Future winters promise less snow, more rain. Nobody’s prepared

Blue veins of ice streaked the snow this January in Salt Lake City, Utah. Snow hydrologist McKenzie Skiles eyed the veins, worried. The blue ice formed where water had flowed, then refrozen. “That’s concerning,” Skiles says, “because it tells us snow is undergoing midwinter melt.” She pulled out a thermometer and found the snow near its melting point of 32 degrees Fahrenheit.

Winter mountains look amiss without thick drifts of snow. In many high-elevation zones, climate change promises fewer flakes and more rain over time. Image credit: Benjamin Hatchett/Colorado State University, Fort Collins.

In Salt Lake, snow shouldn’t melt in January. It typically piles through early April, the historical peak snowpack for cold, high western mountains. Melting snow starts dripping by midmonth, feeding creeks all summer.

But the temperature swings of climate change have arrived in Utah and other snowy places. Long warm stretches now punctuate winter. During a weeklong February heatwave, Salt Lake hit a record 65 degrees Fahrenheit—20 degrees above the winter average. “You can’t help but think, ‘Is this every future winter?’” Skiles says from her office at The University of Utah. “Is it just going to keep getting worse?’

Studies from her lab and others find that less snow is falling on mountains worldwide, and there’s more rain in the forecast. Satellites looking down on the Himalayas have recorded 13 fewer snowy days per year there since 2002 (1). And climate models of California’s Sierra Nevada Mountains predict that, at 3 degrees warming, more than half the range’s precipitation will fall as rain, not snow (2). That would be disastrous for the Golden State, where snowmelt from the Sierras is a third of the water supply. California simply doesn’t have the infrastructure to capture all that water from rain.

More rain will also change flood risks, increasing the danger on high mountains where midwinter melt is rare, but perhaps decreasing it at lower elevations, where less snow falling early in the season will mean less melt during winter rains. Overall, less snow compromises drinking and agricultural water storage in the West. States have already started to consider adaptations such as building new reservoirs, restoring floodplains, and expanding snow monitoring. How to adapt to less snow and less consistency, Skiles says, is “the million-dollar question.”

Flakes Falling

For decades, the primary means of tracking mountain snow has been a series of meteorological stations poking up from clearings, each station outfitted with wind, temperature, precipitation, and snow depth sensors. All the stations have a 6- to 8-foot outdoor pillow filled with fluid that measures the weight of the snow that falls on it. Called the SNOTEL network (short for “snow telemetry”), the stations, numbering over 900, are scattered across the West. Stations collect daily temperature, precipitation, snow depth, and weight data. Hydrologists use these data to estimate how much water the snowpack stores across the network every year. Statistical models then translate those measurements into streamflow forecasts. SNOTEL primarily exists to forecast water supplies based on snowpack. But several decades ago, station data began to hint at a rainier future, a conclusion that newer studies support and refine.

One of the first big papers was published in 2003, when climate scientist Philip Mote asked if historical and SNOTEL data revealed climate impacts of changing temperature on snowpack (3). As far back as the 1910s, he explains, researchers would trek into the mountains in snowshoes, and later on snowmobiles, to stick a pipe into the snow and then weigh that pipe to estimate how much water the snowpack held. This went on until around the 1990s, when SNOTEL sites were installed. “In 2003, I was looking at those measurements and wondering, ‘does temperature have any influence on snowpack?’” Mote says. “And to my surprise, it has profound influence.”

At the time, other studies using aerial surveys, satellite imagery, and historical measurements made by hand had suggested declines in the snow-covered area of the northern hemisphere since the 1960s (4). And, sure enough, when Mote compared temperature and precipitation measurements from SNOTEL stations and manual, historical data in Idaho, Oregon, Washington, and Montana between 1950 and 2000, he found declines in snow cover at most sites.

The most dramatic losses hit coastal regions, such as the Cascade Mountains in Washington. Stations at mid-elevations there had lost more than 40% of their snow cover since 1950. Temperatures hadn’t risen enough to impact snow at high elevations in 2003. But at mid-elevations, where it might snow on a cold day and rain on a hot one, warming sites showed an uptick in rain and a decline in snow over the 50 years.

“I was astonished how strong the signal of a warming climate was showing up,” Mote says from Oregon State University in Corvallis. He repeated the study in 2018 with 15 more years of data (5). By then, snowpack had declined in 90% of the long-term records since 1950, including those from SNOTEL sites.

In high-elevation Utah, near 9,000 feet, snow should accumulate and stick all winter. Researcher McKenzie Skiles and her team dig a pit to track shifts in snow depth and the volume of water stored there. Image credit: Otto Lang and McKenzie Skiles/The University of Utah, Salt Lake City.

Simulating Snow

Today’s climate model studies echo those early warnings of a rainier tomorrow. Dozens of models now estimate climate trends by representing physical processes in the atmosphere, on land, and in the ocean. These models are governed by equations representing fluid flow, conservation of mass, momentum and energy, and other processes.

While models don’t agree on every detail of snowpack projections, there are some broad areas of consensus, according to a 2021 review of more than 300 studies (6). Most predicted that coastal mountain ranges will lose water trapped by snow more quickly than interior mountain ranges, with expected snow cover losses of 20% by 2030 and up to 40% by 2070. Inland ranges, such as the Great Basin and Upper Colorado River, could eventually expect the same degree of snow loss, the review found, but 15 to 20 years later.

Models suggest that water stored by western snowpack could decline by about 25% by 2050. And low-to-no-snow years—which the authors defined as 30% or less of the historical peak water storage—could become the norm within two generations if emissions continue unchecked.

“It’s trying to understand, ‘how much time do we have?’,” says hydrologist and civil engineer Laurie Huning, who coauthored the review from California State University, Long Beach. In the West, she says, “a low-to-no-snow future may become more persistent in 35 to 60 years.”

Drilling Down

By refining climate and earth system models, researchers are starting to hone estimates of just how rainy it may get. On average, models see the world as a series of 110-square-kilometer grid cells, explains climate scientist Elizabeth Burakowski, at the University of New Hampshire in Durham. But that resolution might not suffice to understand what’s happening on a mountain range. Peaks and valleys blur together, so the mountain becomes one smooth, brown, fuzzy surface.

Over the last several years, Burakowski has helped test a “variable resolution community earth system model,” or VR-CESM, with higher resolution from 14-square-kilometer grid cells (7, 8). Now, researchers can discern, say, a jagged hillside where the same storm might dust steep upper slopes with snow but lash slightly lower meadows with rain. Last year, she partnered with Areidy Beltran-Peña, an earth system scientist and postdoc at Stanford University in California, to use the VR-CESM to simulate future snowpack and rain in California’s Sierra Nevada region (2). Beltran-Peña recalls reading an April 1776 diary entry in which Spanish missionary Pedro Font first labeled California’s iconic mountain range on a map in his notebook. He dubbed the snowcapped peaks the Sierra Nevada, or “snowy mountain range.” If Font saw those same mountains 50 years from now, Beltran-Peña wondered, would he call them the Sierra Lluviosa (“rainy range”) instead?

He might. The study, presented at the American Geophysical Union meeting last December, fed the model a high-emissions scenario for the years 2015 to 2100 and predicted snowpack for the Sierras under 1.5 degrees, 2 degrees, and 3 degrees of warming. In a 3-degree warming scenario, the model suggests that more than half of the Sierra’s precipitation could fall as rain. “The Sierra Nevada hydroclimate is shifting,” Beltran-Peña says. It’s moving “from a snow-dominated to rain-dominated system.”

Across the world in high mountain Asia, the vast Himalayan Arc faces similar woes. That range is a “bellwether for what’s going on with snow across the rest of the world,” Skiles says, because the Himalayas receive the most snow worldwide outside the poles. In a 2021 study, she used NASA Moderate Resolution Imaging Spectroradiometer (MODIS) Terra satellite records to track how snowpack had changed in the Himalayas between 2002 and 2017, when data were available. The MODIS satellite images every point on Earth every one to two days, “and that’s the temporal scale we need to monitor snow cover,” Skiles says, because it can change so quickly. “Broadly, we found snow cover was in decline everywhere,” she says, but it was most pronounced at mid-elevations between 4,000 and 5,000 meters, where snowpack is normally the deepest. On average, these elevations experienced up to 13 fewer days with any snow since 2002 (1). The study didn’t look at changes in rainfall, but “broadly, we found snow cover was in decline,” Skiles says, and “most pronounced in elevation bands where snow cover is most likely to be present, and most needed to sustain glaciers.”

Unusually low-snow winters will become more common as the effects of climate change become more pronounced. Here, Donner Ski Ranch, on Donner Summit in California, is dusted in very low snowpack in February 2018. Image credit: Benjamin Hatchett/Colorado State University, Fort Collins.

Water, Water Everywhere

Losing all that snow means losing water storage. Across the West, up to 80% of states’ water comes from melting snow. In spring and early summer, the meltwater slowly soaks into the soil or flows overland until it meets a stream. The icy melt trickles downhill, eventually splashing into lakes or artificial reservoirs that store most of the West’s water. “Pick any water supply reservoir in California, and they all have this issue,” says hydrologist Dennis Lettenmaier, who retired from the University of California, Los Angeles this summer. Western states count on snowmelt to refill their reservoirs every spring.

“We’re not set up to manage that water if it’s instantaneously moveable [as rain] rather than locked up as snow for months,”

—Ben Hatchett

“We’re not set up to manage that water if it’s instantaneously moveable [as rain] rather than locked up as snow for months,” says Benjamin Hatchett, an earth system scientist at Colorado State University in Fort Collins. The crux of the problem is that reservoirs are drawn down in spring and summer to meet demand, when snowmelt helps by topping them up. But with climate change, heavier rains will fall in winter when some reservoirs are already full, risking flooding.

Mote coauthored a 2021 study predicting that less snow and more rain will raise flooding risks along the Pacific Northwest’s Columbia River in the second half of this century (9). The East Coast is poised for flooding, too, says ecosystem ecologist Alexandra Contosta at the University of New Hampshire. That’s especially the case when rain falls on snowy hillsides. The combination can melt the snow and unleash a wall of water. In December 2023, rain fell on deep snow across Maine, New Hampshire, and Vermont, causing catastrophic flooding.

In the aftermath, Burakowski and other researchers advocated for expanding SNOTEL to the East Coast. In 2023, New Hampshire legislators introduced a bill that would give jurisdiction to expand SNOTEL eastward. The bill was reintroduced in March 2025, and the Senate Appropriations Committee advanced it in July, including $2 million toward a feasibility study (10).

Not everywhere is poised for more snow-related flooding, Lettenmaier notes. In coastal places, a warmer and rainier future may ironically mean less flood risk from melting snow. Why? Because the snowline will shift to higher elevations, meaning there’s less snow falling early in winter, and therefore less snowpack to melt and unleash flooding when rain does fall (11). “It’s kind of the flip side of the coin,” Lettenmaier says, compared to increasing flood risks at other elevations. If more water is falling as rain, there could still be flood risk, he cautions, “but it’s not from melting snow.” The factors that cause flooding are changing from place to place.

Mopping it Up

As for adaptation, building more reservoirs is one option, Hatchett says. It will cost billions, but California is already forging ahead with the Sites Reservoir, proposed to break ground in 2026, north of Sacramento. The reservoir is intended to catch and store overflow from the Sacramento River during winter storms. “The trade-off is we’ve dammed an ecosystem to create the reservoir,” Hatchett says. The water it captures won’t follow its natural course to the San Francisco Bay. Downstream, that may have ecological consequences since flows of fresh water maintain the salt balance in the bay’s estuary.

Another idea is to restore floodplains, where overflow from nearby creeks historically soaked through porous soils into groundwater. Restoration can entail everything from removing concrete paving so water can soak in, to reestablishing native wetland plants, to undoing the channelization and incision of a river or stream and reestablishing the natural floodplains along its banks (12).

But that’s not without its challenges. Some soils let water soak in slowly, some not at all. That sounds OK unless your farm is right there. “Who is willing to risk flooding their orchard,” Hatchett asks, if it might be underwater for days, weeks, or even months?

One flood control option might actually be draining reservoirs in some cases—into a man-made lake, another downstream reservoir, or ultimately out to sea—right before a forecasted storm, so that excess water has somewhere to go, Mote says. But there would be political challenges, he notes. If you don’t drain the reservoir enough, you still get flooding, and the project has failed. If you drain it too much, you have less water saved for summer. A flood would be much worse than draining too much water, Mote adds, but reservoirs don’t have the ability to respond to 10-day rain forecasts in real time yet.

California’s government is already trying to plan for a rainier, less snowy future. “Are we thinking about it? Unequivocally, yes,” says Michael Anderson, California’s State Climatologist. The Department of Water Resources and other state agencies are working on several reports, he says, to quantify the water supply and project how it may change. California will use those reports to inform how it allocates water and offsets expected losses. “It’s figuring out how to navigate each year… and being able to deal with the water that does show up,” Anderson says. “As the storms we get become rainier, how do you try and make use of the water?”

In far northern California, places like the Yuba watershed and Lake Mendocino are already draining reservoirs in anticipation of storms, “to make space for the storm to come in,” without causing flooding, Anderson says. California is also trying groundwater recharge, where excess water is diverted into basins; it then soaks into the ground, refilling underground aquifers. In 2023, the state recharged 4.1 million acre-feet, or about the storage capacity of Shasta Lake. California is “looking at opportunities to augment storage around the state,” Anderson adds. “The high-level view is we’re taking it seriously.”

As water storage becomes more fraught, municipalities will need to track every drop in hopes of maximizing efficiency. And researchers will need more monitoring of snow, rain, wind, temperature, and other relevant variables. SNOTEL offers good coverage in some areas, especially at higher elevations, Hatchett says, but does have blind spots. Meteorological and satellite data, paired with high-resolution modeling, can help fill in the blanks. Hatchett is already combining SNOTEL with these other data in snowpack forecasting models. He hopes to predict what “a heat wave will do to our snowpack next week,” as well as “what this watershed will look like in 50 years.”