If The Levees Fail In California…

The economic cost would top Katrina's. But a new tool may prevent catastrophe

If you were to draw up a list of the most worrisome infrastructure risks facing America, the leak-prone network of levees that run east from the San Francisco Bay up to Sacramento would rank right near the top. This 2,600-mile-long system of berms protects half a million people, 4 million acres of farmland, and the drinking water supply for most of Southern California. Vulnerable to either an earthquake or flooding, it is "like a ticking time bomb," warns Lester Snow, director of the California Water Resources Dept.

The good news is that Californians have heard the warnings and are starting to fix the levees. Last November they approved a $4.9 billion bond to fund repairs. But that's not nearly enough to fix the entire system, and residents now face a more vexing problem: deciding where the need is greatest.

This is a challenge that other localities face as they rush to fix aging bridges, roads, rails, and power plants in the wake of the collapse of the I-35W bridge in Minneapolis. Allocating repair money intelligently requires assessing risk accurately. That obliges policymakers to make smart predictions about weather, demographics, and countless other factors.

To guide their choices in the delta, officials are relying on a groundbreaking threat-assessment model devised by a team of 300 top scientists and engineers organized after Hurricane Katrina. It's far from perfect, but it's the most sophisticated tool of its kind ever developed and could one day become a template for guiding infrastructure investment in other areas. "We built a 200-pound bicycle," says team leader Ed (Lewis E.) Link, a senior fellow in civil and environmental engineering at the University of Maryland. "Each time it is used to measure risk in other areas, it will go faster."

Taking a spin on the 200-pound bike gives a sense of the dizzying array of factors to consider in making infrastructure investments. Developed to help organize the emergency reconstruction of New Orleans' flood defenses, the model seeks to analyze a wider range of information than had gone into earlier risk- assessment models.


Link's group started by walking the Crescent City's 350 miles of levees and flood walls to quantify their size, condition, and how they were built. It then loaded in fine-grained details about the types of structures in nearby communities--their elevation and condition as well as residents' income and age level. The team then plugged all the data into what Link calls "the mother of all spreadsheets."

Next, the team whipped up 152 computer-simulated storms to understand how waves and flood waters would swamp a digital New Orleans. Each virtual hurricane demanded 12 hours of supercomputing time. A year in the making, the test runs are helping to determine which levees to fix first and how high to rebuild them.

Many of the people who worked on New Orleans have been working in sync on a risk assessment of Sacramento's delta area. That will require significant customization. The Golden State has seven times the length of levees, and they're in worse condition. And rather than hurricanes, earthquakes are the presumed villain. A big quake, say 6.5 on the Richter scale, would "liquefy" the levees. In that case, "the loss of life would be lower" than it was for Katrina, says Link. "But the loss of economic value could be much worse."

As other communities struggle to decide what infrastructure to fix first, the question they should ask, says Transportation Secretary Mary E. Peters, is "Are we getting the best return on investment?" Link's trailblazing approach demonstrates the sort of painstaking analysis necessary to find the best answers.

By Adam Aston & Michael Arndt, with Tom Sawyer of Engineering News-Record in New York and staff

    Before it's here, it's on the Bloomberg Terminal.