When a full-fledged snowstorm descended on the U.S. Gulf Coast in mid-January 2025, followed by a night of unprecedented cold across some bayous and beaches (see Part I of this two-part post), it did more than turn New Orleans and Mobile into winter wonderlands. It reignited a topic that’s cropped up again and again across science and media landscapes since the early 2010s: Is a warming Arctic affecting winter extremes over North America and Eurasia? And if so, how?
There’s actually a vigorous scientific debate on this topic. The debate might be opaque to casual news readers, but it’s important to everyone, from people who love winter weather to those tasked with preparing for it — including the millions affected by a sharp cold wave this week across the central and eastern U.S., accompanied by a belt of significant snow from Oklahoma to North Carolina.
Those who dismiss or deny climate change sometimes use a high-profile cold wave or snowstorm to inject snark, such as, “Enjoying this global warming yet?” Others sometimes leap from connections still being researched to make what look like ironclad conclusions, as in this often-memed statement: “The frigid temperatures you’re experiencing are happening BECAUSE of a warming climate, not in spite of it.”
While people fling snowballs back and forth on social media, dozens of scientists have been tackling the topic in much greater depth for more than a decade. One group has doggedly analyzed upper-air patterns and how they relate to Arctic amplification, a term for the observation that the Arctic is warming faster than other areas of the world. Their goal: explain why cold and snow extremes don’t seem to be fading everywhere, as one might expect in a warming climate.
Another group has just as doggedly scrutinized decades of observations and computer-model replications of recent climate. They’ve confirmed that the sharpest cold extremes are becoming less frequent across most of the midlatitude Northern Hemisphere, the broad belt between roughly 35 and 65 degrees north of the equator that covers much of the U.S., Canada, Europe, Russia, and China. And they suspect natural climate variation – rather than a rapidly warming Arctic – most likely explains why cold and snow extremes have maintained their edge in a few areas over the last several decades.
There’s also a newer line of research folding high-resolution weather models into global climate projections. It’s starting to yield more clues as to how U.S. snowfalls might evolve later this century, including what one group sees as an eventual near-elimination of Southern snowstorms.
The bottom line: It seems likely that most Americans will see a decrease in winter weather over the coming decades. Yet there are some crucial caveats – including the risk of putting winter’s hazards on the shelf way too quickly.
The evolution of Arctic-midlatitude research
As far back as the 1990s, scientists warned that not all effects of climate change would be straightforward. One example is the “wet get wetter, dry get drier” paradox. A warming atmosphere allows more moisture to evaporate from dry landscapes, worsening drought impacts, as well as from oceans, helping intensify extreme rainfall.
Warmer air and extra moisture aren’t usually good for snowstorms. But in certain cases and places, they might increase snowfall – as long as temperatures stay just cold enough to produce it.
By the 2010s, scientists were starting to analyze how midlatitude winter storms might be affected by accelerated Arctic warming, especially by the decline in Arctic sea ice extent. Sea ice loss as measured in September, the typical time of the annual minimum, intensified in the 2000s and hit its current record low in 2012. Since then, it has oscillated around a “new abnormal” low range (see Fig. 1 below), with more losses to come. Winter losses have been dramatic as well: Arctic sea ice extent in recent days has been at record-low values for early February.
“This persistent new normal, and the related losses of most of the old and thick ice, are prominent characteristics of the new, warmer Arctic – resiliently low in ice cover despite more than a decade of variations in seasonal climate patterns and ocean conditions,” the National Snow and Ice Data Center reported in October 2024.
A review paper in Science published on February 6 and led by Julienne Stroeve of the National Snow and Ice Data Center warns that a plausible global temperature rise of 2.7 degrees Fahrenheit above preindustrial levels would make for an Arctic “transformed beyond contemporary recognition,” with the Arctic virtually ice-free each summer for months.
Does a melting Arctic influence U.S. weather?
Headlines emerged in the wake of a 2012 paper by Jennifer Francis, now at the Woodwell Climate Research Center and Steven Vavrus at the University of Wisconsin–Madison. Francis and Vavrus proposed that Arctic amplification would lead to weaker west-to-east jet-stream winds and an increased frequency of large north-south-oriented upper-level waves in the atmosphere’s circulation. They also hypothesized this shift would allow midlatitude weather extremes – in the U.S. and elsewhere – to become more persistent and the impacts more extreme. Francis and colleagues have since expanded on this work in a number of follow-up papers.
“While it’s clear we’re seeing fewer cold temperature records being broken as the climate warms, the disruption caused by cold spells is being felt in places where debilitating cold is unusual, and so folks and communities are not prepared for it – like this winter in Louisiana, Florida, Greece, and Saudi Arabia,” Francis told Yale Climate Connections by email.
Meanwhile, Judah Cohen at Atmospheric and Environment Research, Inc., had already spent years analyzing sea ice loss in the Barents and Kara Seas, north of Scandinavia and western Russia, looking at how moisture from the open water could help increase autumn snowpack accumulating in Siberia. As mapped out in a series of papers, including this one from 2012, Cohen and colleagues asserted that cold surface high pressure building atop the increased Siberian snowpack could disrupt the jet stream in ways that perturb the stratospheric polar vortex and, over weeks to months, intensify severe winter weather events in northern midlatitudes, including North America.
Studies of Arctic change and midlatitude winter weather now span at least 75 published papers since 2020 alone. These research threads were pulled together in a 2023 workshop and a subsequent sprawling review paper published in December 2024 in Environmental Research: Climate. Along with lead author Edward Hanna at the University of Lincoln in the United Kingdom, the paper includes 19 coauthors (including Francis and Cohen) hailing from eight countries in North America, Europe, and Asia.
Referring to the topic as “a central and controversial area of research and debate,” the paper adds: “Accumulated evidence suggests that recent trends in northern midlatitude winter cold events are due to a combination of influences, including both Arctic warming and interannual variability.”
As highlighted in the review paper, scientists have widened their nets over the years to study not only splits and displacements of the stratospheric polar vortex, but also less-publicized stretching events, when the vortex becomes elongated into an oval or bean shape. These events have been prominent in recent years over North America (see Part I of this series).
In a 2021 Science paper, Cohen presented evidence for an increase in stretching events during the era of Arctic amplification, and in a forthcoming paper, he and colleagues find that extreme winter weather in the central and eastern U.S. tends to occur more often with stretching events than with splits and displacements.
“The impacts from stretching events are not long-lasting, and the changes to the polar vortex are not nearly as dramatic [as with splits and displacements],” Cohen said.
Tropical connections to the Arctic are also being scrutinized more closely. And in a series of studies, Binhe Luo of Beijing Normal University and colleagues have applied nonlinear theory to better understand the complex nature of the atmospheric blocks – upper-level highs that stay in place for days to weeks – that can help keep extreme winter weather, including bitter cold and snow, locked in place for days on end.
“Whenever these blocks form, we see very persistent weather conditions in their vicinity,” Francis explained. “Computer programs that simulate weather and climate struggle to capture these blocks realistically, so understanding them better will lead to more accurate weather forecasts and projections of extreme weather in the future.”
A different take on the Arctic’s role
Assuming that the polar stratosphere is influenced by climate change, has this translated into reliable changes to winter extremes across the U.S. and other midlatitude areas? A number of influential climate scientists aren’t yet convinced. They include Andrew Dessler at Texas A&M University.
“Until someone shows me statistics demonstrating that winter storm severity or frequency is getting worse as the climate warms, I’m not going to believe that any [such] individual event is connected to climate change,” Dessler wrote by email. “There are people I respect who think there is a connection, and we will argue it out the same way the scientific community handles all disagreements. However, given the lack of consensus, I think it’s wrong to confidently connect climate change to these cold events.”
Isla Simpson at the NSF National Center for Atmospheric Research, who studies atmospheric dynamics and how they’re captured in global climate models, said in an email, “I think that the large internal [natural] variability in both the stratospheric polar vortex and in midlatitude weather makes it really challenging to make conclusive statements about whether we are seeing long term trends that are making cold extremes more likely.”
Such natural variability could help explain why winters at some locations have gone decades without having warmed in lockstep with the global trend. The most dramatic case is northeast Asia, where winter temperatures actually cooled slightly from the 1990s to 2010s. There was also little winter warming or slight cooling from the mid-20th century to early this century over parts of the U.S. Ohio Valley and Southeast, part of a long-studied “warming hole” (itself now on the decrease) where multiple factors may have been at work, including reforestation and sun-blocking postwar pollution.
Regional winter temperatures that fail to rise in tandem with global warming can themselves be a noteworthy thing, Cohen argues, as reflected in the title of a 2023 study he led: “No detectable trend in midlatitude cold extremes during the recent period of Arctic amplification.”
Cohen wonders about the risks of, as he puts it, “telling people winters are going to get warmer and you’re never going to see snow again in your life, and then you see snow on palm trees on the Gulf Coast.” He argues for stressing the point that winter events can still happen and still be severe, even if they aren’t getting more common: “I think this increases, not decreases, your credibility.”
In a 2024 paper that’s one of the few in this realm to assess both snowfall and temperature, Cohen, Francis, and Karl Pfeiffer – another researcher at Atmospheric and Environment Research, Inc. – highlight a 1950-to-2023 trend toward more variable winter weather across the northern U.S., southern Canada, and northeast Asia and decreased variability over Europe and the Arctic Ocean. As they did in a 2018 U.S.-focused study, the authors use the Accumulated Winter Season Severity Index, a tool developed by the Midwestern Regional Climate Center that incorporates the intensity and persistence of both cold and snow from fall into spring compared to what’s typical at a given location. It’s also been called the “misery index.” (The index for 2024-25 through Feb. 17 shows “mild”, “moderate”, or “average” for most of the country – levels 1, 2, and 3 out of 5 – with a few pockets of “severe” and “extreme” conditions.)
The index showed widespread decreasing trends throughout the Northern Hemisphere for the period 1950-2023, but increasing trends across much of northern North America and parts of Eurasia during recent decades when the Arctic warmed rapidly (see Fig. 5 below). The study also found (not shown in the figure) that periods of unusual warmth extending a few miles above the Arctic were strongly correlated with high winter severity throughout the northern midlatitudes over the 1950-2023 period. In other words, northern midlatitude winters seem to have trended less severe over the past 75 years, but there’s been an apparent uptick in cold and/or snowy winters over some areas since 2000. And when those winters do get rough, they appear to be tightly linked to Arctic warming.
How do natural variations fit into the picture?
Other experts have long argued that some of the trends analyzed by Cohen and colleagues aren’t prolonged or consistent enough to take them beyond the range of natural variability – and thus they could change at any time.
Russell Blackport, a researcher at Environment and Climate Change Canada, James Screen of the University of Exeter, and colleagues have returned to this theme in multiple studies. In a paper from Science Advances published in October 2024, they found that across northern midlatitudes from 1990 to 2022, the three-month average winter temperature at each point warmed on average by 0.25°C per decade. But as shown in the left-hand side of Fig. 6 (below), the coldest single day per year at each point warmed almost twice as fast as the winter-long average – by 0.42°C per decade. In short, not only are full winters getting warmer on average, but the coldest winter days are getting warmer at an even faster clip.
Left: the lowest daily average temperature for each winter (TMn).
Right: the number of days each winter in which the temperature is in the coldest 5% of the long-term period (TM5p).
Datasets shown include the ERA5 reanalysis of observations (black), Berkeley Earth observations (orange), JRA55 observations (red), and the mean of all model ensemble members that were used to replicate winter conditions (blue). The blue shading indicates the 2.5 to 97.5 percentile range of the model spread from all ensemble members. The anomalies are relative to the long-term period of 1971-2022. (Image credit: From Blackport, Sigmond, and Screen, “Models and observations agree on fewer and milder midlatitude cold extremes even over recent decades of rapid Arctic warming,” Science Advances, 2 October 2024, DOI: 10.1126/sciadv.adp134).
The results differed for the shorter time spans of 1990-2013 and 2000-2013, reinforcing how much the start and end years of a trend analysis can affect the results.
Using climate models in the same study to replicate the last few decades of winter weather, Blackport and colleagues found that every simulation showed a few regions with more frequent or intense cold waves over recent decades.
“However, these occur in different regions in different simulations, so when you average over all simulations (and thus average over the internal variability), what you are left with is a global warming signal of a strong decrease in intensity/frequency everywhere,” Blackport said in an email. “The longer-term trends (50-70 years) show decrease in intensity/frequency everywhere, including the regions where there are increases in more recent, short-term trends.”
“I think we all recognize that cold extremes can arise naturally without Arctic amplification and that they will continue to occur in a warming climate,” Screen wrote by email. “I think we also agree that Arctic amplification could affect midlatitude weather in some ways, but whether it results in more cold extremes is still a point of contention.
“In some respects, the difference of opinion lies with where you put the ‘burden of proof’. We are saying that the observed variability and trends in cold extremes are consistent with natural (internal) variability. In a sense, the Arctic is innocent until proven guilty … But innocence is also hard to prove. Although observed variability and trends in cold extremes are consistent with natural (internal) variability, it doesn’t necessarily mean Arctic warming has no influence. Essentially, we’re somewhat at an impasse because it’s hard to fully prove or disprove a connection. To continue the analogy, the ‘jury’ are split.”
The future of U.S. snow in a 21st-century climate
How top-end snowstorms will change in a warming world is an even tougher nut to crack than extreme cold. What’s fairly straightforward is that rising temperatures will push snow-favored zones poleward over time. But it’s also possible that a few regions could see more and/or heavier snowstorms, at least while those areas are still cold enough for regular snowfall.
New modeling approaches are helping to tackle this problem in more detail. Walker Ashley at Northern Illinois University and colleagues examined the future of snowstorms over central and eastern North America in a paper now under review. He previewed the new work at the American Meteorological Society’s 2025 annual meeting in New Orleans – just a week ahead of January’s historic snowstorm there – as part of a session called “Winter Weather in a Warming World.”
To gauge how global-scale changes could affect local snowfall, Ashley’s team embedded a high-resolution regional weather model (WRF-ARW) in two sets of future-climate simulations from the Community Earth System Model, one based on a high-end emissions scenario (RCP8.5) and one on a midrange scenario (RCP4.5). The regional weather models were run for 15-year-long time slices at midcentury (2040-2055) and late-century (2085-2100). This dynamic-downscaling approach – which automatically incorporates any changes to atmospheric structure and circulation that emerge in a global model – is the same one used by the group in a 2023 study to analyze future changes in supercell thunderstorms.
Overall, the snowfall changes projected in the new work are as one might expect. The number, duration, area, and amount of water held in snowfall all drop by a few percent in both scenarios by midcentury, with the reductions exceeding 25% by late century in the high-end scenario. Snow seasons also grow shorter, with sharp drop-offs in early autumn and mid-spring that take shape as soon as midcentury for both midrange and high-end scenarios.
And in both scenarios, the frequency of the most extreme snowfalls (see Fig. 7) plunges by more than 50% during the next several decades across most of the U.S. South – eventually producing what the researchers call a “near removal of snowstorms” from Texas to the mid-Atlantic.
The one carrot for snow-lovers in this study is an increase in snowfalls that would rank in the top 10% by historical standards (see right-hand column of Fig. 7), albeit only across patchy areas, mainly from the Northern Plains to the Northeast.
If a warmer climate were simply pushing existing snow-favored areas northward across the U.S., one wouldn’t expect such an increase at all.
“The basic pattern I expect with climate warming is increases in snowfall in very cold regions and decreases in warmer regions,” says Paul O’Gorman, an atmospheric science professor at the Massachusetts Institute of Technology who studies how the hydrologic cycle responds to climate change. These decreases are already showing up in annual snowfall totals, O’Gorman added.
Colder air can “hold” less moisture, so the heaviest snow in today’s climate tends to be in moderately cold areas, where temperatures are in the optimal temperature range for snowmaking, rather than in more frigid areas – for example, in southern rather than northern Canada. But as the most snow-favorable sweet-spot regions push northward, they might also draw in more moisture.
It’s also possible, said O’Gorman, that the evolving dynamics of snowstorms could lead to intensified snow rates in some areas, as indicated by global climate models. And since those models tend to underplay recent winter extremes, any intensification could be even sharper than the models suggest.
As Ashley puts it: “Unquestionably, the changing snow landscape across North America is complex and uncertain.” At the New Orleans meeting, he stressed the need for more experiments with different models, while noting the massive demand on computing power and storage they can pose. “Hopefully, in 10 to 20 years, as computing resources improve, we’ll have a dozen or more of these models at any one time to examine,” Ashley said. “This will provide a lot more explanatory power and reduce uncertainty in our projections.”
Jeff Masters contributed to this post.
!function(f,b,e,v,n,t,s)
{if(f.fbq)return;n=f.fbq=function(){n.callMethod?
n.callMethod.apply(n,arguments):n.queue.push(arguments)};
if(!f._fbq)f._fbq=n;n.push=n;n.loaded=!0;n.version=’2.0′;
n.queue=[];t=b.createElement(e);t.async=!0;
t.src=v;s=b.getElementsByTagName(e)[0];
s.parentNode.insertBefore(t,s)}(window, document,’script’,
‘https://connect.facebook.net/en_US/fbevents.js’);
fbq(‘init’, ‘568493883318626’);
fbq(‘track’, ‘PageView’, {“page_title”:”Update: Howu2019s U.S. winter weather changing in a warming world?”,”user_role”:”guest”,”event_url”:”https://yaleclimateconnections.org/2025/02/update-hows-u-s-winter-weather-changing-in-a-warming-world”,”post_type”:”post”,”post_id”:127391});
