out of the Arctic are one cause of CAOs in middle latitudes, especially at more northern locations. In contrast, previous studies of CAOs over North America (Colucci and Davenport 1987; Curry 1987; Colle and Mass 1995; Walsh et al. 2001) often stressed the transport of surface cold air equatorward by surface cyclones and anticyclones.
Our simulations of this event with the atmospheric component of MPAS allowed further insight into TPVs and their relationship to CAOs. Comparison of the control simulations against simulations in which the initial strength of TPV was modified revealed a strong relationship between TPV strength in the Arctic and the subsequent magnitude of the cold surface air over the CONUS. Weaker (stronger) TPVs in the Arctic meant warmer (colder) temperatures to the right of TPV track over CONUS. Changing the initial intensity of the TPV in the Arctic had more modest impact on the TPV track even in a 7-day forecast, although stronger TPVs did have a more southerly track over the CONUS. In terms of predictability for this high-impact events, our control and modified simulation utilizing MPAS all predicted a movement of the TPV and associated CAO into the northern CONUS. The magnitude of the CAO and its southern extent were, however, found to be closely linked to the intensity of the TPV.
The widely utilized definition of Cavallo and Hakim (2005) required that vortices spend greater than 60% of its lifetime within the Arctic to be considered a TPV. Our study and the results of Biernat et al. (2021) suggests a broadening of this lifetime criteria is needed. The importance of CAOs that are linked to TPVs that are ejected relatively rapidly out of the Arctic is an intriguing result given that Papritz et al. (2019) found that TPVS with CAO that move over the Fram Strait tend to have longer residence times in the inner Arctic. This difference supports the suggestion of Papritz et al. (2019) that the linkage between TPVs and CAOs needs to be investigated across various regions.
From our climatological investigation, the path for TPVs associated with CAOs over North America show a movement of TPVs out of the northern edge of the Canadian Arctic (western Canadian Archipelago) south-southeastward into the Midwest and Great Lakes region of the United States and then toward the Northeast toward Labrador. This movement is consistent with the results of Biernat et al. (2021). Our composites of CAOs in the observations suggest a dynamical linkage in that the CAO centroid lies to the southwest of a minimum in tropopause temperatures and poleward of the entrance region of a jet streak. This relationship is expected given the region to the southwest of the TPV cyclonic centroid will be associated with the advection of cold northerly flow. Our case study, however, suggests that the location of the surface CAO relative to the TPV core on the tropopause also results from the vertical tilt of the TPV.
As noted in Papritz et al. (2019), the atmospheric moisture will increase as the Arctic warms so that TPVs may weaken as the impacts of latent heat increase relative to the role of longwave radiation. Given the association between TPVs and CAOs in our study and the increase in the ice-free areas in a warming Arctic, we concur with the hypothesis of Papritz et al. (2019) that the potential weakening of TPVs with climate change could also reduce the intensity and frequency of CAOs, especially for the most extreme events. Screen et al. (2015) showed that in a changing climate, Arctic sea ice loss reduces the risk of extreme cold in North America. In addition, the observations and MPAS simulations in our study revealed that the intensity of the TPV in the lower levels was reduced by diabatic processes as the system moved into middle latitudes with a particularly rapid weakening when the TPV moved over the ice-free Great Lakes. A delay in the freezing of the lakes in a warmer climate is likely to decrease the intensity of CAOs in this region and increase lake-effect snows when TPVs move over ice-free warmer water.
In the event investigated in this study, the TPV did pass over the sea ice region between Greenland and the Yukon. Since TPVs that are associated with CAOs spend more time in the Western Hemisphere side of the Arctic (Fig. 18), the origin and location along the equatorward track will matter. Given that our study showed a common track for TPVs moving out of the Arctic into North America, another implication of this study is whether an increase in the waviness of the PJS will occur with a warming Arctic and, if so, the impact of such changes on the frequency of extreme CAOs due to intense TPVs being ejected out of the Arctic. As noted in Blackport and Screen (2020), the relationship between a warming Arctic and waviness of the PJS is currently being debated. The final implication of our study related to a changing climate is the dependence of the intensity and equatorward movement of the TPV and associated CAO on the horizontal grid of the model. This dependence adds a significant caution on the utilization of coarse-grid model to draw inferences on how a warming Arctic impacts the extreme CAO events in middle latitudes.
Acknowledgments
We appreciate the discussions with members of the AAARG research group, Kevin Biernat (University at Albany, State University of New York), and members of the ONR DRI team such as James Doyle, Daniel Eleuterio, Ron Ferek, and others. We also thank the Office of Naval Research for support of this research under Grants N00014-16-1-2489, N00014-18-1-2223 (SMC), and N00014-18-1-2163 (DP).
Data availability statement
The FNL-GDAS analysis are readily available to the public at National Centers for Environmental Prediction/National Weather Service/NOAA/U.S. Department of Commerce, 2015, updated daily. NCEP GDAS/FNL 0.25° global tropospheric analyses and forecast grids are available from the Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory: https://doi.org/10.5065/D65Q4T4Z. The observations utilized can obtained through NOAA’s National Centers for Environmental Information (NCEI). The MPAS simulations and TPV climatology are available upon request.
REFERENCES
Agee, E. M., and S. R. Gilbert, 1989: An aircraft investigation of mesoscale convection over Lake Michigan during the 10 January 1984 cold air outbreak. J. Atmos. Sci., 46, 1877–1897, https://doi.org/10.1175/1520-0469(1989)046<1877:AAIOMC>2.0.CO;2.
Biernat, K., L. Bosart, and D. Keyser, 2021: A climatological analysis of the linkages between tropopause polar vortices, cold pools, and cold air outbreaks over the central and eastern United States. Mon. Wea. Rev., 149, 189–206, https://doi.org/10.1175/MWR-D-20-0191.1.
Blackport, R., and J. A. Screen, 2020: Insignificant effect of Arctic amplification on the amplitude of midlatitude atmospheric waves. Sci. Adv., 6, eaay2880, https://doi.org/10.1126/sciadv.aay2880.
Bosart, L. F., G. J. Hakim, K. R. Tyle, M. A. Bedrick, W. E. Bracken, M. J. Dickinson, and D. M. Schultz, 1996: Large-scale antecedent conditions associated with the 12–14 March 1993 cyclone (“Superstorm ’93”) over eastern North America. Mon. Wea. Rev., 124, 1865–1891, https://doi.org/10.1175/1520-0493(1996)124<1865:LSACAW>2.0.CO;2.
Boyle, J. S., and L. F. Bosart, 1983: A cyclone/anticyclone couplet over North America: An example of anticyclone evolution. Mon. Wea. Rev., 111, 1025–1045, https://doi.org/10.1175/1520-0493(1983)111<1025:ACCONA>2.0.CO;2.
Browning, K., A. Thorpe, A. Montani, D. Parsons, M. Griffiths, P. Panagi, and E. Dicks, 2000: Interactions of tropopause depressions with an ex–tropical cyclone and sensitivity of forecasts to analysis errors. Mon. Wea. Rev., 128, 2734–2755, https://doi.org/10.1175/1520-0493(2000)128<2734:IOTDWA>2.0.CO;2.
Cavallo, S. M., 2009: The structure and maintenance of tropopause polar vortices over the Arctic. Ph.D. thesis, University of Washington, 148 pp.
Cavallo, S. M., and G. J. Hakim, 2005: Life cycles of tropopause polar vortices. 15th Conf. on Atmospheric and Oceanic Fluid Dynamics, Cambridge, MA, Amer. Meteor. Soc., JP2.1, https://ams.confex.com/ams/Cambridge/webprogram/Paper92780.html.
Cavallo, S. M., and G. J. Hakim, 2009: Potential vorticity diagnosis of a tropopause polar cyclone. Mon. Wea. Rev., 137, 1358–1371, https://doi.org/10.1175/2008MWR2670.1.
Cavallo, S. M., and G. J. Hakim, 2010a: Composite structure of tropopause polar cyclones. Mon. Wea. Rev., 138, 3840–3857, https://doi.org/10.1175/2010MWR3371.1.
Cavallo, S. M., and G. J. Hakim, 2010b: The composite structure of tropopause polar cyclones from a mesoscale model. Mon. Wea. Rev., 138, 3840–3857, https://doi.org/10.1175/2010MWR3371.1.
Cavallo, S. M., and G. J. Hakim, 2012: Radiative impact on tropopause polar vortices over the Arctic. Mon. Wea. Rev., 140, 1683–1702, https://doi.org/10.1175/MWR-D-11-00182.1.
Cavallo, S. M., and G. J. Hakim, 2013: Physical mechanisms of tropopause polar vortex intensity change. J. Atmos. Sci., 70, 3359–3373, https://doi.org/10.1175/JAS-D-13-088.1.
Cellitti, M. P., J. E. Walsh, R. M. Rauber, and D. H. Portis, 2006: Extreme cold air outbreaks over the United States, the polar vortex, and the large-scale circulation. J. Geophys. Res., 111, D02114, https://doi.org/10.1029/2005JD006273.
Colle, B. A., and C. F. Mass, 1995: The structure and evolution of cold surges east of the Rocky Mountains. Mon. Wea. Rev., 123, 2577–2610, https://doi.org/10.1175/1520-0493(1995)123<2577:TSAEOC>2.0.CO;2.
Colucci, S. J., and J. C. Davenport, 1987: Rapid surface anticyclogenesis: Synoptic climatology and attendant large-scale circulation changes. Mon. Wea. Rev., 115, 822–836, https://doi.org/10.1175/1520-0493(1987)115<0822:RSASCA>2.0.CO;2.
Curry, J., 1987: The contribution of radiative cooling to the formation of cold-core anticyclones. J. Atmos. Sci., 44, 2575–2592, https://doi.org/10.1175/1520-0469(1987)044<2575:TCORCT>2.0.CO;2.
Curry, J., J. L. Schramm, W. B. Rossow, and D. Randall, 1996: Overview of Arctic cloud and radiation characteristics. J. Climate, 9, 1731–1764, https://doi.org/10.1175/1520-0442(1996)009<1731:OOACAR>2.0.CO;2.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, https://doi.org/10.1002/qj.828.
Downton, M. W., and K. A. Miller, 1993: The freeze risk to Florida citrus. Part II: Temperature variability and circulation patterns. J. Climate, 6, 364–372, https://doi.org/10.1175/1520-0442(1993)006<0364:TFRTFC>2.0.CO;2.
Gajanan, M., 2019: 21 people died of the cold during the polar vortex. Time, 1 February, http://time.com/5518469/21-people-died-cold-polar-vortex/.
Grell, G. A., J. Dudhia, and D. R. Stauffer, 1994: A description of the fifth generation Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 121 pp., https://doi.org/10.5065/D60Z716B.
Hakim, G. J., 2000: Climatology of coherent structures on the extratropical tropopause. Mon. Wea. Rev., 128, 385–406, https://doi.org/10.1175/1520-0493(2000)128<0385:COCSOT>2.0.CO;2.
Hakim, G. J., and A. K. Canavan, 2005: Observed cyclone–anticyclone tropopause vortex asymmetries. J. Atmos. Sci., 62, 231–240, https://doi.org/10.1175/JAS-3353.1.
Hong, S.-Y., and H.-L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev., 124, 2322–2339, https://doi.org/10.1175/1520-0493(1996)124<2322:NBLVDI>2.0.CO;2.
Hoskins, B. J., M. McIntyre, and A. W. Robertson, 1985: On the use and significance of isentropic potential vorticity maps. Quart. J. Roy. Meteor. Soc., 111, 877–946, https://doi.org/10.1002/qj.49711147002.
Iacono, M. J., J. S. Delamere, E. J. Mlawer, M. W. Shephard, S. A. Clough, and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, https://doi.org/10.1029/2008JD009944.
Klemp, J. B., 2011: A terrain-following coordinate with smoothed coordinate surfaces. Mon. Wea. Rev., 139, 2163–2169, https://doi.org/10.1175/MWR-D-10-05046.1.
Konrad, C. E., II, 1996: Relationships between the intensity of cold-air outbreaks and the evolution of synoptic and planetary-scale features over North America. Mon. Wea. Rev., 124, 1067–1083, https://doi.org/10.1175/1520-0493(1996)124<1067:RBTIOC>2.0.CO;2.
Konrad, C. E., II, 1998: Persistent planetary scale circulation patterns and their relationship with cold air outbreak activity over the eastern United States. Int. J. Climatol., 18, 1209–1221, https://doi.org/10.1002/(SICI)1097-0088(199809)18:11<1209::AID-JOC301>3.0.CO;2-K.
Konrad, C. E., II, and S. J. Colucci, 1989: An examination of extreme cold air outbreaks over eastern North America. Mon. Wea. Rev., 117, 2687–2700, https://doi.org/10.1175/1520-0493(1989)117<2687:AEOECA>2.0.CO;2.
Massey, F. J., 1951: The Kolmogorov-Smirnov test for goodness of fit. J. Amer. Stat. Assoc., 46, 68–78, https://doi.org/10.1080/01621459.1951.10500769.
Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343–360, https://doi.org/10.1175/BAMS-87-3-343.
Miller, K. A., and M. W. Downton, 1993: The freeze risk to Florida citrus. Part 1: Investment decisions. J. Climate, 6, 354–363, https://doi.org/10.1175/1520-0442(1993)006<0354:TFRTFC>2.0.CO;2.
Morgan, M. C., and J. W. Nielsen-Gammon, 1998: Using tropopause maps to diagnose midlatitude weather systems. Mon. Wea. Rev., 126, 2555–2579, https://doi.org/10.1175/1520-0493(1998)126<2555:UTMTDM>2.0.CO;2.
Myrick, D., 2017: Global Forecast Systems (GFS) upgrade: Effective July 19, 2017. National Weather Service Change Notice 17-67, 10 pp.
Nielsen-Gammon, J. W., and D. A. Gold, 2008: Potential vorticity diagnosis of the severe convective regime. Part II: The impact of idealized PV anomalies. Mon. Wea. Rev., 136, 1582–1592, https://doi.org/10.1175/2007MWR2091.1.
NWS, 2019: January 30-31, 2019: Record to near-record cold in northern Illinois. National Weather Service, https://www.weather.gov/lot/RecordColdJan2019.
Palmer, C. E., 1959: The stratospheric polar vortex in winter. J. Geophys. Res., 64, 749–764, https://doi.org/10.1029/JZ064i007p00749.
Papritz, L., E. Rouges, F. Aemisegger, and H. Wernli, 2019: On the thermodynamic preconditioning of Arctic air masses and the role of tropopause polar vortices for cold air outbreaks from Fram Strait. J. Geophys. Res. Atmos., 124, 11 033–11 050, https://doi.org/10.1029/2019JD030570.
Pedlosky, J., 1998: Geophysical Fluid Dynamics. 2nd ed. Springer, 728 pp.
Provenzale, A., 1999: Transport by coherent barotropic vortices. Annu. Rev. Fluid Mech., 31, 55–93, https://doi.org/10.1146/annurev.fluid.31.1.55.
Pyle, M. E., D. Keyser, and L. F. Bosart, 2004: A diagnostic study of jet streaks: Kinematic signatures and relationship to coherent tropopause disturbances. Mon. Wea. Rev., 132, 297–319, https://doi.org/10.1175/1520-0493(2004)132<0297:ADSOJS>2.0.CO;2.
Rogers, J. C., and R. V. Rohli, 1991: Florida citrus freezes and polar anticyclones in the Great Plains. J. Climate, 4, 1103–1113, https://doi.org/10.1175/1520-0442(1991)004<1103:FCFAPA>2.0.CO;2.
Sanders, F., and J. R. Gyakum, 1980: Synoptic-dynamic climatology of the “bomb.” Mon. Wea. Rev., 108, 1589–1606, https://doi.org/10.1175/1520-0493(1980)108<1589:SDCOT>2.0.CO;2.
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, https://doi.org/10.1175/BAMS-D-14-00185.1.
Shapiro, M., T. Hampel, and A. Krueger, 1987: The Arctic tropopause fold. Mon. Wea. Rev., 115, 444–454, https://doi.org/10.1175/1520-0493(1987)115<0444:TATF>2.0.CO;2.
Skamarock, W. C., J. B. Klemp, M. G. Duda, L. D. Fowler, S.-H. Park, and T. D. Ringler, 2012: A multiscale nonhydrostatic atmospheric model using centroidal Voronoi tesselations and C-grid staggering. Mon. Wea. Rev., 140, 3090–3105, https://doi.org/10.1175/MWR-D-11-00215.1.
Stein, A., R. R. Draxler, G. D. Rolph, B. J. Stunder, M. Cohen, and F. Ngan, 2015: NOAA’s HYSPLIT atmospheric transport and dispersion modeling system. Bull. Amer. Meteor. Soc., 96, 2059–2077, https://doi.org/10.1175/BAMS-D-14-00110.1.
Szapiro, N., and S. Cavallo, 2018: TPVTrack v1. 0: A watershed segmentation and overlap correspondence method for tracking tropopause polar vortices. Geosci. Model Dev., 11, 5173–5187, https://doi.org/10.5194/gmd-11-5173-2018.
Thompson, G., R. M. Rasmussen, and K. Manning, 2004: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part I: Description and sensitivity analysis. Mon. Wea. Rev., 132, 519–542, https://doi.org/10.1175/1520-0493(2004)132<0519:EFOWPU>2.0.CO;2.
Thorpe, A. J., 1985: Diagnosis of balanced vortex structure using potential vorticity. J. Atmos. Sci., 42, 397–406, https://doi.org/10.1175/1520-0469(1985)042<0397:DOBVSU>2.0.CO;2.
Thorpe, A. J., 1986: Synoptic scale disturbances with circular symmetry. Mon. Wea. Rev., 114, 1384–1389, https://doi.org/10.1175/1520-0493(1986)114<1384:SSDWCS>2.0.CO;2.
Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117, 1779–1800, https://doi.org/10.1175/1520-0493(1989)117<1779:ACMFSF>2.0.CO;2.
Vavrus, S., J. Walsh, W. Chapman, and D. Portis, 2006: The behavior of extreme cold air outbreaks under greenhouse warming. Int. J. Climatol., 26, 1133–1147, https://doi.org/10.1002/joc.1301.
Wallace, J. M., and D. S. Gutzler, 1981: Teleconnections in the geopotential height field during the Northern Hemisphere winter. Mon. Wea. Rev., 109, 784–812, https://doi.org/10.1175/1520-0493(1981)109<0784:TITGHF>2.0.CO;2.
Walsh, J. E., A. S. Phillips, D. H. Portis, and W. L. Chapman, 2001: Extreme cold outbreaks in the United States and Europe, 1948–99. J. Climate, 14, 2642–2658, https://doi.org/10.1175/1520-0442(2001)014<2642:ECOITU>2.0.CO;2.
Wandishin, M. S., J. W. Nielsen-Gammon, and D. Keyser, 2000: A potential vorticity diagnostic approach to upper-level frontogenesis within a developing baroclinic wave. J. Atmos. Sci., 57, 3918–3938, https://doi.org/10.1175/1520-0469(2001)058<3918:APVDAT>2.0.CO;2.
Waugh, D. W., A. H. Sobel, and L. M. Polvani, 2017: What is the polar vortex and how does it influence weather? Bull. Amer. Meteor. Soc., 98, 37–44, https://doi.org/10.1175/BAMS-D-15-00212.1.
Wheeler, D., V. Harvey, D. Atkinson, R. Collins, and M. Mills, 2011: A climatology of cold air outbreaks over North America: WACCM and ERA-40 comparison and analysis. J. Geophys. Res., 116, D12107, https://doi.org/10.1029/2011JD015711.
Winters, A. C., and J. E. Martin, 2017: Diagnosis of a North American polar–subtropical jet superposition employing piecewise potential vorticity inversion. Mon. Wea. Rev., 145, 1853–1873, https://doi.org/10.1175/MWR-D-16-0262.1.
Winters, A. C., D. Keyser, L. F. Bosart, and J. E. Martin, 2020: Composite synoptic-scale environments conducive to North American polar–subtropical jet superposition events. Mon. Wea. Rev., 148, 1987–2008, https://doi.org/10.1175/MWR-D-19-0353.1.
Zhang, C., Y. Wang, and K. Hamilton, 2011: Improved representation of boundary layer clouds over the southeast Pacific in ARW-WRF using a modified Tiedtke cumulus parameterization scheme. Mon. Wea. Rev., 139, 3489–3513, https://doi.org/10.1175/MWR-D-10-05091.1.
