Every winter, snow piles up across the mountains of Colorado, Wyoming, and Utah, building the frozen reservoir that feeds the Colorado River and supplies water to roughly 40 million people in seven states and dozens of tribal nations. Every spring, water managers watch the melt and wait for it to show up in rivers. For more than two decades, less water has arrived than the snowpack promised. Now, a peer-reviewed study published in April 2026 in Geophysical Research Letters identifies the primary reason: warmer, drier springs are letting plants and soil consume meltwater before it ever reaches a stream channel.
The finding upends a long-held assumption. For years, many scientists suspected sublimation, the process by which snow converts directly into water vapor without melting, was the leading cause of the shortfall. Instead, the new research points to evapotranspiration, the combination of evaporation from soil and water uptake by vegetation, as the dominant culprit in the Upper Colorado River Basin’s headwaters.
What the research found
The study analyzed decades of snowpack measurements and stream gauge records across the Upper Colorado River Basin, comparing what the snowpack should have produced against what actually flowed downstream. Since 2000, the gap between those two numbers has grown. The researchers attribute most of that shortfall to two linked trends: declining spring precipitation and rising evapotranspiration driven by warmer temperatures.
The mechanism is straightforward. As spring temperatures climb, the growing season starts earlier. Vegetation wakes up sooner, draws more moisture from the soil, and releases it into the atmosphere. Soils themselves dry out faster in warmer conditions, absorbing snowmelt like a sponge before it can trickle into creeks. The result is that even a winter with healthy snowpack can produce a disappointing runoff season.
Sublimation has not been ruled out entirely, but its role appears smaller than expected. The U.S. Department of Energy funded a dedicated field effort called the Sublimation of Snow (SOS) campaign from 2022 to 2023, deploying instruments across Colorado’s mountains to directly measure how much water vapor snow loses to the atmosphere. A technical report from the campaign documented seasonal vapor losses and notable blowing-snow events, producing the most detailed direct sublimation measurements ever collected in complex mountain terrain.
A peer-reviewed synthesis of those results, published in the Bulletin of the American Meteorological Society, confirmed that sublimation accounts for a measurable share of water loss but is not the dominant pathway. The synthesis also flagged significant uncertainty in scaling site-level sublimation readings across thousands of square miles of varied terrain, meaning the exact contribution remains an open question at the basin scale.
The broader observational foundation for this work came from the DOE’s Atmospheric Radiation Measurement Surface Atmosphere Integrated Field Laboratory (SAIL) campaign, which operated in the East River watershed in Colorado from 2021 to 2023. SAIL provided wind measurements, boundary-layer profiling, and distributed weather station data that helped researchers separate sublimation losses from evapotranspiration and other processes.
What scientists still do not know
The new study narrows the mystery, but it does not close it. The precise fraction of missing water attributable to each mechanism, whether evapotranspiration, sublimation, soil absorption, or underground flow paths, varies by elevation, slope orientation, and vegetation type. A DOE Environmental System Science project focused on the East River watershed is testing competing hypotheses about whether patchy snow distribution, subsurface water connectivity, or surface evaporation drives the largest losses at finer scales. That work has not yet produced published observational datasets.
Scaling remains a core challenge. The SOS campaign measured sublimation at specific instrumented sites. Extrapolating those readings across the full Upper Colorado Basin, a region spanning tens of thousands of square miles with wildly different terrain, introduces uncertainty that researchers have openly acknowledged.
There is also a gap between the science and the policy tools that depend on it. The U.S. Bureau of Reclamation maintains provisional natural flow estimates at Lees Ferry, Arizona, the key accounting point for the Colorado River, spanning 1906 to 2024. Those records distinguish observed flows from naturalized flows and serve as the baseline for allocation decisions. But the dataset is an operational record, not a predictive model, and it does not yet incorporate the mechanistic insights from these recent field campaigns. Whether federal water managers will update their forecasting frameworks in response to the new evapotranspiration findings has not been publicly announced.
Why the distinction matters for the river’s future
For the states, tribes, cities, and farms that share the Colorado River, the difference between sublimation and evapotranspiration is not academic. Sublimation is driven primarily by wind, humidity, and snow surface conditions, factors that are difficult to manage but relatively stable over time. Evapotranspiration, by contrast, responds directly to temperature, vegetation density, and land cover, all of which are shifting as the climate warms and landscapes change.
That distinction carries a blunt implication: if rising spring temperatures are the primary driver of the river’s losses, then the gap between snowpack and streamflow will likely keep widening even in winters with above-average snowfall. More snow will not necessarily mean more water in the river.
The Colorado River Basin is already operating under renegotiated shortage guidelines, with Lake Powell and Lake Mead hovering at levels that have triggered unprecedented cuts to downstream deliveries in recent years. Water allocation negotiations, reservoir operations, and agricultural planning across the basin have historically relied on the assumption that snowpack is a reliable predictor of runoff. This research suggests that assumption needs revisiting, and that the season between snowfall and runoff, not the snowpack itself, may be the most important variable to watch.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
