Why Natomas levees flunked

When the U.S. Army Corps of Engineers two weeks ago revealed that Natomas levees are not tall enough to contain even a modest storm, it wasn't because the levees had shrunk overnight or because someone misread the yardstick.

Instead, the corps applied a new yardstick.

But the new method is so complex that many flood-control experts are struggling to understand it – even some within the corps itself, said Joe Countryman, former chief of civil works design in the corps' Sacramento District and now president of MBK Engineers, a flood-control consultancy.

"It's not being demeaning to say a lot of people at the corps do not understand this," he said. "It is extremely complex."

Still, the corps intends to use its more complex yardstick to certify all new Central Valley levees. So billions of dollars and thousands of lives may be at stake.

Known as "risk analysis," the new method uses statistical modeling to estimate the range of uncertainty behind water surface calculations. The results may require not just taller levees, but ones that come with a greater assurance of safety.

In traditional measurements, engineers use weather records and experience with how the system behaves to set possible flood elevations. The results may have less certainty, but are based on tangible events.

Questions about the new method may be academic for Natomas, because the basin's levees don't appear to pass muster for other reasons. There weren't even enough data to analyze the Natomas eastern perimeter. This forced the Federal Emergency Management Agency to impose a new flood risk category resulting in a likely building ban.

If the new measurements are applied to levees, such as those in Natomas, Countryman and others worry the risk analysis could exaggerate flood elevations and force communities to waste money building unnecessarily tall levees. But it could also underestimate risk, he said, leaving urban areas more exposed.

Corps officials agree risk analysis could produce taller or shorter levees depending on the situation. And, they say, that's exactly the point.

The new approach is intended as a more comprehensive estimate of risks. It could require taller levees where the variation in flows is greater, or allow shorter levees and cost savings where that is less likely.

"There's a suspicion that risk analysis is a ruse that the corps came up with to build bigger projects," said Darryl Davis, a senior adviser to the corps who helped develop the risk analysis approach before retiring as director of its hydrologic engineering center in Davis.

"All anyone's trying to do is improve the information that goes into the decision," he added.

In the traditional levee-certification approach, years of weather records are analyzed to determine the likelihood of experiencing certain flows. The 100-year flood, or one with a 1 percent chance of striking in any year, is often used to set a target water elevation, and levees are then designed to meet that.

On top of that, an additional 3 feet or so of "freeboard" space is added to levees to defend against catastrophic flows pushed by factors that can't be predicted.

The corps's risk analysis method starts with a similar approach, but also estimates the uncertainty behind the data.

The variables include weather history and watershed behavior, but also much more. In fact, the number of variables is dizzying in a flood control system like the Sacramento River, which includes dozens of reservoirs, more than 1,000 miles of levees, flood channels full of silt and vegetation, and a watershed larger than West Virginia with snowy peaks over 10,000-feet high.

The corps uses statistical methods to calculate uncertainty within each of these variables. Those scenarios are then sampled thousands of times in a mathematical model to reveal the larger range of uncertainty.

This allows the corps to capture the full range of possible 100-year water surface elevations, Davis said.

The levee height required for certification must be enough to provide at least "90 percent assurance" of containing this range of flows. In other words, the corps is stating that 90 percent of the possible 100-year floods will fall below this elevation.

About the writer:

• Call The Bee's Matt Weiser, (916) 321-1264.