On the Levee

Meandering rivers are commonly flanked by earthy embankments higher than the surrounding land: the buildup from floodwaters dropping layers of sediment. Natural levees, geologists call these formations. "As it moves out over the floodplain," Peter Adams explains, "the water loses its turbulent intensity, and the sediment just falls out."

purple, green, black swirls
Courtesy Nasa Jet Propulsion Laboratory

Roll on, big river. A satellite image of the lower Mississippi shows its broad and shifting floodplain. Where the land is flat, natural levees tend to be gently sloped. In deep river valleys, on the other hand, the levees are narrow and steep.

Natural levees are not all alike, however. In cross-section, some are narrow and steep as pup tents, others broad and gently sloped, like inverted dinner plates. What variables could shape them so differently?

According to Adams, who recently completed a master's degree in geosciences at Penn State, two distinct processes are at work. Where levees are steep, he suggests, they have been formed by what he calls turbulent diffusion. "This occurs where the floodplain fills quickly, and the water level of the river and that of the surrounding floodplain are the same. It's like a hose flowing into a bathtub. At the margin between swift current and stagnant water, there is turbulence. This turbulence forms eddies, which carry momentum and sediment off into the floodplain."

Where levees are gently sloped, Adams says, something called advection is in control: a horizontal movement en masse. "Here the floodplain isn't filled as rapidly. There's a difference in elevation. Not a big difference—it only has to be enough to generate a flow. But the mass of water, with its sediment, moves out laterally as a body."

Flood plains fill rapidly, Adams continues, where they are tightly contained—in a valley between two ridges, say. That's where you would expect to find narrow, steep levees formed by turbulent diffusion. Where the land is flat and floodwater can roll out unimpeded, on the other hand, the levees should be broad and gently sloped.

Adams and his adviser, Penn State professor of geosciences Rudy Slingerland, devised mathematical models that supported this hypothesis. To properly test it, however, they needed examples in the field. During the summer of 1998, accordingly, Adams spent seven weeks on the Upper Columbia River, which flows through a narrow, steep-walled valley in the Canadian Rockies.

"It was pretty remote," he remembers. "I was able to find a little beaver cabin, three kilometers by boat from the nearest place to park a car. Its owners said, 'Sure, it's past trapping season. Feel free to stay there.'"

Each day Adams and Remus Lazar, a graduate student from the University of Nebraska, along with David Pinkus, an undergraduate field assistant Adams had recruited back in University Park, spent hours on the river, surveying, "getting the shapes of the levees there with pretty detailed accuracy." They also drilled cores into the levee up to 16-feet deep in order to analyze the sizes of the sedimentary particles the levee contained.

Back in Pennsylvania, Adams compiled his data and compared it against measurements taken the year before, by another of Slingerland's students, then-undergraduate Steve Nelson, on the Lower Saskatchewan River in the plains of eastern Saskatchewan. "It's the Kansas of Canada," Adams explains. "There are no mountains around, nothing confining the river."

He found exactly what he had expected. "In general, the Saskatchewan River levees are wider and more gently sloping, while the Columbia's levees are steeper and narrower." In addition, the data allowed him to rule out a couple of other variables—particle size and channel width—as reliable predictors of levee shape.

"So the models and the data match up quite well," he says. "But does this hold true for other rivers? We need to make a prediction on another river, then get some data and see whether it holds up."

As a preliminary step, he reports, he has been checking out digital elevation models on the Worldwide Web, calculating the levee slopes of several major rivers. "If diffusion versus advection is the principal influence," he says, "we should be able to find some relationship between levee slope and valley width." The data plot so far, including the Tuross in Australia, and the Illinois, the Missouri, and the San Joaquin in North America, "shows a very good correlation—enough to convince us that we really have something valuable to contribute.

"Now, the more data we can gather, the better."

Peter N. Adams completed his master's degree in geosciences at Penn State in June 1999, and is currently pursuing a Ph.D. at the University of California at Santa Cruz. His adviser, Rudy L. Slingerland, Ph.D., is professor and head of the department of geosciences in the College of Earth and Mineral Sciences, 503 Deike Bldg., University Park, PA 16802; 814-865-6842; sling@geosc.psu.edu. Slingerland and colleague Norman Smith, Ph.D., of the University of Nebraska, have received a grant from the National Science Founda-tion to continue the project reported above.

Last Updated January 01, 2000