The U.S. has long built levees to protect communities from floods, but levees have a side effect: by lowering flood risk in one area, they raise the flood risks for neighboring towns. Few scientists study this problem, even as deadly record floods inundate the Midwest. Regional coverage of levee disputes can be reduced to he-said/she-said arguments. We set out to investigate the decision-making behind building levees, showcasing the flaws in planning and the human harm, with a foundation in science. We knew we needed to illustrate the counterintuitive concept that levees could make flooding worse, and we knew we needed to innovate to capture the interest of readers.
So we worked with researchers, who built a scale model of a tiny community, protected from a miniature river by a miniature levee, and we flooded it. The resulting video and data graphics are a clear warning of the potential for big problems.
What makes this project innovative?
We worked with scientists at the University of Minnesota who research fluid dynamics to build a 10-foot by 13-foot scale model. It was scientifically sound, visually attractive and simple enough for the average viewer to grasp, and it included tiny wooden houses, carved by the fathers of two of our journalists and painted by hand as a newsroom-wide project. In partnership with Vox, our video journalists wore hip waders during the four-day, six-camera filming session. The lab's laser scanner captured millimeter-resolution data of the topography of the basin and water heights. The results are a visual interactive that allows the audience to test different levee scenarios and see how the flooding changes as the water flows faster and levee walls grow taller: https://projects.propublica.org/graphics/levees
What was the impact of your project? How did you measure it?
Our project provided local officials and residents information that was otherwise out of reach. Communities had been awaiting a model from the U.S. Army Corps of Engineers that showed that overbuilt levees in Illinois raised flood levels upstream and across the river. The model was a public record, but its format was so complex, it needed to be plugged into a program built for hydraulic engineers. A professor at Southern Illinois University analyzed the model for us, and we published the results in our story as an animated map. We were the only media outlet to do this. We showed how the Army Corps fails to re-evaluate projects after they’re built, paving the way for aggravated flooding. In collaboration with the Alton Telegraph, we exposed a secretive lobbying effort to deregulate federal levees. We used the Corps’ own post-disaster report to bring accountability to its delayed decision to save Cairo, Ill. from a massive 2011 flood -- hesitating to save an impoverished, majority-black city amid lobbying from well-connected farmers who had planted crops in the floodway.
Source and methodology
We created an array of interactive graphics to explain the effects of levees in the real world; much of the data behind these graphics were obtained by or created by us exclusively. River flume model designed, built and operated by Jeffrey Marr, Robert Gabrielson, Aaron Ketchmark and Erik Steen at the St. Anthony Falls Laboratory at the University of Minnesota. SAFL staff collected and processed high resolution topography and water surface elevation data from the model for ProPublica and Reveal. A paper by Heine and Pinter found the levee raised flood levels by more than seven feet, but a review of the Wabash River stream gauge’s history, prompted by ProPublica and Reveal, confirmed the maximum change is closer to five feet. To create pre- and post-levee inundation layers on the Wabash River, Heine applied USGS stream gauge data to USGS high resolution lidar elevation data. To approximate pre-levee inundation data in the area now occupied by the Duke Energy power plant and cooling pond, he blended high resolution data with USGS 1950s 10 ft. contour lines to create a composite that that aimed to represent the pre-levee floodplain features.
10-foot by 13-foot flood plume (hammers, nails, etc.), laser scanner, HTML/CSS, JS
Al Shaw, Lisa Song, Katie Campbell and Ranjani Chakraborty