About a century ago, Franklin D. Roosevelt’s New Deal helped kickstart a decades-long era of large dam construction in America. To this day, a patchwork of infrastructure stores water, controls river flows and generates power across the nation.

But, slowly, their ability to store water is decreasing from the bottom up. 

A new study from former INSTAAR PhD student Abigail Eckland and collaborators modeled sediment accumulation in more than 57,000 American reservoirs. The researchers found that reservoir storage capacity will decrease by a total of 8% by 2025. A large minority of these reservoirs, around 20,000, are on track to lose more than a quarter of their storage capacity on the same timeline.

Sediment has long been a focus for both scientists and water managers, but this is the first time the issue has been addressed at this scale. 

“The project started with this big idea of trying to find out how much water storage capacity is still remaining across U.S. reservoirs and how much sediment is stored inland,” Eckland said. 

The new study's second major insight comes further downstream. Because much river sediment is captured by reservoirs, less sediment is reaching the ocean via coastal deltas. 

In an undammed river, sediment naturally accumulates at a delta, raising or replenishing the seabed. But heavily dammed rivers deliver less sediment, potentially exacerbating saltwater intrusion and coastal flooding caused by sea-level rise and low river flows.

“If all that sediment actually made it to the coast, it would accrete at roughly the same rate as sea-level rise for around half of the nation’s deltas,” Eckland said. 

Abby Eckland and Lavender Merrill collect a shallow sediment core at the Wax Lake Delta in Louisiana in June of 2026. (Courtesy, Abby Eckland)

While the new study highlights a significant problem facing water managers, it also suggests a potential mitigation strategy. According to the paper, around 7% of this trapped sediment is stored in “terminal reservoirs.” These dammed bodies of water are the final stop before the river reaches the coast, making them potential targets for engineering projects that could move trapped sediment downstream.  

Ultimately this decision would fall on water managers — including one agency that was heavily involved in the investigation from the start. Eckland worked with scientists from the Bureau of Reclamation to develop and execute the project. 

“You can always dream about policy implications when you’re at your desk, but working with the Bureau on this made it more concrete,” Eckland explained.

Just last year, Eckland and bureau scientist Melissa Foster collaborated on a dataset developed by CU Boulder PhD student Aaron Hurst called “ResNet.” The tool maps water storage reservoirs onto an existing national dataset of freshwater flows. 

The new study is the first investigation to rely on ResNet to generate scientific insights. The researchers first developed a model to predict sediment accumulation and storage capacity changes, usingResNet’s data. This allowed the scientists to analyze each reservoir within the context of a larger watershed dam network, rather than as an isolated system.

“In the past, researchers would take sedimentation rates from survey data and use that to estimate the loss of storage capacity over time for an individual reservoir,” Eckland’s PhD advisor Irina Overeem said. “But, if you build a big dam further upstream, you need to take that into account.”

Eckland hopes that the big-picture insights from the paper prove useful to water managers.

“We now have a number that shows how much sediment is in target areas for sustainable management,” she said. 

With that information, managers might begin to see sediment not just as a problem accumulating behind dams, but as a resource that might be put back to work downstream.