Blog — S&P Global Sustainable1 — 02 March, 2026
Why climate change doesn’t mean goodbye to snow
The early months of 2026 have seen particularly severe winter weather in the eastern US. A historic winter bomb cyclone struck the Northeast Feb. 23, with blizzard warnings affecting 40 million people from Washington, D.C., to northern Maine. Two feet of snow have been reported in five states. Just a month earlier, a massive winter storm system tracked across the US, impacting nearly two dozen states. It featured a perilous mix of snow, sleet and freezing rain with snowfall totals reaching up to 75 cm (30 inches). Fueled by an Arctic air mass colliding with moisture from the Gulf of Mexico and Pacific Ocean, the storm disrupted travel, damaged infrastructure and exposed millions to dangerously cold temperatures while leaving 800,000 customers without power.
This event is only the latest reminder of the economic and societal impacts of winter storms. National Centers for Environmental Information data shows that in the US alone, the economic cost of winter storms between 1980 and 2024 totaled more than $104 billion. Major snowfall events can disrupt essential services and daily life across many sectors, from road closures, cancelled flights and supply chain delays to widespread power outages. About 23% of weather-related US power outages between 2000 and 2023 were caused by winter storms, according to the research nonprofit Climate Central.
Climate change and winter storms
As the climate continues to warm at an unprecedented pace, should we expect winter storms to become less common or eventually all but disappear? Are we saying goodbye to snow? The truth is, it depends on where you live.
The recipe for major snow calls for two main ingredients: Freezing temperatures on the ground as well as throughout the air column, and sufficient moisture in the atmosphere. With warming temperatures, we at the S&P Global Climate Center of Excellence anticipate the number of days with freezing temperatures to decline overall. However, the atmospheric moisture content is aided by higher temperatures, as warmer air is capable of providing more moisture for snowflake formation.
As a result, in some regions, the drop in the number of freezing days is counteracted by this increased moisture availability, leading to both more frequent and more intense snowfall events. In addition, major snowstorms are often accompanied by cold snaps, and projected changes in the atmospheric circulation may continue to allow Arctic air to escape farther south — conditions that produce intense cold snaps accompanying major snow events.
Modeling challenges
Modeling projected snowfall is challenging and is very different from predicting specific snowfall events on weather time scales. Examining some of the differences between our work in the S&P Global Climate Center of Excellence and that of meteorologists, who predict specific snowstorms, helps illustrate some of the scientific challenges.
Meteorologists focus on how much rain or snow will occur with a given storm and grapple with the exact timing of the event. Climatologists have similar concerns, but the exact date and hour of the event is not a focus. Meteorologists usually have more data to work with, including detailed vertical profiles of atmospheric wind, temperature and moisture from both observations and regional weather models. Additionally, because global climate models have coarser resolution — for example, on the scale of 100 km rather than 10 km — they typically have larger biases than regional weather models.
At the S&P Global Climate Center of Excellence, we have developed surface-based statistical relationships based on ERA5-Land1 reanalysis data. These relationships reveal which specific combinations of climate variables are more likely to result in snow. We pair this information with output from downscaled (25 km) global climate models to produce in-house snowfall projections.
Before highlighting some of our results, it is important to emphasize that our goal is to project and understand long-term trends related to the warming effect of human-caused emissions. One of the ways to show these trends is to model the number of days in a decade that reach a certain threshold for a given climate event. Our approach enables us to make statements such as: “On average in the 2050s, under a given emissions scenario, this location will see a decrease in snow days.” Individual days and years in the future will deviate from the average view presented here.
Additionally, our definition of “snow days” requires that the total snowfall for that day exceed 2.5 mm (0.1 inch) liquid-water-equivalent 2. We choose this threshold deliberately to focus on snow days of practical significance that could yield hazardous conditions. This is a subtle point but explains why the number of snow days can seem lower than expected in some colder regions in our projections (Figure 1).
Take Canada, for example: Despite its high latitude location, the number of snow days there is comparable to the central US when using the higher threshold (Figure 1, left panel). This occurs because high latitudes tend to be moisture limited, meaning it is so cold that the atmospheric moisture is too low for more significant snow days.
Projected snowfall changes
The Snowfall hazard we are developing will join the other climate hazards available in the S&P Global Sustainable1 Physical Risk dataset. Below, we offer a first glance at two of our metrics: the number of projected snow days and the maximum daily total snowfall. We produce these metrics on maps showing the change relative to the baseline climate of 1950 to 2014. The magnitude of each signal will depend on the selected climate change scenario, each of which reflects varying levels of human-driven global warming by end-century. SSP1.2-6 shows the least and SSP5.8-5 shows the most change. Three takeaways stand out:
- Changes in the number of snowfall days and the snowfall amounts show strong regional dependence.
Regional climate variables like proximity to water bodies, elevation and local topography all influence the amount and frequency of projected snowfall events. For instance, we see strong decreases along the US eastern seaboard, the west coast of North America and Central Europe. This geographic variability emphasizes that climate impact and risk assessments cannot be done through global averages alone. Instead, access to high-resolution local data and models that capture these regional nuances are needed. - Counterintuitively, some regions will experience an increase in snow days and snowfall amount as the climate warms.
Temperature and moisture availability are the two important limiting factors in snowfall. In a warming climate, air can transport more moisture and can actually offset the decline in days cold enough for significant snow. We see this modeled in northern parts of Canada, Alaska and Russia. In these regions, projections show a moderate increase in the number of snow days and the snow amount even at the end of the century. The eastern shore of the Hudson Bay in Canada sees an increase in snowfall and snow days, attributable to the moisture available from the bay and its delayed freezing with continued warming. - Beginning and end of century differences matter.
In fact, these differences matter so much that some regions show opposing trends. Take North America as an example: Until roughly mid-century, depending on the scenario, we see some places with increasing or unchanging snow days and amounts, such as the northern US border and southern Canada. After that, between 2050 and 2100, the increase in temperature dominates and we see a decline in snow days and snowfall totals. In the near term, regions like this will still need their snow shovels.
Climate change has a nuanced impact on snowfall. On one hand, we are certainly seeing a decrease in the number of days cold enough to produce snow from a global perspective. On the other hand, this decrease can be offset by the increased moisture content that becomes available from warmer air. In those locales, the added moisture can boost snowfall, and climate models predict modest snowfall increases despite the background warming.
Overall, the strong regional dependence of our snowfall projections underscores a critical takeaway for organizations and policymakers: Access to regional and locally resolved hazard data is essential for accurate climate impact assessments. Climate adaptation planning is more effective when it is based on model and data-driven studies that show how warming reshapes the climate differently across regions, sometimes with counterintuitive results.
1 ERA5-Land is a high-resolution land surface reanalysis dataset that is produced by the European Centre for Medium-Range Weather Forecasts. It provides hourly, gridded, estimates of climate variables (here we use temperature, snowfall, and precipitation) at roughly 0.1° spatial resolution from 1950 to present.
2 The liquid water equivalent tells us the amount of water that would be measured if all the snow were melted in a rain gauge.