Western U.S. headwater streams are projected to carry substantially less groundwater-supported flow by the end of the century, with more of that water arriving earlier in the year. In a new study, Caelum Mroczek, a recent graduate of NAU’s Earth Sciences and Environmental Sustainability Ph.D. program who completed the degree in May 2026, and his colleagues found that annual base flow could decline by roughly 45% to 65% across the region under two future climate scenarios. Mroczek is now a postdoctoral researcher at the University of Arizona’s Arizona Institute for Resilience. The projected changes threaten the late-summer flows that sustain ecosystems, communities, and downstream water users.
Base flow is the portion of streamflow supplied by groundwater or other delayed sources. It acts as a reserve during dry weather, when rainfall and snowmelt are no longer moving directly through a watershed. Mroczek compares the process to a mountain acting like a sponge: winter snow and rain enter soil and rock, then stored water gradually drains into streams. “This is the steady flow that keeps a creek flowing in dry periods and late summer,” Mroczek wrote.
To understand how that reserve is changing, the researchers analyzed 75 years of daily streamflow records from 115 U.S. Geological Survey streamgages, covering 1950 through 2024. They separated the groundwater-supported portion of streamflow from total discharge, then grouped the basins according to shared climate, terrain, and hydrologic characteristics. Finally, they combined statistical analyses with cluster-specific machine-learning models to test how base flow might change through 2099.
The four groups responded differently. Snow-storage mountain catchments held winter precipitation as snow and reached their base-flow peak around June. Large, buffered catchments, many in the Cascades, smoothed out short-term changes and showed smaller percentage losses, but their size meant they could account for the greatest total volume of water lost. Temperature-responsive mountain catchments could briefly gain base flow as warming triggered earlier snowmelt. Low-elevation, rain-dominated catchments had little snow storage, peaked around March, and were especially sensitive to heat and evaporation.
“‘Decline’ means different things in different places,” Mroczek wrote. A single regional average, he said, would hide both the timing differences and the fact that the same warming can increase one basin’s flow temporarily while reducing another’s.
Across all four groups, the strongest positive driver of base flow was antecedent moisture, which the study represented as precipitation accumulated during the previous three months. In everyday terms, it is a measure of how wet the ground already is and whether the soil is prepared to store water and release it later. The historical record showed the most widespread declines in June and July, while the projections showed modest increases in winter and spring followed by pronounced summer losses.
For Mroczek, the timing may be as important as the volume. “It’s not just a story of ‘less water,’” he wrote. “It’s water arriving on a schedule increasingly out of step with when streams, people, and ecosystems actually need it.” In the study’s projections, late-century summer base flow fell by as much as 4% to 5% of annual base flow in some clusters, while winter and spring contributions rose by roughly 1% to 5%.
The results are not a precise forecast for every basin. The projections relied on one global climate model, the authors did not screen out possible trans-basin diversions, and the models were less reliable at very low flows. Mroczek said the next assessment would benefit from multiple climate models, better separation of human and climate influences, and the addition of subsurface geology and soils to the basin classification.
The paper’s central message is that water managers may need to reconsider reservoir operations, drought planning, and environmental-flow targets built on historical seasonal patterns. As the West warms, protecting the groundwater and watershed storage that support late-season flow may become increasingly important.
Paper: Mroczek, C., Springer, A. E., & Lucas, B. (2026). “The Fate of Western Headwaters: Climate Controls on Base-Flow Decline.” Earth’s Future, 14, e2025EF007971. https://doi.org/10.1029/2025EF007971
Reported by SES Research Highlights for the School of Earth and Sustainability.