What do glass, dirty ice, and riverbeds have in common? All are disordered solids, meaning they have malleable internal structures that can rearrange themselves over time.

Materials scientists know that when force is applied to a disordered solid, structural changes will occur, impacting the material’s resistance to flow. Using a custom laboratory apparatus, Penn geophysicists have shown that riverbeds behave similarly, with sediment particles rearranging in response to fluid forces. Moreover, these structural changes cause particles within the bed to slowly creep forward over time, representing a new type of sediment erosion in rivers.

“This is a new state of behavior that no one has ever seen for grains in a fluid,” says Douglas Jerolmack, an associate professor in the Department of Earth & Environmental Science in the School of Arts & Sciences.

These insights appear in a Nature Communications paper co-authored by Jerolmack, physics professor Douglas Durian, and postdoctoral researchers Morgane Houssais and Carlos Ortiz. By treating riverbeds as disordered solids, geologists may improve their predictions of erosion and landscape evolution. The new perspective may also inform future civil engineering projects.

Rivers drive the evolution of Earth’s surface by moving sediment—eroding it upstream and depositing it downstream. But for decades, theoretical models, which use fluid dynamics to predict how much sediment a river will move, have rarely matched what geologists measure in nature.

Jerolmack started discussing this problem with Durian at interdisciplinary research meetings held at Penn’s Materials Research Science and Engineering Center. The strange behaviors geologists observe in rivers, Jerolmack discovered, are similar to what materials scientists see in dry sand and glass. Jerolmack and Durian came to suspect that riverbeds behave like disordered solids, and that the reason models do not accurately predict river transport is because their exclusive focus on the fluid ignores what is happening in the riverbed.

Durian and Jerolmack tested that hypothesis by constructing an idealized laboratory “river,” in which the motion of individual grains within a bed could be precisely tracked over days. Sure enough, the researchers found that when fluid forces are applied, structural rearrangements of the riverbed occur, resulting in slow particle creep beneath the bed’s surface. This is not only exactly the sort of behavior one would predict for a disordered solid, it supports the notion that current sediment transport models do not account for important bed dynamics.

How much this riverbed “creep” contributes to sediment transport in actual rivers remains to be tested. But for calm rivers where the fluid force is low, Jerolmack suspects creep may be significant.

“If you want to predict the erosion of a river or the change in a landscape, you need to predict how much stuff moves,” Jerolmack says. “Historically, researchers have focused on the fluid and treated the granular part of this problem as a black box. Bringing knowledge from materials science and physics to bear, we can understand rivers and landscapes in a fundamentally different way.”